Description
Employment is a relationship between two parties, usually based on a contract, one being the employer and the other being the employee.
ABSTRACT
Title of Document:
ORGANIZATION-EMPLOYEE RELATIONSHIPS MODEL: A TWO-SIDED STORY Hongmei Shen, Doctor of Philosophy, 2009
Directed By:
Professor Elizabeth L. Toth Department of Communication
The purpose of this study was to develop a theory of internal relationship management, and to propose a new way of measuring organization-public relationships by simultaneously examining the organizations’ as well as their employees’ perceptions of the quality of their relationships. It sought to contribute to theory-building on the process of relationship management from its maintenance through its quality to the consequences. An online survey was used to collect data. Usable questionnaires totaled 785 from 30 organizations. Data analytic methods included missing value analysis, exploratory factor analysis, confirmatory factor analysis, multivariate regression, polynomial regression, surface response tests, mediation tests, and reliability tests.
The proposed measures of symmetrical relationship maintenance strategies, asymmetrical relationship strategies, organization-employee relationship characteristics, turnover intention, and contextual performance were found to be valid and reliable. The major findings included: first, the more organizations used symmetrical relationship maintenance strategies with their employees, the more likely both employees and the organizations reported greater trust, control mutuality, commitment, satisfaction, and less distrust in the relationship; and vice versa for asymmetrical strategies. Second, employees would have higher turnover intention when both employees and their organizations perceived higher distrust and lower trust, control mutuality, commitment, and satisfaction. Also, when employees were more optimistic than their organizations about their relationships, employees were more likely to leave the organization. Third, employees’ contextual performance would rise as both these employees and their organizations reported greater level of commitment and satisfaction. However, employees’ level of contextual performance would drop when incongruence increased. Lastly, mediation tests showed that the effects of symmetrical and asymmetrical relationship maintenance strategies on turnover intention and contextual performance were partially mediated by congruence of perceived relationship characteristics, excluding the dimension of distrust regarding the effect of relationship maintenance strategies on contextual performance. This study contributed to public relations theory by 1) clarifying and refining the conceptualizations and operationalizations of relationship maintenance strategies, congruence of perceived relationship characteristics, and organizational effectiveness, 2)
proposing a new way to evaluate two sides of organization-public relationships, and 3) empirically testing a relationship-building model within organizations to develop a theory of internal relationship management.
ORGANIZATION-EMPLOYEE RELATIONSHIPS MODEL: A TWO-SIDED STORY
By Hongmei Shen
Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park, in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2009
Advisory Committee: Professor Elizabeth L. Toth, Chair Professor Meina Liu Professor Xiaoli Nan Professor Leah Waks Professor Cheri Ostroff
© Copyright by Hongmei Shen 2009
Dedication
I dedicate this dissertation to my family in China and my husband Shan Zhu. I am grateful for their love and support that made this work possible.
ii
Acknowledgements
Academic research is a road less travelled by. One needs courage and support to fend off self-doubts on the way to excellence. Only a novice on the road, I owe immense debt to many people who have encouraged and supported me. Special thanks go to my advisor, Dr. Elizabeth L. Toth for her continued support, encouragement, and guidance throughout the process. Thank you for holding me to higher standards. Thank you for guiding me into the academic world. Thanks also are due to my committee members: Dr. Meina Liu, Dr. Xiaoli Nan, Dr. Leah Waks, and Dr. Cheri Ostroff. Dr. Liu, Dr. Nan, and Dr. Waks shared their wisdom and knowledge whenever I asked. I am particularly grateful for Dr. Ostroff who has shared her expertise in quantitative methods and knowledge in organizational psychology and provided timely feedback on my data analysis even at midnight, without which I could not have completed this dissertation. I also thank Dr. Edward L. Fink who has been my mentor in quantitative methodology, teaching, and life. I thank Dr. Linda Aldoory whose academic rigor has pushed me forward. I am grateful for Dr. Gregory R. Hancock who patiently discussed with me possible ways to analyze my data. I thank my friend Dr. Jeong-Nam Kim who has spent so much time helping with my data analysis. Dr. Dale Hample, Dr. Brecken Chinn Swartz, and Mr. Richard J. Toth selflessly assisted my data collection. I am indebted to my friends at Maryland: my sisters Huang Jiang, Ai Zhang and Xiaodong Hou whose friendship made the journey more enjoyable, my buddies Xiaoying Xie and Bing Han who shared numerous insights on the phone in my research, Lucinda
iii
Austin and Lindsay Becker who offered tremendous help in my data collection, Shuo Yao, Sabine Chai, Nance McCown, Ying Wei, Yi Luo, Jingwei Yu, Teng Zhang, Lin Zhu, Lin Julie Zhu, Ling Na, Paula Weissman, Hyunhee Kim, Jarim Kim, Ahnlee Jang, Katie Place, Renata Schloss, Brian Smith, Jen Vardeman, Sanja Sipek, Jill Underhill, Erica Lamm, Sejal Patel, and David Payne. I thank my family in China: my parents Changchun Shen and Jiaxiu Du, my parents-in-law, Guangheng Zhu and Mingzhe Feng, my sister Qiurong Shen and my brothers Jianping Shen and Dong Xiang. Last, but not least, I owe the deepest gratitude to my husband Shan Zhu for his love, sacrifice, and support.
iv
Table of Contents
List of Tables ................................................................................................................... viii List of Figures .................................................................................................................... xi Chapter I.............................................................................................................................. 1 Overview of Study .............................................................................................................. 1 Purpose............................................................................................................................ 1 Theoretical Framework................................................................................................... 4 The Dialectical Approach ........................................................................................... 4 Extant Relationship Models........................................................................................ 5 Relationship Maintenance Strategies.......................................................................... 7 Congruence of Perceived Organization-Employee Relationship Characteristics....... 9 Organizational Effectiveness Outcomes ................................................................... 11 Method and Research Design ....................................................................................... 13 Significance of the Study .............................................................................................. 14 Chapter II .......................................................................................................................... 16 Conceptualization ............................................................................................................. 16 Defining Organization-Public Relationships ................................................................ 17 Why Organization-Employee Relationships?............................................................... 20 Research on Relationship Management........................................................................ 21 Application of the Relational Perspective................................................................. 22 Relationship Maintenance Strategies........................................................................ 23 Dimensions, Types, and Models of Organization-Public Relationships .................. 24 Conclusion ................................................................................................................ 31 Toward A Theory on Organization-Employee Relationships ...................................... 33 Systems Theory......................................................................................................... 33 Resource Dependence Theory .................................................................................. 35 Social Exchange Theory ........................................................................................... 36 The Dialectical Approach ......................................................................................... 37 Relationship Maintenance Strategies............................................................................ 42 Definition and a Terminological Issue...................................................................... 42 Relationship Maintenance Strategies in the Literature ............................................. 44 Symmetrical Maintenance Strategies........................................................................ 44 Congruence of Perceived Relationship Characteristics ................................................ 51 Trust .......................................................................................................................... 53 Distrust...................................................................................................................... 54 Commitment ............................................................................................................. 58 Satisfaction................................................................................................................ 60 Control Mutuality...................................................................................................... 61 Congruence of Perceived Relationship Characteristics ............................................ 62 Measurement Issues Related to Congruence of Perceived Relationship Characteristics........................................................................................................... 66
v
Organizational Effectiveness ........................................................................................ 74 Organizational Effectiveness: Definition and Approaches....................................... 74 Contextual Performance............................................................................................ 78 Turnover Intention .................................................................................................... 81 Model and Hypotheses: Linking the Variables............................................................. 82 Research on Relationship Maintenance Strategies as Antecedents .......................... 82 Summary ....................................................................................................................... 91 Chapter III......................................................................................................................... 95 Methodology ..................................................................................................................... 95 Survey ........................................................................................................................... 95 Why Survey?............................................................................................................. 95 Strengths and Weaknesses of Online Surveys .......................................................... 96 Sampling ....................................................................................................................... 98 The Appropriateness of Non-Probability Sampling ................................................. 98 Sample Size Estimation and Procedures................................................................... 99 Pilot Studies ................................................................................................................ 100 Pilot Study 1............................................................................................................ 100 Pilot Study 2............................................................................................................ 102 Measures ..................................................................................................................... 103 Antecedents: Relationship Maintenance Strategies ................................................ 103 Mediator: Congruence of Perceived Organization-Employee Relationship Characteristics......................................................................................................... 104 Dependent Variables............................................................................................... 105 Data Collection Procedures......................................................................................... 106 Data Analysis .............................................................................................................. 108 Data Aggregation .................................................................................................... 108 Statistical Procedures for Data Analyses ................................................................ 108 Chapter IV....................................................................................................................... 111 Results............................................................................................................................. 111 Description of Participants.......................................................................................... 111 Missing Values Analysis............................................................................................. 115 Data Aggregation ........................................................................................................ 117 Means, Standard Deviation, and Correlations ............................................................ 121 Research Question 1: How do organizational members perceive relationship maintenance strategies used by their organizations? .............................................. 122 Research Question 2: How do non-management employees and managers perceive the organization-employee relationships?............................................................... 124 Turnover Intention and Contextual Performance from the Non-Managerial Employees’ Perspective .......................................................................................... 126 Reliability Tests and Factor Analyses......................................................................... 127 Reliability Coefficients for Initial Measures........................................................... 127 EFA and Reliability Results.................................................................................... 128 CFA Results ............................................................................................................ 144 Relationship Maintenance Strategies...................................................................... 145 Turnover Intention and Contextual Performance ................................................... 152 Perceived Relationship Characteristics................................................................... 158
vi
Construct Reliability and Variance Explained........................................................ 169 Summary ................................................................................................................. 171 Hypotheses Testing..................................................................................................... 172 Multicollinearity Test.............................................................................................. 172 Multivariate Regression for Hypothesis 1 .............................................................. 174 Polynomial Regression Steps.................................................................................. 181 Overall Effect of Congruence of Perceived Relationship Characteristics on Dependent Variables............................................................................................... 183 Congruence of Perceived Relationship Characteristics and Turnover Intention .... 183 Congruence of Perceived Relationship Characteristics and Contextual Performance ................................................................................................................................. 186 Testing the Mediation Effect ...................................................................................... 208 Summary ..................................................................................................................... 214 Chapter V ........................................................................................................................ 216 Conclusions and Implications ......................................................................................... 216 Summary of Results.................................................................................................... 216 Data Aggregation .................................................................................................... 216 Descriptive Statistics Results for Research Questions............................................ 217 Reliability and Validity of Measuring Instrument .................................................. 218 Hypotheses Testing................................................................................................. 222 Mediation Effect Test Results................................................................................. 227 Theoretical Implications ............................................................................................. 228 Refined Conceptualization...................................................................................... 229 Relationship Building within Organizations........................................................... 233 Linking Public Relations to Organizational Effectiveness ..................................... 236 Relationship as a Two-Sided Story: Methodological Implications ............................ 237 Practical Implications.................................................................................................. 239 Limitations .................................................................................................................. 241 Future Research .......................................................................................................... 242 Appendix A. Questionnaire ............................................................................................ 246 Appendix B. Contact Email Template ............................................................................ 255 References....................................................................................................................... 256
vii
List of Tables
Table 1 Trust and Distrust Prototypical Relationship Conditions .................................... 56 Table 2 Descriptions of Participating Organizations ...................................................... 112 Table 3 Demographic Information of Participants (N = 785*)....................................... 114 Table 4 Within-Group Agreement (rwg) ........................................................................ 118 Table 5-I ICC (1) and ICC (2) Results for Managers’ Data (n = 195) ........................... 120 Table 5-II ICC (1) Results for Employees’ Data (n = 583) ............................................ 121 Table 6 Means, Standard Deviations, and Correlations of Independent Variables (n = 583) ............................................................................................................................. 123 Table 7 Means, Standard Deviations, and Correlations of Dependent Variables (n = 583) ............................................................................................................................. 124 Table 8 Means, Standard Deviations, and Correlations of Congruence of Perceived Relationship Characteristics (n = 195)................................................................ 125 Table 9 Organizational-Level Means, Standard Deviations, and Correlations of Congruence of Perceived Relationship Characteristics (n = 30) ........................ 126 Table 10 Cronbach’s Alpha of Initial Measures ............................................................. 128 Table 11 Items, Factor Loadings, and Reliability for Relationship Maintenance Strategies (n = 583).............................................................................................................. 131 Table 12 Items, Factor Loadings, and Reliability for Relationship Characteristics (n = 785) ............................................................................................................................. 136 Table 13 Items, Factor Loadings, and Reliability for Turnover Intention and Contextual Performance (n = 583) ........................................................................................ 142 Table 14 Symmetrical Relationship Maintenance Strategies CFA Results (n = 583).... 146
viii
Table 15 Asymmetrical Relationship Maintenance Strategies CFA Results (n = 583).. 149 Table 16 Turnover Intention CFA Results (n = 583)...................................................... 154 Table 17 Contextual Performance CFA Results (n = 583) ............................................. 156 Table 18 Trust Distrust CFA Results (N = 785) ............................................................. 159 Table 19 Perceived Relationship Characteristics CFA Results (N = 785) ..................... 162 Table 20 Correlations between Items with Added Error Covariance for Relationship Characteristics Measurement Model (N = 785).................................................. 165 Table 21 Comparison of Correlations between Relationship Characteristics Dimensions in the Initial Five-Factor Oblique Model vs. the Final Five-Factor Oblique Model with Error Covariances (N = 785) ...................................................................... 168 Table 22 Variance Explained and Coefficient H for All Measurement Models............. 169 Table 23 VIF of Independent Variables (n =583)........................................................... 174 Table 24 Multivariate Results of Symmetrical Relationship Maintenance Strategies Predicting Congruence of Perceived Organization-Employee Relationships (n =583)................................................................................................................... 178 Table 25 Multivariate Results of Asymmetrical Relationship Maintenance Strategies Predicting Congruence of Perceived Organization-Employee Relationships (n =583)................................................................................................................... 181 Table 26 Polynomial Regression Results for Congruence of Perceived Relationship Characteristics and Turnover Intention (n =583)................................................ 186 Table 27 Polynomial Regression Results for Congruence of Perceived Relationship Characteristics and Contextual Performance (n =583) ....................................... 188
ix
Table 28 Polynomial Regression Results for Congruence of Perceived Relationship Characteristics and Altruism (n =583)................................................................ 196 Table 29 Polynomial Regression Results for Congruence of Perceived Relationship Characteristics and Conscientiousness (n =583) ................................................ 203 Table 30 Mediation Effect Results for Relationship Maintenance Strategies and Turnover Intention (n =583) ............................................................................................... 211 Table 31 Mediation Effect Results for Relationship Maintenance Strategies and Contextual Performance (n = 583)...................................................................... 214
x
List of Figures
Figure 1. The structural model of internal relationship management theory...................... 4 Figure 2. The coorientation model adapted from Broom and Dozier (1990) ................... 68 Figure 3. Final four-factor oblique model of symmetrical relationship maintenance strategies. ............................................................................................................ 148 Figure 4. Final second-order factor model of asymmetrical relationship maintenance strategies. ............................................................................................................ 152 Figure 5. Final one-factor model of turnover intention. ................................................. 154 Figure 6. Final two-factor oblique model of contextual performance. ........................... 158 Figure 7. Final five-factor oblique model of perceived relationship characteristics. ..... 167 Figure 8. Congruence of perceived trust on contextual performance............................. 190 Figure 9. Congruence of perceived control mutuality on contextual performance. ....... 191 Figure 10. Congruence of perceived commitment on contextual performance.............. 192 Figure 11. Congruence of perceived satisfaction on contextual performance................ 193 Figure 12. Congruence of perceived trust on altruism.................................................... 198 Figure 13. Congruence of perceived commitment on altruism....................................... 200 Figure 14. Congruence of perceived satisfaction on altruism......................................... 201 Figure 15. Congruence of perceived trust on conscientiousness. ................................... 205 Figure 16. Congruence of perceived control mutuality on conscientiousness................ 206 Figure 17. Congruence of perceived commitment on conscientiousness. ...................... 207 Figure 18. Congruence of perceived satisfaction on conscientiousness. ........................ 207 Figure 19. Final model of internal relationship management. ........................................ 228 Figure 20. Final model of internal relationship management. ........................................ 228
xi
Chapter I Overview of Study Purpose Since the 1970s, organizations have demanded that public relations practitioners demonstrate the value of communication programs, as public relations is an intangible management function (L. Grunig, J. Grunig, & Dozier, 2002). Practitioners have to justify why organizations should spend millions of dollars in their work (J. Grunig & Y. Huang, 2000; Lindenmann, 1997, 2003). Public relations practitioners often face downsizing when organizations hit financial crises because of the difficulty to demonstrate public relations’ value (L. Grunig et al.). As the former President and CEO of the Institute for Public Relations Jack Felton said, “For years we have been told that we can never expect to get proper credit for what we do in public relations until we can find an effective way to measure our effectiveness” (Lindenmann, 1997, p. 1). To address this question, the strategic management approach posited that the public relations function adds values to organizations by scanning the environment, identifying stakeholders, and segmenting different publics from stakeholder categories (e.g., Dozier, L. Grunig, & J. Grunig, 1995; J. Grunig, 2006; L. Grunig et al., 2002). Public relations also adds value by using two-way symmetrical communication to cultivate long-term quality relationships with strategic publics (e.g., J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999; Y. Huang, 1997; Hung, 2001, 2002, 2004, 2007; H.-S. Kim, 2005; Ledingham, 2003; Ledingham & Bruning, 1998, 2000; Ni, 2006; Yang, 2005).
1
These strategic management scholars particularly contended that the value of public relations lies in relationships. Since Ferguson’s (1984) call for a relational perspective of public relations, relationship management has emerged as a new paradigm in public relations research (Ledingham, 2003). Research on organization-public relationships has identified different types of relationships (e.g., Bruning & Ledingham, 1999; Hung, 2004, 2005, 2007), models of organization-public relationships (e.g., Broom et al., 2000; J. Grunig & Y. Huang, 2000; Toth, 2000), dimensions of relational outcomes (e.g., Bruning & Ledingham, 1999, 2000; Hon & J. Grunig, 1999; Y. Huang, 2001; Y. Kim, 2001), strategies to maintain relationships (e.g., J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999; Hung, 2007; Ki & Hon, 2007a), and the effect of relationships on reputation (J. Grunig & Hung, 2002), publics’ attitudes, behavioral intention (Ki & Hon, 2007b), and actual behavior (Bruning, 2002). It has been applied to community relations, media relations, employee relations, public affairs, issues management, and crisis communication (e.g., Bridges & Nelson, 2000; Coombs, 2000; Ledingham & Bruning, 1998; L. Wilson, 2000). One problem with this body of literature is that most organization-public relationship research views relationship quality as the ultimate end product and purpose of public relations programs. With its basic premises tied to organizational effectiveness, L. Grunig et al.’s (2002) research has already demonstrated how excellent public relations can contribute to organizational effectiveness. Still to be examined is how public relations relationship maintenance programs that help build relationships with publics can contribute to organizational effectiveness.
2
Second, despite the growing interests in organization-public relationships, few scholars have focused on the relationships an organization has with its employees, a most important public to organizational survival (Cutlip, Center, & Broom, 1985; H.-S. Kim, 2003, 2005, 2007; Ni, 2006; Wright, 1995). A primary purpose of this study therefore was to develop a theory of internal relationship management. To that end, I tested a model consisting of relational antecedents (symmetrical and asymmetrical relationship maintenance strategies), congruence of perceived organization-employee relationship characteristics1 and organizational effectiveness outcomes. In so doing, I sought to enhance our understanding of relationship building within organizations, from its maintenance through its characteristics to its consequences. The third problem with current research on organization-public relationships is its asymmetrical nature. Although relationships involve more than one party, very limited research (Christensen, 2005; Kelly, Thompson, & Waters, 2006; Seltzer, 2005; Seltzer & Mitrook, 2009; Shin & Cameron, 2005; Waters, 2007) has been devoted to the measurement of both sides of a relationship. A second purpose of this dissertation thus is to propose a new way of measuring organization-public relationships by looking at both the organization’s and employees’ perceptions of the quality of their relationships. To solve these problems, this study tested a model of internal relationship management (Figure 1) by evaluating publics’ as well as organizations’ perceptions of
Many studies, including Hon and J. Grunig (1999), called dimensions of relationship characteristics as “relational outcomes” or “outcomes of relationships,” which is confusing because these are attributes of the quality of a relationship, rather than results of the relationship. Simply put, they are not effects and the relationship is not the cause. Also, some (e.g., Ki & Hon, 2007b) used “indicators” and “outcomes” interchangeably. To avoid confusion, this study therefore will not use the terms of “relational outcomes” or “outcomes of relationships” and the like.
1
3
relationship quality. I discuss the underlying theory and the conceptualization of each component of the model in the following section.
?
H2 H1a
?
Symmetrical Relationship Maintenance Strategies
Congruence of Perceived Relationship Characteristics
Turnover Intention ?
H3
H1b
Asymmetrical Relationship Maintenance Strategies
Contextual Performance
Figure 1. The structural model of internal relationship management theory. Theoretical Framework The Dialectical Approach Relying on Y. Huang’s (1998) definition, I defined organization-employee relationships as the extent to which an organization and its employee publics trust one another, agree each party has legitimate power to influence, report satisfaction with one another, and commit to one another. The main theoretical framework used in this study was the dialectical approach that consists of four key assumptions, namely, contradiction, change, praxis, and totality (Baxter & Montgomery, 1996). Contradiction is “the dynamic interplay between unified oppositions”, in which oppositions are “tendencies or features of a phenomenon” (Baxter & Montgomery, p. 8). Change arises from contradiction. It is “the interplay of stability and flux” (Baxter & Montgomery, p. 10). Praxis means that individuals are both
4
proactive in making decisions and choices and reactive to their partners’ decisions and choices. Totality refers to the notion that “social phenomena are defined by the relations among their characteristics, not by the characteristics themselves” (Montgomery, 1993, p. 206). Extending this approach to organization-public relationships, I viewed organizations and their publics (e.g., employees) as two “unified oppositions” that interact with and influence each other’s decisions and choices. Their different agendas inevitably lead to the dynamic interplay between them. I further developed a general proposition of relationship management—the dynamic and constant tension and change in organization-public relationships around negotiated interests and goals lead to benefits of varying degrees for the organization and its publics and better understanding of each other. Extant Relationship Models Based on the above proposition and models in the literature (Broom et al., 1997, 2000; Bruning & Ledingham, 1999; J. Grunig & Y. Huang, 2000; Hung, 2007; Toth, 2000), I proposed to test an organization-employee relationship model. Particularly, the two three-stage models proposed by Broom et al. (1997, 2000), and J. Grunig and Y. Huang (2000) inspired the initial conceptualization of this study. Adopting the systems theory, social exchange and resource dependence theory, Broom et al.’s (1997, 2000) relationship model includes antecedents, concept and consequences. Antecedents are the “perceptions, motives, needs, behaviors, and so forth” (e.g., perceptions of uncertain environment, needs for resources, and collective perceptions and expectations) that are “causes in the formation of relationships” (Broom
5
et al., 1997, p. 94). In light of the notion of open systems, Broom et al. (2000) noted that these antecedents are “the sources of change, pressure, or tension on the system derived from the environment” (p. 16), though they did not clarify the meaning of “environment.” Concept represents the attributes of the exchanges or transfers of information, energy or resources in a relationship. Consequences are outputs of such exchanges or transfers (e.g., goal achievement, dependency/loss of autonomy, and routine and institutionalized behavior). Although it is subject to the limitations of the systems theory, social exchange theory, and resource dependence theory, Broom et al.’s model suggests that organizations and publics build relationships to achieve certain outcomes, such as goal achievement, akin to the reasoning of my study. J. Grunig and Y. Huang’s (2000) model comprises situational antecedents, relationship maintenance strategies, and relationship outcomes. The situational antecedents indicate when and how relationships form. Pointing out the inadequacy of the resource dependence theory and social exchange theory in accounting for all the change pressures from the environment, J. Grunig and Y. Huang argued that relationship antecedents are situational. The term situational emphasizes that “publics come and go and change as situations change” (J. Grunig & Y. Huang, p. 35). Once a relationship is forged between an organization and a public, or multiple organizations and multiple publics, organizations develop communication strategies to maintain the relationship. Relationship maintenance strategies include symmetrical as well as asymmetrical strategies. The symmetrical ones are geared towards mutual or joint gains whereas the asymmetrical ones are focused on maximization of self-gains. Examples of symmetrical maintenance strategies are shared tasks and integrative
6
negotiation. Instances of asymmetrical ones are contending and distributive negotiation. Relationship outcomes resultant from the maintenance strategies are control mutuality, commitment, satisfaction, trust, and goal attainment (J. Grunig & Y. Huang, 2000). This model overcomes the inadequacy of the theoretical framework of Broom et al’s (1997, 2000) in explaining all the tensions and pressures from the environment. Nevertheless, it remains unclear what the “environment” means and constitutes in J. Grunig and Y. Huang’s model (2000). In contrast with Broom et al., J. Grunig and Y. Huang regard relationship quality as the end product, which is analogous to assuming that a clothing factory buys silk and thread only to make clothes instead of further selling such clothes for profit. Integrating these two models and using the dialectical approach to internal relationship management, I proposed a three-stage model of organization-employee relationships (see Figure 1), including two antecedents (symmetrical and asymmetrical relationship maintenance strategies), the congruence of perceived relationship characteristics (trust, distrust, satisfaction, control mutuality, and commitment) as the mediator, and organizational effectiveness outcomes (contextual performance and turnover intention). Relationship Maintenance Strategies Consistent with the dialectical approach, I defined relationship maintenance strategies as planned and intentional organizational behavioral attempts or efforts “to keep a relationship in a satisfactory condition” and “to keep a relationship in repair” (Dindia, 2003, pp. 3-4). Researchers debated on whether it is necessary to differentiate these strategies from routine behaviors. I concurred that intent does distinguish strategic
7
from routine relationship maintenance behaviors, but argued that such distinctions are practically difficult and unnecessary. Relationship maintenance strategies encompass symmetrical and asymmetrical ones (Hon & J. Grunig, 1999; J. Grunig & Y. Huang, 2000). Symmetrical maintenance strategies are disclosure or openness, assurances of legitimacy, networking, and advice. The asymmetrical ones include distributive negotiation, avoiding, and compromising/accommodating. They are respectively connected with the two-way symmetrical model and two-way asymmetrical model in public relations. Only a limited number of studies have concentrated on relationship maintenance strategies in public relations research (Ki & Shin, 2005). These studies and research in other areas of study such as interpersonal communication and conflict management have linked relationship maintenance strategies to dimensions of perceived organizationemployee relationship characteristics (trust, distrust, satisfaction, control mutuality, and commitment). Furthermore, a study by Weigel and Ballard-Reisch (1999) of 129 married couples provided evidence that individuals’ use of certain symmetrical maintenance strategies (positivity, openness, assurances of legitimacy, networking, and shared tasks) were positively associated with the couples’ joint marital relationship quality (satisfaction, commitment, and love), where the notion of joint marital relationship quality is similar to my conceptualization of congruence of perceived relationship characteristics. Therefore, by extending these pieces of work, I hypothesized relationship maintenance strategies as antecedents of congruence of perceived organization-employee relationship characteristics in my study; specifically anticipating a positive relationship between symmetrical relationship maintenance strategies and congruence of perceived
8
organization-employee relationship characteristics and a negative relationship between asymmetrical relationship maintenance strategies and congruence of perceived organization-employee relationship characteristics. Congruence of Perceived Organization-Employee Relationship Characteristics Existing public relations literature acknowledges the multidimensionality of the concept of organization-public relationships, proposing four such dimensions—trust, satisfaction, control mutuality, and commitment (e.g., J. Grunig & Hung, 2002; Hon & J. Grunig, 1999; Y. Huang, 2001; Ki & Hon, 2007b, 2007c; Ki & Shin, 2005; H.-S. Kim, 2005, 2007; Y. Kim, 2001; Yang & J. Grunig, 2005; Yang, 2007). Various problems exist with the oft-cited conceptualization of these dimensions by Hon and J. Grunig (1999): 1) Is trust a perception or a behavioral intention? 2) Is trust the opposite of distrust? 3) Is normative commitment an additional component of commitment? 4) Is satisfaction towards the job or the organization or both? 5) Are these four dimensions the only dimensions to measure relationship characteristics? 6) Why has public relations research used only publics’ perceptions to represent organization-public relationship characteristics? I addressed these problems by clarifying the conceptualization of the aforementioned dimensions, elucidating the differences and similarities between trust and distrust, and proposing the notion of congruence of perceived relationship characteristics from the dialectical perspective. For example, I added the new dimension of distrust from research on business management (e.g., Lewicki, McAllister, & Bies, 1998) as an additional component of relationship characteristics because it has been shown as a concept distinct from trust. Trust pertains to integrity, dependability, and competence.
9
Commitment encompasses the components of continuance and affective commitment. I also suggested normative commitment as another component of commitment (see Gruen, Summers, & Acito, 2000). Lastly, I laid out a few key problems with the existing coorientation method in public relations research to assess both the organization’s and its publics’ perceptions of their relationships, and provided an alternative approach. The concept of congruence has been examined for at least three decades by organizational psychologists (e.g., Chatman, 1989; Ostroff & Schulte, 2007; Schneider, 1983, 1987; Vancouver & Schmitt, 1991). Extending their conceptualization of congruence as compatibility between individual and organizational characteristics (e.g., values, beliefs, attitudes, and perceptions), I argued from the dialectical perspective that it is imperative to understand the interaction between organizations and their publics (e.g., employees) in order to best capture the essence of organization-public relationships. Congruence of perceived organization-public relationship characteristics is not exactly the same as consensus or similarity, but reflects the dynamic interaction and process of interdependence between unified oppositions, namely, forces such as an organization and its publics. It changes and adapts over time. However, at a given point in time, compatibility and/or consensus may emerge. Based on the person-organization fit literature in organizational psychology, I operationalized congruence as five variables: the perceptions of employees (one opposition) as the person variable, the perceptions of the organization (another opposition) as the organization variable, the squared person variable, the squared organization variable, and the product of the person and organization variable (Edwards, 1994, 1995; Ostroff, Atwater, & Feinberg, 2004). Considering that this was the first study that adopted such an approach to studying
10
organization-public relationship characteristics, I emphasized its exploratory nature in examining the ways in which non-management employees and managers perceive their organization-employee relationships although I used research from other disciplines to support my reasoning. Organizational Effectiveness Outcomes One pivotal development in the relationship paradigm was the identification of the linkage between organization-public relationship quality and public attitudes, perceptions, and behavior (Ledingham, 2003). Yet scant attention has been devoted to the behavioral consequences of organization-public relationship quality (Ki & Hon, 2007b). To add to this line of research, I focused on organizational effectiveness outcomes. Organizational effectiveness lies at the heart of most organizational research (Robbins, 1990). L. Grunig et al.’s (2002) landmark study of public relations excellence also began with the quest of organizational effectiveness. They suggested that the public relations function contributes to organizational effectiveness by building quality, longterm relationships with strategic publics. One such strategic public is the employee public. Thus, building upon this body of research on public relations excellence, I proposed that the public relations function may demonstrate its contribution to organizational effectiveness through building quality relationships with employees. By considering the goal attainment approach, the systems approach, the strategic constituencies approach, and the competing values approach to studying organizational effectiveness (Robbins, 1990), I suggested an integration of these approaches and thus presenting my definition of organizational effectiveness—the degree to which an
11
organization incorporates the values and demands of its strategic constituencies into its long-term (ends) and short-term (means) goals that it achieves. Employees are one of an organization’s strategic constituencies. As Campbell (1977) demonstrated, employees’ turnover and productivity are two key criteria of organizational effectiveness. Furthermore, these two criteria have been widely studied in research on human behavior in organizations (e.g., Borman & Motowidlo, 1993, 1997; Goodman & Svyantek, 1999; J. Hogan, Rybicki, Motowidlo, & Borman, 1998; Motowidlo, 2000; Motowidlo & Van Scotter, 1994; Van Scotter, 2000). Lastly, relationship management researchers such as Ledingham (2003) have uncovered evidence linking organization-public relationship quality to publics’ loyalty, similar to the intention to stay with an organization by employees, namely, turnover intention. Therefore, I chose turnover intention and contextual performance as two behavioral outcomes of organizational effectiveness in this study. The term contextual performance was developed to tap non-task-related performance by employees, which is often not part of their formal job responsibilities (Borman & Motowidlo, 1993). It overlaps with organizational citizenship behavior (Bateman & Organ, 1983). Examples of contextual performance include assistance to coworkers and initiatives to support organizational objectives (Borman & Motowidlo, 1997). On the other hand, turnover intention has been conceived of as deliberate willingness to leave an organization (Tett & Meyer, 1993). I operationalized it as intention to leave within a particular time interval. Based on and extending research from organizational behavior and business management, I anticipated that the level of congruence of perceived organization-
12
employee relationship characteristics is linked to the above two organizational effectiveness outcomes (e.g., Allen & Meyer, 1990; Z. Chen & Francesco, 2000; Goodman & Svyantek, 1999; Morgan & S. Hunt, 1994; Saks & Ashforth, 2002; Schneider, 1983, 1987; Vancouver & Schmitt, 1991; Van Vianen, 2000). Method and Research Design The online survey method was used in this study due to its efficiency, convenience, and ensured confidentiality and anonymity. A combination of random sampling and non-probability sampling was employed in the selection of participating organizations. As theoretical generalizability other than statistical generalizability was the focus, non-probability sampling was appropriate (Shapiro, 2002). Undergraduate students in the Department of Communication were paid to recruit organizations during the non-probability sampling stage. The same template email was used for organization recruitment, containing information on the purpose of the study, the investigators, expected participation time, and potential harm and benefits, among other content that has been approved by the Human Subjects Review Board at the University of Maryland. Each organization was provided a separate link to the same questionnaire. Template reminder emails were sent each week following the indication of willingness to participate, until at least 25 participants were recruited from a particular organization. Confidentiality and anonymity were ensured, with the exception that the organization names were known to the researcher and the respective student recruiter for an organization. Forty-eight organizations participated in the study, but 18 of them were excluded from data analyses because of the small number of responses (< 10) from each
13
organization. The organizations were in different industries and of diverse sizes to ensure sufficient variance between organizations. A total of 1293 participants agreed to participate. The final sample size was 785, consisting of 195 managers, 583 nonmanagerial employees, and seven with no indication of position level. Willing participants (n = 263) were entered in a raffle to reward their participation. Significance of the Study This study contributed to the continued search for the value of public relations. As many have contended, the value of public relations lies in relationship management. But what comes into relationships? How do employees form relationships with their organizations? What do organizations do to maintain relationships with their employees? What do we build relationships for? How can relationship management contribute to the overall organization effectiveness? This study provided answers to these questions, important to research and practice alike. Second, despite the burgeoning body of literature on relationship management, it remains a fragmented area fraught with research that often does not explicitly state its conceptualization of relationship quality. Conceptual clarity is necessary for theory development (Dindia, 2003). Taking the dialectical approach and relying on congruence research in organizational psychology, this study clarified the conceptualization of organization-public relationships, its operationalization and measurement, and its antecedent and outcomes, thus potentially advancing relationship research to a higher theoretical level. Third, this study refined existing measurement of organization-public relationships by distinguishing trust from trustworthiness and distrust, adding new
14
dimensions to the current widely used measuring instrument by Hon and J. Grunig (1999), problematizing the coorientation method, and proposing a new way to evaluate two sides of the relationship. In so doing, it moved relationship management research to a new operational level. Fourth, this study provided insights into the process of how organizations build relationships with one of their most important publics—employees—by developing a theory of internal relationship management. The study demonstrated the ways in which organizations’ relationship maintenance strategies influence the congruence of organizations’ and their employees’ perceptions of relationship characteristics, which eventually impacts organizational effectiveness outcomes—employees’ contextual performance and turnover intention. Few studies on relationship management have presented a comprehensive view of this process.
15
Chapter II Conceptualization J. Grunig and T. Hunt (1984) defined public relations as the “management of communication between an organization and its publics” (p. 6), which represents the strategic management paradigm of public relations. As part of this paradigm, the excellence study sought to determine how public relations makes an organization more effective, how much this contribution is worth economically, and the characteristics of this public relations function (L. Grunig et al., 2002). This study that spanned 17 years and surveyed three countries found that public relations can make an organization more effective “when it identifies the strategic publics that develop because of the consequences that organizations and publics have on each other” and “when it uses symmetrical communication programs to develop and maintain quality long-term relationships with these strategic publics” (L. Grunig et al., 2002, p. 548). In other words, the value of public relations to organizations lies in the relationships it builds with strategic publics. Built upon the excellence study and extant literature on relationship management, this study posited that the public relations function uses symmetrical relationship maintenance strategies to develop quality relationships between organizations and their employees, which consequently leads to organizational effectiveness, as assessed by employees’ turnover intention and contextual performance. In this section, I first provide conceptualization of organization-public relationships, an explanation for my focus on organization-employee relationships, and a review and critique of current research on relationship management. Next, I offer conceptualization of focal constructs: relationship maintenance strategies (symmetrical
16
and asymmetrical), perceived relationship characteristics, congruence of perceived relationship characteristics, organizational effectiveness, contextual performance, and turnover intention; and reviews of relevant literature. I conclude with review of literatures linking the focal constructs and introduction of the conceptual model. Defining Organization-Public Relationships Public relations as a scholarly discipline has struggled for an identity that distinguishes it from marketing, advertising, and journalism (Bruning & Ledingham, 1999; Cheney & Christensen, 2001; L. Grunig et al., 2002). Since more than two decades ago when Ferguson (1984) first advocated a relational perspective of public relations, a multitude of research has examined organization-public relationships. Nearly two decades later, Ledingham (2003) observed “pivotal developments which spurred emergence of the relational perspective as a framework for public relations study, teaching, and practice” (p. 286). He reasoned that relationship management has become a general theory that “unifies … [the public relations] discipline” (Ledingham, 2003, p. 192). He further concluded, “The relational theory of public relations also serves as the foundation for a relationship-grounded research paradigm” (Ledingham, 2003, p. 193). Under the umbrella of this relational theory are a wide range of relational dimensions, types, and models identified by researchers. Despite the celebration of the emerging research on relationship management, it remains a fragmented paradigm according to Ledingham and Bruning (2000) and Ledingham (2003). To begin with there still is no agreed-upon definition of “relationship” (Broom et al., 1997; Ki & Shin, 2005). Ledingham and Bruning (1998a) referred to an organization-public relationship as a state that “exists between an
17
organization and its key publics, in which the actions of either can impact the economic, social, cultural or political well being of the other” (p. 62). Broom et al. (2000) conceptualized it as characteristics of relationship-related activities—“the patterns of interaction, transaction, exchange, and linkage between an organization and its publics” (p. 18). Thomlison (2000) stated that a relationship is “a set of expectations two parties have for each other’s behavior based on their interaction patterns” (p. 178). Y. Huang (1998) defined organization-public relationships from the perspective of its characteristics, referring to relationships as “the degree that the organization and its publics trust one another, agree [each party] has rightful power to influence, experience satisfaction with each other, and commit oneself to one another” (p. 12). She regarded organization-public relationships as relationships between a corporate person and another corporate person or between a corporate person and a group of individuals (Y. Huang, 1997a). Hung (2005) developed her conceptualization of relationships from the perspective of systems theory, resource dependence theory and exchange theory. She stated, “[organization-public relationships] arise when organizations and their strategic publics are interdependent, and this interdependence results in consequences to each other that organizations need to manage constantly” (Hung, p. 396). Most of these definitions have one thing in common—they all defined a concept by laying out its characteristics, exemplified by sets of expectations, or specific properties such as trust, power, satisfaction, commitment (Y. Huang, 2001), or interdependence (Hung, 2005). In a similar vein, Wood’s (1995) exhaustive review of more than 700 articles and books on relational communication led her to conclude with four dimensions of characteristics of relationships—investment, commitment, trust, and comfort with
18
relational dialectics. Another interpersonal communication researcher Wilmot (1980) also used two dimensions of characteristics to define the nature of dyadic relationships: love/hate, and dominant/submissive. Broom et al.’s (2000) extensive review further revealed other dimensions of characteristics of relationships that interpersonal communication scholars commonly use. In addition, Ledingham and Bruning (1998a) cited research from marketing and social psychology that proposed various dimensions of relationship characteristics. Relying on these bodies of literature from different disciplines, this study thus proposed to conceptualize organization-public relationships from the perspective of its characteristics, and used Y. Huang’s (1998) definition of organization-public relationships as a tentative one: “the degree that the organization and its publics trust one another, agree [each party] has rightful power to influence, experience satisfaction with each other, and commit oneself to one another” (p. 12). In addition to the lack of a clear definition, researchers disagreed on whether an organization-public relationship is a subjective reality or an objective entity. Some scholars (Duck, 1973, 1986; J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999; Y. Huang, 1997a, 2001; Ki & Hon, 2007b; Ledingham & Bruning, 1998) conceived of it in a subjective way. Duck (1973) considered relationships inseparable from relational partners’ cognition and values. Others (Broom et al., 1997, 2000; Katz & Kahn, 1967; Oliver, 1990; Rogers, 1998) viewed it in an objective manner, independent of relational partners. Rogers stated, “The relationship lies in the connection [of relational partners], not in one or the other but in the between” (p. 77). She further suggested, “Relationships are socially performed, constructed, maintained, and altered in the reciprocal actions of
19
the relational members….Relationships do not have feelings, make imputations, engage in sense making, and the like; only the [relational] members have these capacities” (p. 78). Still others (Andersen, 1993; Cappella, 1991) saw a relationship as a combination of subjective perceptions and objective qualities independent of relational partners. Andersen argued, “Relationships are the combined product and producers of both the interpersonal interactions and the cognitive activity of the interactants” (p. 2). Most empirical research on organization-public relationships in the public relations literature to date has measured organization-public relationships as perceptions. Ki and Hon (2007b) reasoned that it is no surprise to treat relationships as relational parties’ perceptions. Broom and Dozier (1990) proposed a coorientational approach to measure both an organization’s and its publics’ perceptions. Hon and J. Grunig (1999) also asked for perceptions of an organization’s key publics. Only Broom et al. (2000) proposed assessing organization-public relationships as independent of relational members. They used a third party to observe organizational members’ interaction with each other. These observations are perceptions and judgments of the third party. Therefore, this study defined organization-public relationships as relational members’ perceptions of their relationship characteristics. Why Organization-Employee Relationships? Cutlip et al. (1985) pointed out that organization-employee relationships are the most important among all the relationships organizations have with their publics. Holtzhausen (2002) also noted that an important aspect of public relations managers’ job responsibilities is to maintain quality relationships with internal publics—employees. J. Grunig (1992) and L. Grunig et al. (2002) discussed in great detail the significance of
20
internal communication for organizational excellence. Rhee (2007) made a similar point citing literature from internal marketing. She wrote, “According to the proponents of internal branding, a corporation must first align its employees with its brand before it can make brand promises to its customers” (p. 8). Likewise, Guaspari (2002) claimed that internal marketing can retain employees whose increased loyalty then helps organizations build better relationships with external publics. It appears well understood that organizations cannot be what they are without support from their employees. Notwithstanding the significance of organization-employee relationships, Sallot, Lyon, Acosta-Alzuru, and K. Jones (2003) assessed the articles found in three major public relations journals from inception through 2000, and found that only two percent of the total number of published studies examined organization-public relationships. Among the two percent, even fewer focused on organization-employee relationships. Public relations scholars have continued to express concerns that researchers and practitioners alike have neglected employees as a type of public, assuming that their loyalty and commitment can always be counted on (D’Aprix, 1984; Hung, 2002; H.-S. Kim, 2003, 2005; Ni, 2006; Wright, 1995). It is clear that little is known of organization-employee relationships—a key area of research in the field of public relations, with the exceptions such as Rhee (2004), Ni (2006), and H.-S. Kim (2005). This study sought to provide more insights into this area. Research on Relationship Management Ledingham (2003) identified four pivotal developments of current research on organization-public relationships, which were: (1) the “recognition of the central role of relationships in public relations” exemplified by Ferguson’s (1984) article; (2)
21
reconceptualization of public relations as a strategic management function; (3) “identification of components and types of organization-public relationships, their linkages to public attitudes, perceptions, knowledge and behavior, and relationship measurement strategies”; and (4) “construction of organization-public relationship models that accommodate relationship antecedents, process, and consequences” (pp. 182183). These pivotal developments included articles and book chapters on definitions of organization-public relationships reviewed in preceding sections; application of the relational perspective to different contexts; maintenance strategies; and dimensions, types, and models of organization-public relationships. Application of the Relational Perspective The relational perspective has been used by organization-public relationships researchers in contexts such as public affairs, community relations, employee relations, media relations, manufacturer-retailer relations, university-student relations, issues management, and crisis communication (e.g., Bridges & Neilson, 2000; Coombs, 2000; Jo, 2003; Ki & Hon, 2007b; Ledingham & Bruning, 1998b; Ni, 2006; L. Wilson, 2000). For instance, Coombs (2000) integrated stakeholder theory and attribution theory as an organizing framework for a relational approach to crisis management. He suggested the ways in which existing stakeholder-organization relationships influence crisis managers’ selection of crisis response strategies. L. Wilson (2000) used a case study of Novell, Inc. to examine the linkage between perceptions of dimensions of organization-public relationships and a behavioral outcome loyalty. She argued that the company’s local community publics and employee publics expressed their loyalty because they perceived openness, trust, commitment and investment in the company’s actions.
22
Relationship Maintenance Strategies Relationship maintenance was commonly defined as “to keep a relationship in a satisfactory condition” and “to keep a relationship in repair” (Dindia, 2003, pp. 3-4). Existing public relations literature derived two types of relationship maintenance strategies—symmetrical and asymmetrical ones—from the interpersonal communication literature (e.g., Y. Huang, 1997a; J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999; Hung, 2007). The symmetrical strategies encompass disclosure/openness, assurances of legitimacy, participation in mutual networks, shared tasks, integrative negotiation, cooperation/collaboration, being unconditionally constructive, and win-win or no deal. The asymmetrical ones include distributive negotiation, avoiding, contending, compromising, and accommodating (J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999). There is a paucity of research focusing on relationship maintenance strategies (Hung, 2007; Ki & Hon, 2007a; Ki & Shin, 2005). Only a handful of studies examined relationship maintenance strategies through qualitative interviews, case studies, and content analysis of company Web sites (Bortree, 2003; Hong & Kiousis, 2007; Hung, 2003, 2004; Hung & Y. Chen, 2007; Kelleher & Miller, 2006; Ki & Hon, 2003, 2006, 2007a; Wigley, 2003). These studies verified several strategies theorized by Hon and J. Grunig (1999) and J. Grunig and Y. Huang (2000). For example, Ki and Hon (2006) discovered in an analysis of Fortune 500 companies’ Web sites that openness and access are two most commonly employed strategies by these organizations. It appears that the long list of strategies proposed by Hon and J. Grunig (1999) was often accepted as given. However, several questions remain, such as: Do all of the
23
strategies exist in all kinds of relationships organizations have with their various publics or does it depend on the type of publics? Do the symmetrical strategies have the same effect on quality of organization-public relationships as the asymmetrical ones? As researchers posited that relationship maintenance strategies are predictive of quality of organization-public relationships (J. Grunig & Y. Huang, 2000; Hon & J. Grunig), it is imperative to further understand the process of relationship maintenance. A first step can be looking into the above questions. Dimensions, Types, and Models of Organization-Public Relationships Dimensions. According to Ki and Shin’s (2005) analysis of published articles on organization-public relationships, dimensions of relationship quality were a primary focus of existing research on relationship management. Ferguson (1984) first proposed some dichotomous dimensions of quality of organization-public relationships, such as dynamic versus static, open versus closed, and satisfactory versus unsatisfactory. L. Grunig et al. (1992) theorized six dimensions: reciprocity, trust, mutual legitimacy, openness, mutual satisfaction, and mutual understanding. Y. Huang (1997a) combed through public relations models and conflict resolution strategies and developed four dimensions of relational quality in her dissertation: trust, control mutuality, relational commitment, and relational satisfaction. Ledingham and Bruning (1998a) presented five dimensions: trust, openness, involvement, investment, and commitment. Based on Y. Huang’s work, Hon and J. Grunig’s (1999) report on how to measure relationship quality in public relations outlined six dimensions: trust, commitment, satisfaction, and control mutuality, communal and exchange relationships. Len-Rios (2001) used these six dimensions in two online surveys to look at online organization-public relationships. J.
24
Grunig and Y. Huang (2000) contended that trust, commitment, satisfaction, and control mutuality (excluding the two types of relationships) seem to best represent the essence of organization-public relationships. Y. Huang (2001) produced a cross-cultural, multi-item scale of quality of organization-public relationships and added face and favor as a fifth dimension in addition to the four dimensions originally in her dissertation. Y. Kim (2001) suggested four relational dimensions: trust, commitment, local and community involvement, and reputation. Among the different versions of dimensions of the quality of organization-public relationships, J. Grunig and Y. Huang’s (2000) four dimensions were widely adopted in different studies. Ki and Shin’s (2005) review of 38 published articles in a dozen major academic journals from 1990 to 2005 revealed that satisfaction, commitment, trust, mutual understanding, control mutuality, and benefit were primary indicators of relationships reported by researchers. A key problem with the extant research on dimensions of the quality of organization-public relationships is that despite the different versions of dimensions and measures of these dimensions, only a few studies have measured the reliability and validity of the proposed scales (Y. Huang, 2001; Ki & Hon, 2007c; Y. Kim, 2001). Secondly, relationship quality has been treated as a one-sided concept, i.e., quality as perceived by publics, which is ironic considering the name organization-public relationships. Research is needed to address these problems. Types. Fewer studies have focused on the types of organization-public relationships (Hon & J. Grunig, 1999; Hung, 2002, 2005, 2007). Hon and J. Grunig distinguished communal relationships from exchange relationships. In communal
25
relationships, relational parties offer benefits to one another out of concerns for their welfare, expecting nothing in return. In exchange relationships, parties provide benefits to one another to return the favor they received in the past. The level of trust, control mutuality, satisfaction, and commitment that was developed in exchange relationships are not comparable to that in communal relationships. Most relationships begin as exchange relationships and then mature to communal ones (Hon & J. Grunig). Hung (2005, 2007) proposed six more types of relationships: exploitative relationships, manipulative relationships, symbiotic relationships, contractual relationships, convenantal relationships, and mutual communal relationships. Exploitative relationships happen when one relational party takes advantage of the other or does not fulfill his or her obligation when expected to. Manipulative relationships arise when an organization uses asymmetrical or pseudo-symmetrical communication on publics to serve its own interests after learning about the actual needs and interests of publics. Symbiotic relationships develop when organizations and publics work towards mutual goals (such as survival), with the knowledge of their interdependence in the environment and awareness of the consequences of their behavior on one another. Contractual relationships occur when relational parties come to an agreement on what each should do in the relationship, analogous to signing a contract. Such relationships are not necessarily equal; power is a salient factor. Convenantal relationships are characterized by both parties committed to a common good through open exchanges of communication (e.g., suggestions and criticisms). The norm of reciprocity is applicable in such relationships. An example of convenantal relationships is between a teacher and students. Mutual communal relationships are those in which relational parties are concerned about the
26
welfare of the other. This focus on psychological intention differentiates mutual communal relationships from convenantal relationships as the latter emphasize open exchanges. Hung (2005) argued that Hon and J. Grunig’s (1999) communal relationships should be called “one-sided communal relationships,” which are more sophisticated than mutual communal, because of the expectation of nothing in return (reciprocity becoming unnecessary). Hung (2005) cited her personal communication with J. Grunig and explained that J. Grunig developed a continuum of types of relationships ranging from concern for oneself (exploitative relationships) to concern for others (communal relationships). As the continuum moves towards concern for others, there exist contractual relationships, exchange relationships, and convenantal relationships. Hung (2005, 2007) modified this continuum: with exploitative relationships on the end of “concern for self interests,” followed by manipulative relationships, contractual relationships, symbiotic relationships, exchange relationships, convenantal relationships, and mutual communal relationships, and one-sided communal relationships on the end of “concern for others’ interests.” She called exchange relationships, convenantal relationships, and mutual communal relationships the “win-win zone” for relational parties. Models. There are several models of organization-public relationships in the public relations literature (Broom et al., 1997, 2000; Bruning & Ledingham, 1999; J. Grunig & Y. Huang, 2000; Hung, 2007; Toth, 2000). Broom et al. (1997, 2000) were among the first to develop such a model. They suggested a three-stage model of relationship management with antecedents, concept and consequences. Posited as causes of the formation of relationships, antecedents include social and cultural norms,
27
collective perceptions and expectations, needs for resources, perceptions of uncertain environment, and legal/voluntary necessity. Representing the process of relationship formation, concept is comprised of properties of exchanges, transactions, communication, and other interconnected activities. As outputs of the relationships, consequences are goal achievement, autonomy, routine and institutionalized behavior. Broom et al. noted that in the long run consequences eventually will result in changes in the antecedents, which then influence the concept. Concept ultimately leads to changes in consequences. Relying on the strategic management theory and the excellence theory, J. Grunig and Y. Huang (2000) developed a parallel three-stage model consisting of situational antecedents, maintenance strategies, and relationship outcomes. The situational antecedents describe how relationships come into being, typically when organizations and publics have consequences on each other, such as organization(s) affecting a public/publics and vice versa. The middle part of the model is maintenance strategies or specific activities to maintain a relationship, which are categorized as symmetrical and asymmetrical. These strategies are derived from the interpersonal communication literature, particularly work by Stafford and Canary (1991), as well as the conflict resolution literature that is applicable to public relations, based on Y. Huang’s (1997a) and Plowman’s (1995) research. The symmetrical strategies are posited as more effective than the asymmetrical ones, analogous to the symmetrical and asymmetrical model in the excellence theory by L. Grunig et al. (2002). Examples of the symmetrical strategies include disclosure (openness), shared tasks (helping to solve problems of interest to the other party), and integrative negotiation (maximizing joint gains); whereas the asymmetrical ones consist of distributive negotiation (maximizing self-gain), avoiding,
28
and contending (trying to convince the other to accept one’s position). Relationship outcomes resulting from such maintenance strategies are control mutuality, commitment, satisfaction, trust and goal attainment. Toth (2000) presented and contrasted a “pure personal influence” type of public relations practice in which interpersonal communication is utilized to “dominate individuals, to accept either the organization’s or public’s position” and a “pure interpersonal influence” type of practice in which interpersonal communication is used to “influence to find mutual definitions, mutuality of understanding, agreement, [and] consensus” (p. 214). The former is “closed and static in attributes” whereas the latter is “open and dynamic in attributes” (Toth, p. 214). She proposed the adoption of the “pure interpersonal influence” model in relationship management. Bruning and Ledingham (2000) developed a five-step process of relationship management: scan, map, act, rollout, and track (SMART). Akin to J. Grunig and Y. Huang’s (2000) notion of environment scanning, scan entails “surveil[ling] the environment to better understand the current state of organization-public relationships and the communication patterns that exist in those relationships, as well as to determine the current state of key public member knowledge, opinions, attitudes, and behaviors” (p. 92). Map means working out a strategic plan with strategies, tactics, objectives, and goals. Act includes conducting a field test of the strategic plan to make modifications as needed. Rollout is implementing the modified strategic plan on key publics. The last step track refers to tracking the influence of the organization’s activities on key publics’ perceptions and behaviors through the use of both quantitative and qualitative methods.
29
Incorporating both the social exchange theory and the dialectical perspective, Hung (2007) developed a relationship management model that views relationship as an ongoing dynamic process, in which each stage of relationship development is subject to changes of the previous stage. Her model consists of relationship objectives, cultivation, development and change strategies, and relationship outcomes. Hung (2007) posited that the types of relationships (under relationship objectives) determine the types of cultivation, development and change strategies, i.e., symmetrical versus asymmetrical, the choice of which affects relationship outcomes. Relationship objectives have four components. The first component is the type of relationships an organization intends to have with its publics resulting from its acknowledgement of interdependence with publics and subsequent intention and motivation to survive in the environment. She listed eight types of possible relationships based on Hung (2005)—one-sided communal, mutual communal, convenantal, exchange, manipulative, contractual, symbiotic, and exploitative relationships. The second component is the quality of relationships that the organization has with its publics in the previous stage (trust, control mutuality, commitment, and satisfaction). The third and fourth components respectively are goal attainment and reputation perceived by the publics. Cultivation, development and change strategies consist of the symmetrical and asymmetrical categories. The symmetrical ones are openness/disclosure, positivity, legitimacy, networking, sharing tasks, keeping promises, cooperation, being unconditionally constructive, and win-win or no deal. The asymmetrical ones include distributive negotiation, avoiding, compromising, contending, and accommodating.
30
Relationship outcomes comprise the quality of relationships (trust, control mutuality, commitment, and satisfaction), the extent of goal attainment, reputation perceived by publics, and types of relationships that may differ from the organization’s initial expectation and intention. They are the same components of relationship objectives. Conclusion Ledingham (2003) contended that the above reviewed four pivotal developments of current research on how organization-public relationships contribute to a theory of relationship management. He summarized the findings of extant research on relationship management as the following: 1. In the relational perspective, relationship presuppositions act as a framing mechanism for theory building, teaching, and practice. The use of communication output as the measure of programmatic accountability is superceded by measurement of relational and behavioral outcomes. 2. The appropriate unit of measurement of public relations impact is the organization–public relationship. 3. Analysis of organization–public relationships is grounded in interpersonal relationship building. Moreover, many interpersonal relationship principles— complete with guidelines for initiating, maintaining, and improving relationships—serve as a foundation for the exploration of organization–public relationships. 4. Organization–public relationships involve an ongoing interchange of needs, expectations, and fulfillment.
31
5. Ratings of relationship dimensions can define the state of an organization– public relationship, which, in turn, can act as a predictor of public perceptions and choice behavior. 6. Organization–public relationship types include symbolic and behavioral, as well as personal, professional, and community-related. 7. Organization–public relationships change over time. 8. The outcome of effective relationship management is mutual understanding and benefit. 9. Successful organization–public relationships develop around common interests and shared solutions to common problems. 10. Relationship state reflects perceptions of needs and expectations fulfillment. 11. Mutual benefit strategies can generate economic, societal, and political gain both for organizations and publics. Based on these findings, Ledingham (2003) argued that a theory of relationship management has emerged, which is that “Effectively managing organizational[sic]-public relationships around common interests and shared goals, over time, result in mutual understanding and benefit for interacting organizations and publics” (p. 190). In other words, the value of public relations exists in effective management of relationships. Ledingham’s (2003) piece was commendable with regard to providing a useful framework for research on relationship management. Nevertheless, it was rather vague what he meant by “common interests and shared goals”, and “mutual understanding and benefit” and what “effective management” entailed. Furthermore, although Ledingham explained that this theory of relationship management is a general theory that
32
accommodates middle-range theories, such as using the situational theory of publics by J. Grunig (1997) to identify and categorize publics, it was not clear what specific middlerange theories can be considered as components of the general theory of relationship management. Toward A Theory on Organization-Employee Relationships As is shown above, despite the burgeoning research on relationship management, more theory-building endeavors are necessary. Because of the interdisciplinary nature of the research on relationship management, it is important to first lay out the theoretical foundations. Extant research has relied on the systems theory, resource dependence theory, social exchange theory, and relational dialectics theory as their frameworks (Broom et al., 1997, 2000; J. Grunig & Y. Huang, 2000; Y. Huang, 1997; Hung, 2007). Systems Theory A system is “a set of interacting units with relationships among them” (Miller, 1978, p. 16). Organization theorists using the systems theory by Bertalanffy see organizations as open systems, characterized by “a continuous cycle of input, internal transformation (throughout), output, and feedback (whereby one element of experience influences the next)” (G. Morgan, 1997, p. 40). They emphasize the environment in which organizations exist, define organizations in terms of interrelated subsystems, and seek “to establish congruencies or ‘alignments’ between different systems and to identify and eliminate potential dysfunctions” (G. Morgan, p. 39). For instance, Lee and Jablin (1995) contended that work relationships are subsystems nested within larger organizational systems. Likewise, Waldron (2003) argued:
33
Any given work relationship is nested within a complex system of vertical and horizontal networks. The communication of co-worker peers is influenced in part by their individual relationships with those in power and their perception of the supervisory relationships their peers enjoy. (p. 165) Two relevant concepts from the systems approach are interdependence and homeostasis (Broom et al., 1997, 2000; Vogl-Bauer, 2003; Waldron, 2003). Interdependence means that the environment and the organizations are seen interacting with and dependent upon each other. The essence of such interactions is mutual adaptation (Broom et al., 2000, p. 15). The notion of interdependence has been widely used in research on relationship management (Broom et al., 1997, 2000; L. Grunig et al., 1992; Hung, 2005; Yang, 2005). Hung defined organization-public relationships as interdependence, assuming “when an organization realizes the interdependence with its publics, it either competes or collaborates with its publics in acquiring the resources for its survival” (p. 396). Homeostasis is “self-regulation and the ability to maintain a steady state” (G. Morgan, 1997, p. 40). Systems use homeostatic control processes to sustain themselves. With regard to relationship management, relationship maintenance strategies are examples of such homeostatic control processes by systems to stay in a steady state (Dindia & Canary, 1993; Lee & Jablin, 1995). Nonetheless, the problem with homeostasis is its association with stability, negating the fact that social relationships involve both stability and change (Baxter & Montgomery, 1996). A complete acceptance of stability makes it unlikely for relationships to grow or adapt to new circumstances.
34
Resource Dependence Theory Resource dependence theory was one of the most influential theories of organization-environment relations by Pfeffer and Salancik (Hatch, 1997). The assumption is that organizations need resources from their environment for survival, growth and achievement of goals, such as raw materials, labor, capital, and technology (Hatch, 1997; Broom et al., 2000). This is consistent with the notion of interdependence of the systems theory. The basic argument of the theory is that managers’ knowledge of the power/dependence relationships between their own organization and the network it is in can help the organization anticipate sources of influence from the environment and work out ways to offset such influence (Hatch, p. 78). Often managers are required to sort the resources they need according to their criticality and scarcity. Both critical and scarce resources should have the highest priority (Hatch). Furthermore, managers can help organizations develop sustained competitive advantage by creating or obtaining resources that are difficult to be imitated by competitors (Barney, 1991). One example of managing labor and knowledge dependence is organizational strategies to recruit executives from competitors (Hatch). This theory has been applied to understanding inter-organizational relationships. Van de Ven (1976) posited that organizations form relationships with one another because they depend on each other for resources. Specifically, an organization may form a relationship with another because the resources it needs are scarce, or it needs certain specialized skills or services, or because it competes with another organization for clients, resources, etc (Van de Ven & Walker, 1984).
35
Employing this theory to organization-employee relationships, one can see that organizations need labor and knowledge from employees for their survival and growth. Organizations’ strategies to maintain their relationships with employees are efforts to counter the dependence on one resource—employees, and to develop unique competitive advantage over competitors. Social Exchange Theory Originated in psychology and sociology, social exchange theory was used for relational communication (Liska & Cronkhite, 1995; Roloff, 1981). The theory holds that social relationships involve exchanges of relationships, such as status, information, services, goods, money, love, companionship, friendship, and social acceptance (Devito, 1996, Hinde, 1997). People develop relationships to maximize profits. Profit is rewards minus costs. Usually individuals in a relationship have certain expectations of the kind of profits to be obtained from the relationship, referred to as comparison level. The theory suggests that whether a person will stay in a relationship depends on his or her comparison level for alternatives. Individuals compare current relationship profits with anticipated profits from alternative relationship choices. In essence, individuals tend to be less satisfied if they have many relationship options available, unless their expectation of the current relationship is met or exceeded (Thomlison, 2000). When applied to relationship maintenance, the activities to cultivate a relationship can be conceived of as costs. If the comparison level as well as the comparison level for alternatives can be determined, organizations then may predict the amount of rewards necessary to maintain a relationship (Thomlison, 2000). According to Thomlison, public
36
relations practitioners should be constantly monitoring the comparison level and the comparison level for alternatives of their publics. This theory is similar to resource dependence theory in its emphasis on exchanging resources. Nevertheless, with regard to explaining the formation of relationships, J. Grunig and Y. Huang (2000) contended that neither the resource dependence theory nor the social exchange theory can account for all the change pressures from the environment. They stated: These [change] pressures may have little or nothing to do with resources or with exchanges. Instead, they may come from publics, activist groups, government, or media who simply want to change the behavior of the organization in specific situations, thus interfering with the behavior the organization wants to implement. (p. 35) Further, J. Grunig and Y. Huang (2000) claimed, “[T]he antecedents of relationships are situational, just as publics are situational … publics come and go and change as situations change” (p. 35). The Dialectical Approach The systems theory, resource dependence theory and social exchange theory all emphasize the notion of interdependence, a defining nature of organization-public relationships (including organization-employee relationships). However, the notion of environment is too ambiguous in the systems theory. Moreover, the implication of stability by the concept of homeostasis in the systems theory neglects the fact that social relationships are characterized by both stability and change. The problem with resource dependence theory and social exchange theory is that employees are not just resources.
37
Employees’ decisions to enter or exit the relationships with their organizations may not be simply a calculation of rewards minus costs. Therefore, I relied on the dialectical approach that embraces the notion of interdependence but also better captures the quality of organization-public (including employees) relationships. Applying Marx’s concept of dialectic and Bakhtin’s extension of Marxist dialectic—dialogism, Baxter, Montgomery, and Rawlins proposed the dialectical approach to relationship studies (Baxter, 1988; Baxter & Montgomery, 1996; Montgomery, 1993; Rawlins, 1989, 1994). These researchers view the dialectical approach as a perspective, other than a theory of prediction and causal explanation (Baxter & Montgomery). It does not describe an array of axioms or propositions; rather, it provides a set of assumptions—contradiction, change, praxis, and totality (Baxter & Montgomery). The dialectical approach maintains that change, tension, and instability are ever present qualities of social relationships. Relational partners spiral between opposing yet interdependent communicative tendencies, in contrast with the traditional linear model that expects individuals to seek a certain state of satisfaction. That is, “partners are constantly being pulled together and simultaneously being pushed apart” (Dainton, Zelley, & Langan, 2003, p. 83). The dialectical approach consists of four key assumptions: contradiction, change, praxis, and totality (Baxter & Montgomery, 1996). Without the traditional negative connotation, contradiction is defined as “the dynamic interplay between unified oppositions”, in which oppositions are “tendencies or features of a phenomenon” (Baxter & Montgomery, p. 8). A contradiction is formed when two opposing yet interdependent
38
forces or tendencies mutually negate each other. One cannot understand one force without understanding the other (Dainton et al., 2003). Dialectical theorists commonly use contradiction as the unit of analysis in their research. Extending this to organizationpublic relationships, oppositions can be regarded as opposing yet interdependent relational parties that have shared as well as opposing interests. A contradiction then is the dynamic interaction between these opposing relational parties who negate one another at the same time that they are interdependent on one another. It resembles L. Grunig et al.’s (2002) idea of symmetry—not consensus, but “the give-and-take of persuasion and collaboration that organizations and publics use when they must interact with each other” (p. 551). This concept also fits squarely with the notion of congruence of organizationpublic relationships, which captures the interaction between an organization and its publics. The second assumption change arises from contradiction. Because of the tug-ofwar nature, it is impossible to fulfill both of the opposing yet interdependent forces or tendencies. Change then is always present; it forms a dialectical unity with stability (Baxter & Montgomery, 1996). In other words, “dialectical change is the interplay of stability and flux” (Baxter & Montgomery, p. 10). The change process is like a spiral that contains elements of cyclical change (recurrence) as well as linear change (non-repeating changes). Praxis means that “people are at once actors and objects of their own actions” (Baxter & Montgomery, 1996, p. 12). This concept suggests that people are both proactive in making decisions and choices and reactive to partners’ decisions and choices. Such choices and decisions lead to different patterns of dialectical change. For example,
39
subordinate employees and managers who act as representatives of an organization may perceive too little trust and too much distrust in their relationships. Consequently, they may choose to ignore the tension, or try to increase trust and decrease distrust, or simply accept it as inevitable, or redefine the nature of their relationships. They are both proactive and reactive. Furthermore, their present decisions and choices are bound to constrain their future interactions. Totality refers to the idea that “social phenomena are defined by the relations among their characteristics, not by the characteristics themselves” (Montgomery, 1993, p. 206). Put simply, social phenomena are assumed to be understood only in relation to other phenomena. Consequently, relationships “are created by individuals’ interactions with other elements in society, but they also create individuals and societies” (Hung, 2007, pp. 452-453). The dialectical approach is consistent with the organization-public relationships theory summarized by Ledingham (2003). Ledingham stated that “organization-public relationships involve an ongoing interchange of needs, expectations, and fulfillment” and “organization-public relationships change over time” (p. 188), which is commensurate with the concept of change and contradiction. Also, the notion of interdependence lies at the core of the dialectical approach, and constitutes the heart of relationships in the relationship management literature. Furthermore, the dialectical approach requires us to understand organization-public relationships from both the organization’s and the public’s perspective, and to analyze the dynamic interaction between them. The dialectical approach also allows us to revisit the organization-public relationships theory—“Effectively managing organization[sic]-public relationships
40
around common interests and shared goals, over time, results in mutual understanding and benefit for interacting organizations and publics” (Ledingham, 2003, p. 190). Contradiction spurs change. Organizations and publics may not always have the same goals or interests. They often have opposite goals or interests. However, the interdependent yet opposing tension in their relationships give rise to interaction and change. During the process, they can find common grounds. Both the organization and its publics consciously make their own decisions but also react to each other’s decisions. This constant tension and change in their relationships lead to benefits of varying degrees for the organization and publics as well as better understanding of each other—they may not agree with each other. Thus, the above statement can be changed to a major proposition about managing relationships and organizational effectiveness: Proposition 1: The dynamic and constant tension and change in organizationpublic relationships around negotiated interests and goals lead to benefits of varying degrees for the organization and its publics and better understanding of each other. In this light, relationship maintenance strategies can be part of the tension and change. The organization and its publics’ perceptions of their relationships show their “understanding of each other.” In the context of organization-employee relationships, such benefits include organizational effectiveness. As Ledingham (2003) stated, “models are an illustration of theories in action” (p. 186). Based on the models in the literature (Broom et al., 1997, 2000; Bruning & Ledingham, 1999; J. Grunig & Y. Huang, 2000; Hung, 2007; Toth, 2000) and the above discussion, I proposed a three-stage model of organization-employee relationships to shed new light on theory-building of relationship management.
41
In the next sections I offer conceptualization of focal constructs: relationship maintenance strategies (symmetrical and asymmetrical), perceived relationship characteristics, congruence of perceived relationship characteristics, organizational effectiveness, contextual performance, and turnover intention; and reviews of relevant literature. I conclude with review of literatures linking the focal constructs and introduction of the conceptual model. Relationship Maintenance Strategies Definition and a Terminological Issue As Canary and Dainton (2003) stated, “Most sane people know that [interpersonal] relationships require work. That is, partners need to spend time and effort to maintain functional, satisfying relationships. Without such efforts, relationships tend to deteriorate” (p. xiii). This also applies to organization-public relationships. There are four common definitions of relationship maintenance strategies in the interpersonal communication literature: 1) “to keep a relationship in existence;” 2) “to keep a relationship in a specified state or condition;” 3) “to keep a relationship in a satisfactory condition;” and 4) “to keep a relationship in repair” (Dindia, 2003, pp. 3-4). The first definition does not concern the quality of a relationship. A dissatisfying relationship that is kept in existence is not worth maintaining. The second definition is at odds with the dialectical approach that emphasizes the constant change and adaption to opposing tensions in a relationship. In contrast, the third and fourth definitions allude to the endeavors by an organization to make its publics satisfied and prevent the relationships from decaying. Therefore, this study combined the last two definitions to refer to relationship maintenance strategies.
42
Consistent with the dialectical approach, an organization constantly and strategically needs to keep its relationships with the publics satisfying, trusting, and so on; and use preventive and corrective maintenance to prevent something from going awry with the relationship. These planned organizational behavioral attempts or efforts to keep a relationship in a satisfying condition and in repair are called relationship maintenance strategies2. A related terminological issue is whether these strategies include routine behaviors. Dainton and Stafford (1993) distinguished these two by labeling strategic maintenance as planned and conscious whereas routine behaviors as not intentional and automatic. Slightly differently, Dindia (2003) admitted that the dichotomy between routine and strategic relational maintenance behaviors may not be so clear-cut, even though she advocated a delineation of the two. Canary (2001, personal communication cited in Stafford, 2003) further departed from Dainton and Stafford’s position and contended that it was unnecessary to separate routine and strategic, stating: For me, a strategy is defined as an approach someone takes. In other words, strategic communication is implicitly learned and often mindlessly enacted. This is a broad definition of “strategic” that encompasses a lot of behavior. Strategic approaches are often routinized but become more cognitively processed when the routine plan does not work. (p. 54) It is true that in reality strategic and routine relational maintenance behaviors are often intertwined. Some maintenance behaviors may be strategic initially but become
Montgomery (1993) rejected the term relational maintenance and advocated relational sustainment in that the former portrays relationships as static to her. Similarly, J. Grunig (2006) replaced maintenance with cultivation strategies. I argue that maintenance does not necessarily imply static. Organizations must constantly monitor their maintenance strategies to keep the relationships in a satisfying condition and in repair. This means the same as sustainment and cultivation.
2
43
routine over time. Some organizations’ or people’s strategic maintenance behaviors may be perceived as routine by others, and vice versa. Stafford (2003) thus suggested using the more encompassing term maintenance behaviors in lieu of maintenance strategies to avoid controversy. I adhered to Dindia’s position that theoretically speaking, intent differentiates strategic from routine relational maintenance behaviors, and agreed that in practice this distinction appears difficult and not as meaningful and necessary. Relationship Maintenance Strategies in the Literature Prior research has proposed many relationship maintenance strategies. For example, drawn from the literature of interpersonal communication (e.g., Stafford & Canary, 1991) and conflict resolution and negotiation, Hon and J. Grunig (1999) proposed a set of relationship maintenance strategies. J. Grunig and Y. Huang (2000) modified those strategies and categorized them into two types: symmetrical and asymmetrical. The symmetrical strategies are disclosure or openness, assurances of legitimacy, participation in mutual networks, shared tasks, integrative negotiation, cooperation/collaboration, being unconditionally constructive and win-win or no deal. The asymmetrical ones encompass distributive negotiation, avoiding, contending, compromising, and accommodating. Hon and J. Grunig pointed out that the symmetrical (mutually beneficial) strategies are more effective in maintaining relationships than the asymmetrical (self-gain interested only) ones, saying “the most productive relationships in the long run are those that benefit both parties in the relationship rather than those designed to benefit the organizations only” (p. 11). Symmetrical Maintenance Strategies
44
Disclosure or openness. Canary and Stafford (1994) defined disclosure or openness as “direct discussion about the nature of the relationship and setting aside times for talks about the relationship” (p. 12). Hon and J. Grunig (1999) conceived of this strategy as being open about one’s thoughts and feelings. For example, a relational party may encourage the other to disclose thoughts and feelings to him or her, try to discuss the quality of relationship with the other, and remind the other of past relationship decisions they made (Stafford, 2003). In the context of organization-public relationships, both organizations and publics are supposed to be open and frank about what they think about certain situations and decisions, and share with each other concerns they hold (L. Grunig et al., 2002). J. Grunig and Y. Huang (2000) suggested that public relations managers might measure publics’ disclosure by “counting suggestions, complaints, inquiries, and other contacts that members of publics, the media, government, or leaders of activist groups make with the organization” (p. 41). J. Grunig and Y. Huang (2000) considered disclosure or openness logically connected with the two-way symmetrical model of public relations, and positive relationship outcomes. Although Hung (2000) pointed out that openness or disclosure may not guarantee quality relationships, Bok (1989) noted that sharing information can lead to better equity of power in a relationship because access to information is power and often the more powerful can access more information than the powerless. Networking. In interpersonal relationships, networking means spending time with mutual friends, focusing on common friends and affiliations, and expressing willingness to do things with partners’ friends and family (Stafford, 2003). From the perspective of building organization-public relationships, also referred to as participation in mutual
45
networks, this strategy asks organizations to build “networks or coalitions with the same groups that their publics do, such as environmentalists, unions, or community groups” (Hon & J. Grunig, 1999, p. 15). Likewise, publics may try to join networks or coalitions to which organizations belong. J. Grunig and Y. Huang (2000) proposed to assess networking through counting organizations’ contacts with networks of activist groups. Assurances of legitimacy. Assurances of love and commitment are common in interpersonal relationships. For example, married couples stress their commitment to each other and show themselves to be faithful to each other; relational partners indicate that the relationship has a future (Stafford, 2003).With regard to organization-public relationships, assurances of legitimacy are attempts by relational parties to assure each other that their needs, concerns, and problems are legitimate and should be addressed (Hon & J. Grunig, 1999). This is also in line with the two-way symmetrical model (L. Grunig et al., 2002). Hon and J. Grunig (1999) noted that demonstration of commitment to maintaining a relationship can help assure legitimacy. They provided an example of a university demonstrating assurances by developing a new land grant system. The director of external relations for this university’s agricultural extension office explained that university programs catered to the needs of publics and tried to deliver in the way publics wanted. Similarly, L. Grunig’s (1992) research on activist groups confirmed that assurances of legitimacy by organizations are necessary when dealing with activists. Shared tasks. This strategy typically means sharing household responsibilities in interpersonal relationships (Canary & Stafford, 1992, 2001; Stafford, 2003). Messman, Canary, and Hause (2000) in their examination of non-romantic relationships identified
46
shared activity as a maintenance strategy, consisting of sharing special rituals, specific routine activities, and time with the other relational parties. Extending this notion of shared tasks to organization-public relationships, Hon and J. Grunig (1999) conceptualized shared tasks as “organizations’ and publics’ sharing in solving joint or separate problems” (p. 15). They included examples of tasks for organizations and publics such as “reducing pollution, providing employment, making a profit, and staying in business” that interest either the publics or the organizations or both (p. 15). As the organizations’ devotion to solving these problems can be recognized as the organizations’ attempts to ensure the publics that their concerns and problems are legitimate and should be addressed, assurances of legitimacy can subsume shared tasks. Advice. Another maintenance strategy in the interpersonal communication context that can be used in organization-public relationships is advice (Stafford, Dainton, & Haas, 2000)—telling a relational party what he or she needs to do about his or her problems or giving opinions on things going on in his or her life. Given that Hon and J. Grunig’s (1999) and J. Grunig and Y. Huang’s (2000) were based on the earlier work of Stafford and Canary (1991), advice can be added to the collection of maintenance strategies. For example, an organization’s management, as the representative of the organization, can advise about subordinate employees’ careers and/or personal lives when relevant to their career. In addition to these interpersonal strategies, Hon and J. Grunig (1999) and J. Grunig and Y. Huang (2000) suggested a list of symmetrical strategies originated from the conflict resolution and negotiation literature.
47
Integrative negotiation. Putnam (1990) described integrative negotiation as attempts to use “open information exchange and joint decision making” to accomplish joint gains or mutually beneficial outcomes (p. 3). Hon and J. Grunig (1999) considered integrative negotiation a symmetrical strategy because common/mutual interests and winwin solutions are sought out. Nevertheless, “open information exchange” overlaps with disclosure or openness, and “joint decision making” can be subsumed by assurances of legitimacy. Therefore this strategy was not included in this study. Cooperation/collaboration. Cooperation or collaboration involves both the organization and the public trying to reconcile their interests toward building a mutually beneficial relationship (J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999). Such attempts are similar to disclosure or openness, which also aims to reconcile different interests through honest discussions. This strategy accordingly can be incorporated into disclosure or openness. Being unconditionally constructive. This strategy means that “the organization does whatever it thinks is best for the relationship, even if it means giving up some of its positions and even if the public does not reciprocate” (Hon & J. Grunig, 1999, p. 17). It has been often used by organizations when faced with resistant activists who do not compromise or even refuse to negotiate (J. Grunig & Y. Huang, 2000). Win-win or no deal. This strategy is used when the organization and the public fail to find a mutually beneficial solution and agree to disagree — no deal (Hon & J. Grunig, 1999). It is viewed as symmetrical because agreeing to disagree leaves open the potential that the organization and public may still be able to work together to find a winwin solution next time (Covey, 1989; J. Grunig & Y. Huang, 2000; Hon & J. Grunig).
48
This strategy as well as being unconditionally constructive examines how an organization works with its publics for a mutually beneficial solution, which overlaps with disclosure or openness and assurances of legitimacy. Hence, this study excluded these two strategies. Asymmetrical Maintenance Strategies In addition to these symmetrical maintenance strategies, organizations and publics also use asymmetrical ones: distributive negotiation, avoiding, contending, compromising, and accommodating. Distributive negotiation. This strategy aims for self-gains, involving attempts to maximize gains and minimize losses for oneself (J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999). Specific tactics of this strategy include imposing one’s position onto the other relational party without concern for the other, threatening, dominating, and arguing (Hon & J. Grunig; Morill & Thomas, 1992; Putnam & C. Wilson, 1982). This win-lose or self-gain perspective indicates the asymmetry of this strategy. Dual concern strategies. Plowman identified four dual concern strategies, namely, avoiding, contending, compromising and accommodating. They were called “mixed motive or collaborative advocacy” by Hon and J. Grunig (1999, p. 16). An organization and its publics do take each other’s interests into consideration when employing these strategies, but emphasize self interests over others’ interests; thus they are still asymmetrical in nature. Avoiding was defined as “the organization leaves the conflict either physically or psychologically” (Hon & J. Grunig, p. 16). Hess (2003) put forward similar strategies in the interpersonal communication context—avoidance (“[t]rying not to be in the presence of the other person”), ignoring (“[a]cting as if the other person is not there”), and
49
inattention (“[g]iving as little attention as possible to the other person”) (p. 115). Put in the organization-employee relationships context, avoiding/ignoring/inattention can be classified as one strategy that either the organization or public or both use to distance from the other. Contending refers to the extent to which an organization tries to convince the public to accept its position or vice versa (J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999). The notion of contending is analogous to some aforementioned distributive negotiation tactics, such as threatening, dominating, and arguing. Therefore this study used distributive negotiation only. Compromising and accommodating were respectively conceptualized as “the organization meets the public part way between its preferred positions, but neither is completely satisfied with the outcome” and “the organization yields, at least in part, on its position and lowers its aspirations,” which were virtually the same (J. Grunig & Y. Huang, p. 40). Therefore, I will labeled them as one strategy compromising. In sum, there are two kinds of relationship maintenance strategies in the public relations literature: symmetrical and asymmetrical ones. The symmetrical strategies consist of disclosure or openness, assurances of legitimacy, networking, and advice. The asymmetrical ones encompass distributive negotiation, avoiding, and compromising. J. Grunig (2006) considered relationship maintenance strategies “the heir to the models of pubic relations and the two-way symmetrical model, in particular” (p. 168). Truly the symmetrical relationship maintenance strategies are consistent with the two-way symmetrical model, underscoring collaboration, two-way communication, the use of both mediated and interpersonal forms of communication, and nearly always ethical practice.
50
In contrast, the asymmetrical relationship maintenance strategies are logically connected with the two-way asymmetrical model, characterized by advocacy, one-way communication, the use of either mediated or interpersonal forms of communication, and sometimes ethical or unethical practice. Accordingly, I end the above discussion with the second research proposition and first research question on relationship maintenance strategies: Proposition 2: Symmetrical relationship maintenance strategies are composed by disclosure or openness, assurances of legitimacy, networking, and advice. Asymmetrical relationship maintenance strategies include distributive negotiation, avoiding, and compromising. Research Question 1: How do organizational members perceive relationship maintenance strategies used by their organizations? Congruence of Perceived Relationship Characteristics Using the congruence between two constructs or two sides of a construct (e.g., non-management employees vs. managers’ perceptions) is not uncommon in organizational psychology (e.g., Ostroff et al., 2004; Saks & Ashforth; London & Wohlers, 1991; Shore & Bleicken, 1991). As noted earlier, I intended to measure both sides of organization-employee relationships, thus using the construct of congruence of perceived relationship characteristics. It refers to the degree of congruence between employees’ and their organizations’ perceptions of their relationship characteristics. Such characteristics describe the quality of organization-employee relationships. I first review the dimensions of relationship quality and then discuss issues related to congruence.
51
Prior research on relationship management has shown that the quality of an organization-public relationship is multidimensional. Extensive scholarly work has been devoted to identifying these dimensions. Ferguson (1984) was among the first researchers to suggest a few dichotomous dimensions of organization-public relationships, including dynamic versus static, open versus closed, and satisfactory versus unsatisfactory. L. Grunig et al. (1992) proposed six dimensions of perceived relationship state: reciprocity, trust, mutual legitimacy, openness, mutual satisfaction, and mutual understanding. These pioneering scholars nonetheless did not specify how to measure these dimensions or test these theoretical dimensions. Hon and J. Grunig (1999) were among the first to empirically test these dimensions. They derived six dimensions from the interpersonal communication and psychology literature: trust, commitment, satisfaction, control mutuality, communal relationships, and exchange relationships. The University of Maryland research team invited 200 people from randomly chosen email addresses to take an online survey of their perceptions of relationships with five large organizations. The sub-scales (for each dimension) had good reliability (? > .80 on average). Communal relationships and exchange relationships were dropped later by J. Grunig and Y. Huang (2000). Ever since, the remaining four dimensions (trust, commitment, satisfaction, and control mutuality) have been widely used in a variety of studies (e.g., J. Grunig & Hung, 2002; Y. Huang, 2001; Ki & Hon, 2007b, 2007c; Ki & Shin, 2005; H.-S. Kim, 2005, 2007; Y. Kim, 2001; Yang & J. Grunig, 2005; Yang, 2007). Y. Huang added a fifth dimension to Hon and J. Grunig’s four-dimensional measuring instrument—face and favor, which is presumably unique in the Taiwanese culture. This study used these four
52
dimensions (trust, commitment, satisfaction, and control mutuality) to assess characteristics of organization-employee relationships and proposed a fifth dimension-distrust. I present a detailed explanation of these terms below. Trust Trust as a dimension of relationship characteristics in organizational settings has been widely acknowledged (see Mayer, Davis, & Schoorman, 1995; Schoorman, Mayer, & Davis, 2007 for more details). Hon and J. Grunig (1999) defined trust as “one party’s level of confidence in and willingness to open oneself to the other party” (p. 5). Similarly, Moorman, Deshpande and Zaltman (1993) viewed trust as “a willingness to rely on an exchange partner in whom one has confidence” (p. 82). On a different note, the well-cited marketing scholars R. Morgan and S. Hunt (1994) opposed to having “willingness” in the conceptualization of trust because willingness implies behavioral intention. They asserted, “Just as behavioral intention is best viewed as an outcome of attitude and not as part of its definition … ‘willingness to rely’ should be viewed as an outcome … of trust and not as a part of how one defines it” (p. 24). As Hon and J. Grunig (1999) considered relationship characteristics as perceptions rather than behavioral intentions, this study removed the “willingness” component from Hon and J. Grunig’s (1999) definition and defined trust as one party’s level of confidence in the other party. According to Hon and J. Grunig (1999), trust has three sub-dimensions: integrity, dependability, and competence. Integrity shows one party’s belief of consistency and fairness in the other’s behaviors. An example of integrity is “This organization treats people like me fairly and justly” (Hon & J. Grunig, p. 28). Dependability describes the consistency between one’s words and behaviors. An example of dependability is “This
53
organization can be relied on to keep its promises” (Hon & J. Grunig, p. 28). Competence indicates the extent to which one party believes the other has the ability to do what it says it will do. An example of competence is “This organization has the ability to accomplish what it says it will do” (Hon & J. Grunig, p. 28). Cho (2006), Mayer et al. (1995) and Schoorman et al. (2007) contended that ability (equivalent to competence), integrity, and benevolence are dimensions of a person’s perceptions of another’s trustworthiness and thus antecedents to the amount of trust this person can have in another. Benevolence was referred to as “the extent to which a party is believed to want to do good for the trusting party, aside from an egocentric profit motive” (Schoorman et al., p. 345). Likewise, Cho (2006, 2007) treated benevolence, integrity, and competence as dimensions of trustworthiness, antecedents to customers’ perceptions of trust. The question central to trust then is: are the three sub-dimensions proposed by Hon & J. Grunig (1999) sub-dimensions of trust or actually measures of a party’s trustworthiness? Many researchers on relationship management tend to take Hon and J. Grunig’s scale for granted without clarifying the conceptualization. This study examined the validity of trust in Chapter III to address this question. Distrust In relation to the dimension of trust, there are some new developments in the management literature. Based on Luhmann’s work (1979), Lewicki et al. (1998) were among the first (e.g., Kramer, 1994; Sitkin & Roth, 1993; McKnight & Chervany, 2001) to propose distrust as distinct from trust. One would think that distrust is simply the opposite of trust. To the contrary, Lewicki et al. argued that trust and distrust are
54
“separate but linked dimensions” (p. 439). Trust is not simply the absence of trust. They defined trust “in terms of confident positive expectations regarding another’s conduct” and distrust “in terms of confident negative expectations regarding another’s conduct” (p. 439). More specifically, “another’s conduct” includes “words, actions, and decisions;” such “confident positive expectations” are “a belief in, a propensity to attribute virtuous intentions to, and a willingness to act on the basis of another’s conduct;” (Lewicki et al., p. 439). This is quite similar to Hon and J. Grunig’s definition of trust. Lewicki et al. elaborated that the “negative expectations” involved in distrust are “a fear of, a propensity to attribute sinister intentions to, and a desire to buffer oneself from the effects of another’s conduct” (p. 439). It is possible for relational parties in organizations to have both trust and distrust in one another, analogous to harboring both love and hatred. In this light, low distrust cannot be the same as high trust and high distrust not the same as low trust. To better illustrate their conceptualization, Lewicki et al. produced a 2 by 2 table of high and low trust and high and low distrust (p. 445). I reproduced their table (see Table 1). Based on the two-dimensional framework, Lewicki et al. (1998) proposed four prototypical relationship conditions: low trust/low distrust, high trust/low distrust, low trust/high distrust, and high trust/high distrust. Under the condition of low trust/low distrust, the individual has no reason to be confident or vigilant. One example is casual acquaintances. Under the condition of high trust/low distrust, one is confident and does not suspect the other. The relationship is characterized by increased interdependence, where relational parties try to pursue new common initiatives and opportunities. Under the condition of low trust/high distrust, the person is not confident in but very watchful of
55
his or her relational partner. Relational parties attribute harmful motives to the other, attempt to exploit potential vulnerabilities of the other, and dedicate considerable resources to monitor the other’s behavior. Under the condition of high trust/high distrust, one is confident in another in some aspects but suspicious of another in other aspects. Partners have separate and shared objectives. They trust but also verify the each other’s words, actions, and decisions. They pursue new opportunities but also monitor risks involved. Table 1 Trust and Distrust Prototypical Relationship Conditions High Trust Characterized by hope, faith, confidence, assurance, initiative High-value congruence; Interdependence promoted; Opportunities pursued; New initiatives Trust but verify; Relationships highly segmented & bounded; Opportunities pursued & down-side risks/vulnerabilities continually monitored Low Trust Characterized by no hope, no faith, no confidence, passivity, hesitance Low Distrust Characterized by no fear, absence of skepticism, absence of cynicism, low monitoring, no vigilance Casual acquaintances; Limited interdependence; Bounded, armslength transactions; Professional courtesy Undesirable eventualities expected & feared; Harmful motives assumed; Interdependence managed; Preemption; best offense is a good defense; paranoia High Distrust Characterized by fear, skepticism, cynicism, wariness and watchfulness, vigilance
56
Lewicki et al.’s (1998) proposition is intuitively appealing, particularly considering the relationship an employee has with his or her organization (represented by the management). In the age of corporate scandals, even the most satisfying employee’s relationship with his or her organization may be high trust/high distrust, other than high trust/low distrust. From an organizational point of view, it is important to investigate trust as well as distrust because the strategies to build and maintain trust do not necessarily reduce or eliminate distrust; and outcomes as a result of trust as opposed to distrust may differ. A few researchers have begun to empirically test the distinction between trust and distrust (e.g., Adler, 2007; Benamati & Serva, 2007; Cho, 2006, 2007; H. Huang & Dastmalchian, 2006; Lai & Mollering, 2007). For example, Cho (2006) used Lewicki et al.’s (1998) conceptualization of trust and distrust to test antecedents to and consequences of trust and distrust. She posited that core business operations and relationship investments influence a customer’s judgments of an e-vendor’s trustworthiness (i.e., competence and benevolence), which in turn influence differently consumers’ perceptions of trust and distrust toward the e-vendor. Trust and distrust were expected to have different effects on behavioral intentions (i.e., self-disclosure and relationship commitment). Using newly developed measures of distrust, Cho (2006) found that benevolence can be considered as a “trust-producing factor” that has a more significantly positive effect on trust than its negative influence on distrust, but competence is rather “a distrust-avoidance factor” that has a more significantly negative impact on distrust than a positive effect on trust (p. 32).
57
Further, she concluded that a high level of distrust is instrumental in affecting selfdisclosure and high trust more vital to consumer commitment. Cho (2007) investigated the differing effects of trust and distrust on information privacy concerns within the context of obtaining loan services over the Internet. The Web study of 662 responses indicated that distrust has a greater positive effect on information privacy concerns than the negative impact of trust, moderated by customers’ risk proneness (risk-seeking or risk-averse) and trust propensity. Similar to Cho’s (2006) previous study, findings established that distrust is a concept distinct from trust. In lieu of embracing the new concept of distrust, Schoorman et al. (2007) charged that extant theorizing as well as empirical evidence on distrust is not convincing enough to demonstrate that the concept of distrust is actually theoretically or empirically distinct from trust. They criticized researchers (e.g., McAllister, Pang, Tan, & Ruan, 2006) who argued for this distinction but only reverse scored trust items to measure distrust. Nevertheless, a lack of effort in developing and testing valid measures of distrust does not necessarily “prove” that distrust is not a separate construct. Cho’s (2006, 2007) studies provided evidence of the distinctness of distrust. This study accordingly investigated whether distrust is a new dimension of characteristics of organizationemployee relationships. Commitment Hon and J. Grunig (1999) defined commitment as “the extent to which the public [or the organization] feels that the relationship is worth spending time and energy to maintain” (p. 3). This differs from others’ definitions of relational commitment as a behavioral intention (e.g., E. Anderson & Weitz, 1992; Cho, 2006; Michaels, Acock, &
58
Edwards, 1986; Wulf, Odekerken-Schroder, & Iacobucci, 2001). Michaels et al. (1986) stated that commitment is “one’s desire and intent to maintain, rather than terminate, a relationship” (p. 162). In a similar vein, Cho (2006) referred to it as “a lasting or enduring intention to build and maintain a long-term relationship” (p. 28). Likewise, Canary and Zelley (2000) used this conceptualization—“the extent to which a person wants to remain in the relationship and has feelings of attachment” (p. 308). These definitions of commitment as behavioral intention are conceptually analogous to turnover intentions, which are a criterion of organizational effectiveness (Robbins, 1990), distinct from relationship characteristics. Furthermore, because this study viewed relational characteristics as perceptions, I chose Hon and J. Grunig’s definition rather than others’ that focused on behavioral intention. Marketing researchers and public relations scholars alike have recognized commitment as a multidimensional construct (e.g., J. Brown, Lusch, & Nicholson 1995; Gruen et al., 2000; Gundlach, Achrol, & Mentzer, 1995; Hon & J. Grunig, 1999; Hung, 2001; K. Kim & Frazier,1997; Kumar, Scheer, & Steenkamp 1995). Hon and J. Grunig proposed two sub-dimensions—continuance and affective commitment. They derived these sub-dimensions from Meyer and N. Allen’s (1984) work in which continuance commitment indicates the degree to which a relational party (e.g., an employee) is psychologically bonded to an organization out of concerns for the perceived costs associated with leaving the organization; and affective commitment means emotional attachment. In addition to these two sub-dimensions, Gruen et al. proposed another subdimension of commitment that members have towards their professional associations— normative commitment, which describes the extent to which a member is
59
“psychologically bonded to the organization on the basis of the perceived moral obligation to maintain the relationship with the organization” (p. 37). Their study of member-association relationships confirmed the validity of normative commitment as a sub-dimension of commitment. Such obligation-based commitment makes intuitive sense; for example, an employee of a charity organization or activist organization may feel a sense of moral obligation to stay in a relationship with his or her organization if this employee joined the organization to promote charity or a cause. Although Gruen et al. only focused on member-association relationship, it is intriguing to investigate whether normative commitment can be extended to other contexts. This study explored such a possibility. It is noteworthy that Hon and J. Grunig’s (1999) definition of continuance commitment lacks clarity because they stated that it “refers to a certain line of action,” implying a behavior, which contradicts their conceptualization of commitment and other indicators as perceptions of relationships (p. 25). Such lack of clarity was present in their measures of commitment; items such as “I have no desire to have a relationship with this organization” pertain to behavioral intentions. Accordingly, refinement and addition to the existing measures of commitment were included in this study. Satisfaction Hon and J. Grunig (1999) viewed satisfaction as “the extent to which one party feels favorably toward the other because positive expectations about the relationship are reinforced” (p. 20). From a social exchange perspective, both parties are satisfied with the relationship when they achieve a balance of rewards and costs (Stafford & Canary, 1991). J. Grunig and Y. Huang (2000) observed that satisfaction involves affection and emotion
60
whereas trust and control mutuality consist of cognitive dimensions. Initially used to evaluate intimate relationships (Hendrick, 1988), romantic relationships (Stafford, 2003), and work relationships (Waldron, 2003), satisfaction has been widely used to assess organization-public relationships (e.g., J. Grunig & Y. Huang, 2000; Y. Huang, 2001; Hung, 2001; H. Kim, 2005; Ledingham, 2003; Yang, 2007). Organizational psychologists distinguish job satisfaction and satisfaction with the organization. Job satisfaction indicates an employee’s “subjective evaluation of the degree to which his or her requirements are met by the work environment” (Bretz & Judge, 1994, p. 32). It often has to do with factors such as complexity of the job. Public relations scholars (e.g., D’Aprix, 1996; J. Grunig, 1992; L. Grunig et al.) maintained that the satisfaction with the organization overall is a better indicator of the effect of the organization’s communication efforts because employees can be satisfied with their jobs but not with the organization. In the event of low satisfaction with the organization but high job satisfaction, it will be misleading to use measures of job satisfaction to evaluate satisfaction with the organization. Accordingly, the concept satisfaction in this study does not involve job satisfaction. Control Mutuality Control mutuality represents “the degree to which parities agree on who has rightful power to influence one another” (Hon & J. Grunig, 1999, p. 19). Stafford and Canary (1991) more specifically suggested that such “influence” is over “relational goals and behavioral routines” (p. 224). Similar ideas abound in the literature, such as Ferguson’s (1984) “distribution of power in the relationship,” Aldrich’s (1979) notion of reciprocity, and Kelley’s (1979) bilateral control over the relational partner, Bruning and
61
Ledingham’s (1999) “mutual legitimacy” and Bortree and Waters’ (2007) “balanced power.” Power imbalance seems inevitable in interpersonal, inter-organizational and organization-public relationships. It exists in three forms (J. Grunig & Y. Huang, 2000). First, one party is completely powerless and consequently both parties try not to antagonize each other because of fear of escalated tension. Second, when the power difference is minimum, relational parties engage in intensive power struggle that does not end well for either one. Third, when the power difference is large, a third party can be introduced to help with the less powerful and balance out the difference. Research has shown that excessive attempts to exert control are negatively associated with perceived competence of communicators (Bochner, Kaminski, & Fitzpatrick, 1977) and increase levels of activism (L. Grunig, 1992). On the other hand, power does not have to be completely equally distributed to maintain a stable relationship (L. Grunig et al., 1992). Hon and J. Grunig (1999) noted that as long as organizations and publics exert some control over each other, the relationship will be quite stable and positive. Research has linked attempts to control a relationship unilaterally to relational dissatisfaction and misunderstanding (Canary, Weger, & Stafford, 1991; Courtright, Millar, & Rogers-Millar, 1979). Congruence of Perceived Relationship Characteristics Defining congruence. As I discussed earlier, this study adopted the dialectical approach to relationships, a core assumption of which is contradiction. One needs to understand the interaction between organizations and their publics in order to best capture
62
the essence of organization-public relationships, hence the notion of congruence of perceived relationship characteristics. Congruence, also known as fit, in organizational psychology originated from the idea that individual behavior can only be understood as an interaction between individuals and their organizational context (Ostroff & Schulte, 2007). Based on this interactionist perspective, scholars proposed the notion of person-environment fit or congruence that suggests matching individual characteristics with their corresponding environmental characteristics (e.g., Schneider, 1983, 1987). This matching between persons and organizations occurs when individuals’ personality matches the organizational climate (Christiansen, Villanova, & Mikulay, 1997; A. Ryan & Schmitt, 1996; Tom, 1971), or when their values agree with the organizational culture (e.g., Chatman, 1989; O’Reilly, Chatman, & Caldwell, 1991), or when their goals are consistent with those of organizations (e.g., Vancouver & Schmitt, 1991), or when their needs are congruent with characteristics of the environment (Pervin, 1968), to name just a few. This similarity-based person-organization fit or congruence is built upon Schneider’s (1987) Attraction-Selection-Attrition framework, which posited that similar people tend to be attracted to, selected by, and retained by organizations. However, the caveat is that the resultant homogeneity from similarity-based fit can lead to lack of creativity, diversity, and inability to change (Schneider, Goldstein, & Smith, 1995). In light of the dialectical approach to relationships, I argue that the interactionist perspective on in organizational psychology is consistent with the dialectical assumptions. A dialectical approach to relationship management views organizations and their publics, such as employee publics, as opposing yet interdependent forces (contradiction) whose
63
agenda differ considerably at times. Their contradiction serves as the agent of change. Due to the constant negotiation of interests and goals, change occurs in their relationships. During this process of negotiation, organizations resort to an array of relationship maintenance strategies, which result in different perceptions in employees. Their proactiveness and reactiveness in the decisions and choices organizations and employees make (praxis) are shown in both sides’ perceived relationship quality and employees’ behavior at work. Understanding organization-public relationships as a process of relating exhibits the applicability of totality to relationship management research. Integrating both the dialectical approach and the interaectionist perspective and applying it to organization-public relationships, it is clear that “relationships” is a “relating” process that highlights the interaction between publics (e.g., employees) and their corresponding organizations. That is, the concept of congruence of perceived organization-public relationship characteristics is the key to understanding how organizations and publics (e.g., employees) “relate” to each other. Congruence of perceived organization-public relationship characteristics does not refer to similarity or consensus, but represents the dynamic interaction and process of interdependence between unified opposing relational parties. As Broom et al. (2000) noted, “Relationships are the dynamic results of the exchanges and reciprocity that manifest themselves as the relationships develop and evolve, yet they can be described at a given point in time” (p. 17). At a given point in time, the interaction may result in compatibility and consensus. But it will change (change from the dialectical approach) as the two opposing forces in the relationship continue to negotiate and adapt to each other’s reactions (praxis and totality from the dialectical approach).
64
The matching or fitting that happens between the organization and its employees in this study is give-and-take of perceptions. Here employees’ perceptions are akin to the person variable (individual characteristics), and the organization’s perceptions are the organization variable (aggregate assessment of person’s characteristics to represent organizational characteristics) in the aforementioned person-organization fit research. The reasoning for using aggregate personal characteristics (e.g., perceptions) to represent organizational characteristics is that organizations are made up of people and people’s perceptions constitute the environment (e.g., R. Hogan & Roberts, 2000; Schneider, 1987). According to Edwards (1994, 1995), congruence can be operationally expressed by five variables: the perceptions of employees (one opposition in the dialectical approach) as the person variable, the perceptions of the organization (another opposition in the dialectical approach) as the organization variable, the squared person variable, the squared organization variable, and the product of the person and organization variable. Level of analysis. Person-environment fit researchers assume that individuals can be fit to different levels of the environment, such as supervisor, job, group, and organization (Ostroff & Schulte, 2007). Outcomes of fit are primarily at the individual level, including job satisfaction, employee turnover and employee performance (Ostroff & Schulte). It is necessary for researchers to indicate what they are fitting to and at. Fitting to is related to the level of the environment, whereas fitting at means the level of analysis for the outcomes of fit. One can fit employee characteristics to those of their organizations and analyze the individual-level consequences. Conversely, one can fit employee characteristics to those of co-workers (dyadic) while examining effectiveness outcomes at the group or organizational level. This study was fitting employees’
65
perceptions of organization-employee relationship characteristics to their organizations’ perceptions (represented by aggregation of managers’ perceptions), and focusing on effectiveness outcomes at the individual level. Measurement Issues Related to Congruence of Perceived Relationship Characteristics Extant literature primarily assesses relationships from a one-sided perspective (Broom et al., 1997; Christensen, 2005; J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999; Kelly et al., 2006; Seltzer, 2005; Seltzer & Mitrook, 2009; Shin & Cameron, 2005; Waters, 2007). Researchers typically ask publics’ perceptions of their relationships with a given organization. Leading scholars on organization-public relationships such as Hon and J. Grunig and J. Grunig and Y. Huang have long acknowledged the necessity of examining both parties in a relationship. Waters posed the question: “[W]hy does our literature encourage symmetrical communication yet our research is still asymmetrical” (p. 25)? Seltzer used an interpersonal example to illustrate, Albert may like Betty, [and] he may think that she likes him, and … that they have a healthy relationship, but if Betty doesn’t like Albert, then their relationship is not healthy, regardless of Albert’s perceptions. Simply put, the saying that it takes two to tango is directly applicable to public relations relationship measurement. (p. 13) Measuring both relational parties in a relationship is not news in the interpersonal communication literature (e.g., Wilmot, 1980). However, few scholars have made such attempts in public relations research. More recently, a handful of studies employing the coorientation method surfaced in the public relations literature (Christensen, 2005; Kelly et al., 2006; Seltzer, 2005; Seltzer & Mitrook, 2009; Shin & Cameron, 2005; Waters,
66
2007). The coorientation method was initially introduced by J. Grunig and T. Hunt (1984) and Broom and Dozier (1990) into public relations. It consists of four parts: 1) the organization’s view of a relationship; 2) the public’s view of a relationship; 3) the organization’s estimate of the public’s view; 4) and the public’s estimate of the organization’s view. Agreement shows the extent to which the organization’s view matches that of the public’s. Accuracy represents the degree to which the organization’s or the public’s estimate of each other’s view was correct. Perceived agreement is the degree of congruence between the organization’s (or the public’s) view and its perception of the public’s view (or the organization’s). Based on this conceptualization, Broom and Dozier suggested using difference scores or correlation coefficients between two parties’ mean ratings on each questionnaire item. The empirical studies using this method calculate mean scores for each relational dimension for both the organization and the public, then use t-tests, analysis of variance (ANOVA), or multivariate analysis of variance (MANOVA) to see if the mean differences are significant. Significant mean differences represent different levels of accuracy, agreement, and perceived agreement, which are categorized into four states: (a) consensus, (b) dissensus, (c) false consensus, and (d) false conflict (see Figure 1).
67
Organization’s Views of Relationship
Agreement
Public’s Views of Relationship
Perceived Agreement
Accuracy
Perceived Agreement
Organization’s Estimate of Public’s Views of Relationship
Public’s Estimate of Organization’s Views of Relationship
Figure 2. The coorientation model adapted from Broom and Dozier (1990).
68
Though conceptually intriguing, the coorientation method is not without problems. Organizational psychologists Edwards and Parry (1993) and Edwards (1994, 1995) were the first ones to discuss the substantial problems associated with using difference scores as dependent variables as well as independent variables. I first discuss the problems with using difference scores in examining organization-public relationships as a dependent variable herein. Problems with difference scores representing relationships as a dependent variable. As a dependent variable, the use of difference scores to represent agreement, accuracy and congruence between perceived organization-public relationships introduces four major problems (Edwards, 1995). First, as it often happens, the component measures (i.e., two sides’ views and estimates) are positively correlated, which results in difference scores less reliable than their component measures. When one uses components of the difference as separate dependent variables, the reliability tends to be higher. In the case of a decreased reliability by using the difference scores, the standard error of estimate will go up and statistical power will drop (Edwards, 1995; Pedhazur, 1982). Second, the difference-score approach of the coorientation method essentially collapses views of two sides into a single index (differences of mean scores). It “conceal the relative contribution of the component measures to variance in the composite score” (Edwards, 1995, p. 308). One should not forget that the variance of a difference score is a function of the variance as well as covariances of the component measures. The difference-score approach does not allow for estimation of the two sides’ independent effects (Ostroff et al., 2004).
69
Third, difference scores confound the influence of the independent variables on the components (i.e., each side’s views and estimates). Edwards (1995) illustrated this problem using regression equations (pp. 308-309): Y1 – Y2 = b0 + b1X + e (Equation 1), in which Y1 and Y2 are measures of the two sides, X is an independent variable, b1 is the unstandardized regression coefficient, and e is random disturbance. One can express the difference scores of the two sides independently in two equations: Y1 = b10 + b11X + e1 (Equation 2); and Y2 = b20 + b21X + e2 (Equation 3), from which the first equation can be rewritten as: Y1 – Y2 = (b10 - b20) + (b11 - b21) X + (e1 - e2) (Equation 4) It is clear that estimating equations 2 and 3, one can obtain b1 in equation 1. Yet, estimations based on equation 1 alone cannot yield information on the signs or magnitudes of b11 and b21. When interpreting data, one cannot draw conclusions about the effects of the independent variables on each side of relational parties. It is possible that X has an equal but opposite effect on Y1 and Y2 when b11 = - b21, or an effect on Y1 but not on Y2 when b11 = b1, b21= 0, etc. But relying on equation 1 alone cannot offer such information. Lastly, Edwards (1995) pointed out that using difference scores as a dependent variable “transforms an inherently multivariate model into a univariate model” (p. 310). This again can be shown from the above equations. By jointly estimating equation 2 and 3, one can obtain information on the independent effects of X on Y1 and Y2, the joint effects of X on Y1 and Y2 (using multivariate tests of significance), and correlations
70
between residuals (e1 and e2). Such tests are not possible in a univariate model shown in equation 1. Problems with difference scores representing relationships as an independent variable. Similarly, when algebraic difference indices are used to represent relationships as an independent variable, four major problems emerged (Edwards, 1994). First, algebraic difference indices are difficult to interpret (Edwards). They would appear to represent equal but opposite contributions of each component, i.e., each of the two sides of a relationship3, which could only be assumed in the event that both sides had the same variance. Yet, unequal variances are often the case when multiple subordinating employees are compared with the same managerial employee. Second, algebraic difference indices conceal the relative contribution of the components (each side of a relationship) to the relationship between the index and a dependent variable (Edwards, 1994). Sometimes, it is likely that the relationship between the index and the dependent variable is driven by one of the components only. Third, algebraic difference indices cannot explain more variance than when the components are considered jointly (Edwards, 1994). Graphically, an algebraic difference index is a plane with two components having equal but opposite linear relationships with a dependent variable. It is simply an approximation of the actual surface relating the components to the dependent variable, and hence unable to account for more variance than its components. Lastly, the coorientation method using difference scores to represent relationships as an independent variable usually involves testing whether the coefficient shared by the two components in its regression equation is significantly different from zero.
3
Z = b0 + b1 (X-Y) + e
71
Nevertheless, it neglects to also examine whether the implied constraint is valid, i.e., if the components do share the coefficient (Edwards, 1994). Another test is necessary to show that the two components’ coefficients have opposite signs but do not differ significantly in absolute magnitude, which has not been shown in current research using the coorientation method in public relations. As noted earlier, this study used congruence of perceived organization-employee relationship characteristics as a mediator. Thus, the problems associated with using difference scores to represent relationships as dependent as well as independent variables would be certainly applicable should the coorientation method using the difference-score approach. To avoid the preceding problems, Edwards (1994, 1995) proposed an alternative multivariate procedure. Regarding analyzing the components as dependent variables, multivariate regression analysis is suggested. The two components are retained separately and are tested jointly, enabling researchers to estimate “the effects of each antecedent variable on each component measure as well as the multivariate association between the predictors and component measures as a set” (Ostroff et al., 2004, p. 346). Suppose there is dissensus or disagreement between the two relational parties; this approach can determine the sources of such discrepancy. Specifically, this approach entails first using an omnibus multivariate test (based on the overall Wilks’ ?) to examine whether the relationship between the antecedents and the two sides of organization-employee relationships was significant overall. When the test is significant, one can further examine regression coefficients with the organization’s perceptions and the employees’ perceptions as separate dependent variables.
72
With congruence of perceived organization-employee relationship characteristics as the predictor, Edwards (1994, 1995) suggested using the polynomial regression procedure and surface response tests. This procedure is based on the following three guiding principles. First, the relationship between congruence and its outcomes should be viewed in three dimensions. Perceived organization-employee relationship characteristics have two components (employees vs. organizations), the distinctiveness of which need to be retained in data analysis. Second, the relationship between congruence and a particular outcome can be shown as a three-dimensional response surface, another focus of analysis. Third, any constraints imposed on the data, such as those in the difference-score approach, need to be included in hypotheses and tested (Edwards, 1994). Using this procedure, one needs to regress each outcome variable on the components, the squared components, and the product of two components, and examine their response surfaces. These analytic procedures proposed by Edwards were used in this study. To sum up my discussion, I present the third proposition and second research question on congruence of perceived organization-employee relationship characteristics: Proposition 3: Congruence of employees’ and organizations’ (represented by managers’ perceptions) perceptions of their relationship characteristics best captures the relational dynamics. These relationship characteristics are made up of five components: trust, distrust, control mutuality, satisfaction, and commitment. Research Question 2: How do non-management employees and managers perceive the organization-employee relationships?
73
Organizational Effectiveness Organizational Effectiveness: Definition and Approaches “Organizational effectiveness is the ultimate question in any form of organizational analysis” (Hall, 1980, p. 536). The earliest approach to organizational effectiveness in the 1950s viewed organizational effectiveness as “the degree to which an organization realizes its goals” (Robbins, 1990, p. 49), which is evidently ambiguous. The goals can be either short-term or long-term or both; non-management employees’ goals or top executives’ or both; officially stated goals or actual goals. With the proliferation of organizational effectiveness studies in the next two decades, researchers agreed that organizational effectiveness may have no fixed operational definition because organizations can be evaluated as effective or ineffective on many different criteria that sometimes are independent of each other (Robbins, 1990). Such criteria included productivity, profit, absenteeism, turnover, motivation, morale, and stability, among others (Campbell, 1977). Without having to agree upon one standard definition, organization theorists adopted four major approaches to studying organizational effectiveness, each with a distinct definition (Robbins, 1990; J. Grunig & Y. Huang, 2000; L. Grunig et al., 1992). They are the goal attainment approach, the systems approach, the strategic constituencies approach, and the competing-values approach. The goal attainment approach. According to Robbins (1990), the goal attainment approach posits that an organization is effective when it accomplishes its goals (e.g., profit maximization). It assumes organizations as rational and goal-seeking entities. Other assumptions associated with this approach are:
74
First, organizations must have ultimate goals. Second, these goals must be identified and defined well enough to be understood. Third, these goals must be few enough to be manageable. Fourth, there must be general consensus or agreement on these goals. Finally, progress toward these goals must be measurable. (Robbins, p. 54) This approach is fraught with problems. Specifically, the assumed consensus on goals is hard to find in organizations because of the political nature of organizations (L. Grunig et al., 1992). The dominant coalition may desire goals different from the organization’s external publics, such as community leaders and activist groups. The officially stated goals of the organization may not be the actual goals. These different goals may contradict each other, which presents difficulty in evaluation of organizational effectiveness (Robbins, 1990). The systems approach. The systems approach suggests that organizational effectiveness is appraised by an organization’s ability to acquire resources from the environment that are critical to its survival, sustain itself internally as a social organism, and interact successfully with the external environment. Unlike the goal attainment approach that emphasizes the ends, the systems approach underscores the ends less but the means more (Robbins, 1990). The systems approach has its limitations too, mainly related to measurement and the issue of whether the means to achieve ends matter (Robbins, 1990). Measuring end goals is much easier than assessing process variables, such as “flexibility of response to environmental changes” (Robbins, p. 61). Furthermore, critics questioned whether organizations care about what means are used to achieve the end goals if the end goals
75
are accomplished (Robbins). In addition, the ultimate goal as defined in the systems approach is survival, which is extremely vague (J. Grunig & Y. Huang, 2000). The key concept environment is also vaguely defined. The strategic constituencies approach. The strategic constituencies approach states that organizations are effective when they satisfy the demands of the constituencies in their environment that are crucial for their continued existence (Robbins, 1990; J. Grunig & Y. Huang, 2000; L. Grunig et al., 1992). With an emphasis on the environment similar to the systems approach, the strategic constituencies approach however is not concerned with all elements in the organization’s environment. It is only concerned with elements of the environment that can threaten the organization’s survival. In contrast with the goal-attainment approach that assumes organizations as rational and goal-seeking entities, the strategic constituencies approach regards organizations as political entities wherein multiple interests compete for control over resources (Robbins, 1990). It also assumes that organizations have multiple constituencies in the environment who have varying levels of power and different agenda. Last, the strategic constituencies approach assumes that managers in the organization seek various goals, none of which is value free. Robbins (1990) pointed out three problems with the strategic constituencies approach: 1) difficult to separate the strategic constituencies from the larger environment; 2) hard to further separate the strategic from the “almost” strategic constituencies; 3) challenging to accurately identify the expectations of these strategic constituencies about the organization. Although the strategic constituencies approach did not suggest how to overcome these problems, public relations researchers provided solutions (Dozier et al.,
76
1995; L. Grunig et al., 1992; J. Grunig, 1997). These difficult tasks are precisely the job of public relations managers. They use environmental scanning to “identify emerging or existing publics that could either help or hurt” organizations, who are also known as strategic constituencies (Dozier et al., p. 199). The situational theory also suggests ways to segment publics into all-issue publics, apathetic publics, single-issue publics, and hotissue publics; or into latent publics, aware publics, and active publics in terms of their development stages (J. Grunig, 1997). The competing-values approach. Unlike the systems approach that emphasizes means and goal attainment approach that underscores ends, the competing-values approach focuses on both means and ends (Robbins, 1990; L. Grunig et al., 1992). This approach posits that “the criteria you value and use in assessing an organization’s effectiveness—return on investment, market share, new-product innovation, job security—depend on who you are and the interests you represent” (Robbins, p. 68). This approach assumes the existence of multiple criteria of evaluating organizational effectiveness and the unlikelihood of having the “best” criterion. It also assumes that these different criteria can be organized. Given the multiple criteria of effectiveness and conflicting interests, the management needs to use the criteria of effectiveness espoused by its strategic constituencies if the organization is to survive and prosper (Robbins, 1990). But strategic constituencies change over time. Therefore organizational theorists (e.g., Cameron & Whetten, 1981; Quinn & Cameron, 1983) connected this approach with organizations’ life-cycle stages, stating that organizations should emphasize different strategic constituencies at different stage of its life span. Robbins held that the competing-values
77
approach faces the same problems that the strategic constituencies approach encounters. As discussed previously, the public relations literature provided ways to solve these problems. In sum, the above four approaches all have strengths and problems. It is not necessarily one approach is superior to another. An integration of these approaches led to my conceptualization of organizational effectiveness as the degree to which an organization incorporates the values and demands of its strategic constituencies into its long-term (ends) and short-term (means) goals that it achieves. Employees are certainly strategic constituencies to an organization. Their values and demands should be incorporated in the goals of organizations. Just as Campbell (1977) listed, employees’ turnover and productivity are two important organizational effectiveness criteria. Therefore, I chose contextual performance and contextual performance as the variables to reflect organizational effectiveness. Contextual Performance As a criterion of organizational effectiveness, contextual performance matters markedly to an organization because high levels of activities involved in one’s contextual performance contribute to the accomplishment of organizational goals (Borman & Motowidlo, 1993, 1997; Goodman & Svyantek, 1999; J. Hogan et al., 1998; Motowidlo, 2000; Motowidlo & Van Scotter, 1994; Van Scotter, 2000). Furthermore, contextual performance-related activities tend to be similar across jobs (Borman & Motowidlo, 1997), rendering comparisons between organizations possible. Contextual performance is a distinct component of individuals’ job performance behaviors (e.g., Borman & Motowidlo, 1993, 1997; J. Hogan et al., 1998; Motowidlo &
78
Van Scotter, 1994). Job performance in essence is “the degree to which an individual helps the organization reach its goals” (Motowidlo, Borman, & Schmit, 1997, p. 72). Contextual performance has been defined as individuals’ behavior that “creates and maintains the psychological, social, and organizational environment” in which one’s work is performed (J. Hogan et al., 1998, p. 190). Borman and Motowidlo (1993) introduced the term contextual performance out of concern that research and practice in employee selection largely ignored or downplayed the non-task-related performance by employees, which are often not in one’s formal job descriptions. They emphasized, “[These activities] contribute to the organization’s technical core either directly by implementing a part of its technological process, or indirectly by providing it with needed materials or services” (p. 73). The concept of contextual performance that Borman and Motowidlo (1993) proposed has five components (Borman & Motowidlo, 1997; J. Hogan et al., 1998). They are: 1) persisting with enthusiasm and extra effort to achieve task goals; 2) volunteering to carry out actions that are not formal responsibilities of an employee; 3) helping and cooperating with others; 4) adhering to organizational rules/procedures even when inconvenient for an employee; and 5) supporting organizational objectives (Borman & Motowidlo, p. 102). Typical examples of these components of contextual performance include: extra work on one’s job, suggestions to improve organizational performance, assistance to coworkers, meeting deadlines, and complying with organizational policies. It is noteworthy that, as Motowidlo (2000) explicated, these components are not caused by contextual performance; rather, aggregation of all these behaviors or behavioral episodes by employees is contextual performance.
79
Contextual performance overlaps considerably with organizational citizenship behavior that was introduced by Organ and his colleagues (Bateman & Organ, 1983; Smith, Organ, & Near, 1983). Yet these two constructs have different roots and seek to answer different questions. Organ formally defined organizational citizenship behavior as “individual behavior that is discretionary, not directly or explicitly recognized by the formal reward system, and that in the aggregate promotes the effective functioning of the organization” (p. 4). Measures of organizational citizenship behavior were developed by asking managers what they would want their employees to do but could not require them to do. Central to the definition is “discretionary” and “unrewarded.” In contrast, contextual performance is not necessarily discretionary or nonrewarded (Organ, 1997). As indicated earlier, contextual performance taps what is not covered by task performance. Now it seems these two constructs are clearly different. Organ (1997) would not let the dust settle. He redefined organizational citizenship behavior as “contributions to the maintenance and enhancement of the social and psychological context that supports task performance (or the technical/technological/production system” (p. 91). This definition is virtually the same as contextual performance. Built upon Organ and his colleagues’ work, studies that use the term organizational citizenship behavior typically have a number of similar (if not identical) patterns of behavior included in contextual performance. Therefore, when connecting contextual performance to dimensions of organization-employee relationships, I relied on literature in areas of contextual performance as well as organizational citizenship behavior.
80
Turnover Intention Employees make up organizations. As many management researchers regard employees the most important asset to an organization (e.g., Collins, 2001; C. Pettijohn, L. Pettijohn, & Taylor, 2007; Van Scotter), their intention to leave, i.e., turnover intention is a crucial criterion of organizational effectiveness. Whether an employee intends to stay with or leave an organization shows the extent to which the organization incorporates employees’ demands and values in their goals that it achieves, namely, organizational effectiveness as defined in this study. Also, turnover intention is commonly studied as the ultimate outcome in research on human behavior in organizations (e.g., Brashear, Boles, Bellenger, & Brooks, 2002; Chandrashekaran, McNeilly, Russ, & Marinova, 2000; Z. Chen & Francesco, 2000; Lambert, N. Hogan, & Barton, 2001; Tett & Meyer; C. Pettijohn et al., 2007). Core turnover models in management literature have incorporated turnover intention (behavioral intention); and empirical research has shown turnover intention as one of the strongest predictors of actual turnover behavior (e.g., Mobley, 1977; Hom, Caranikas-Walker, Prusia, & Griffeth, 1992; Steers & Mowday, 1981). Therefore, this study used turnover intention as a variable of organizational effectiveness. Turnover intention has been conceived of as “a conscious and deliberate willfulness to leave the organization” (Tett & Meyer, 1993, p. 262). It is one’s behavioral intention. In terms of measurement, it is often associated with a particular time interval, such as intention to leave within the next six months or within the next year.
81
Model and Hypotheses: Linking the Variables Research on Relationship Maintenance Strategies as Antecedents In public relations research, few researchers have focused on relationship maintenance strategies. Only two studies (i.e., Y. Huang, 1997; Ki & Hon, 2007a) developed a measuring instrument of relationship maintenance strategies to date. A handful of studies examined relationship maintenance strategies mainly through qualitative interviews, case studies, and content analysis of company Web sites (Bortree, 2003; Hong & Kiousis, 2007; Hung, 2002, 2003; Hung & Y. Chen, 2007; Kelleher & Miller, 2006; Ki & Hon, 2003, 2006, 2007a; Wigley, 2003). In a comprehensive content analysis of journal articles on organization-public relationships from 1985 through 2004, Ki & Shin (2005) located only 38 articles from leading communication and management journals, such as Public Relations Review, Journal of Public Relations Research, Journalism & Mass Communication Quarterly, Academy of Management Review, and Journal of Communication Management. They pointed out that 18 articles suggested at least one maintenance strategy. Six of them have been studied in most of these articles: access, positivity, openness, assurances (of legitimacy), networking (i.e., participation in mutual networks), and sharing of tasks (i.e., shared tasks). Networking was the most frequently used strategy. With regard to linking relationship maintenance strategies to perceived organization-public relationship characteristics, Hung (2003) found that the maintenance strategies of positivity and shared tasks were used in achieving all four dimensions of relationship characteristics (trust, satisfaction, control mutuality, and commitment) by multinational companies in China. Similar findings were drawn by Hung and Y. Chen
82
(2007) in a meta-analysis of three studies that they conducted. Likewise, Jo and Shim’s (2005) study on organization-employee relationships indicated that employees’ perceived support from the management (analogous to positivity) who engage in empathetic, meaning-making and direction-giving communication is positively associated with the formation of trusting relationships with the management. Hung (2002) studied how multinational and Taiwanese companies built relationships with their publics in China. She found that symmetrical strategies (assurances of legitimacy and openness) were helpful in achieving integrity, a component of trust; shared tasks (as part of assurances of legitimacy) contributed to gaining competence, another component of trust. Both multinational and Taiwanese companies employed symmetrical strategies (shared tasks as part of assurances of legitimacy and networking) to show commitment to and ensure satisfaction of their publics. Multinational companies utilized symmetrical strategies such as networking and assurances of legitimacy to obtain control mutuality whereas the Taiwanese companies resorted to both symmetrical strategies (e.g., openness, assurances of legitimacy, and shared tasks as part of assurances of legitimacy) and asymmetrical ones (e.g., obligation to others, similar to compromising and accommodating). Specific to employee publics, the excellence study by L. Grunig et al. (2002) found that symmetrical communication with employees makes it more likely that employees are satisfied with their organizations, and consequently less willing to leave the organization (akin to turnover intention). In a structural equation model including symmetrical communication, job satisfaction, participative culture, organic structure, and mechanic structure, L. Grunig et al. found that symmetrical communication was
83
significantly and positively related to job satisfaction. Their measures of symmetrical communication (e.g., “Most communication between administrators and other employees in this organization can be said to be two-way communication”; “This organization encourages differences of opinion”) were similar to those of some relationship maintenance strategies (openness/disclosure and assurances of legitimacy). The indicators of job satisfaction (e.g., “I have found this organization to be a good place to work”) are similar to those of satisfaction, a component of perceived relationship characteristics (“Most people like me are happy in their interactions with this organization”). In addition, as J. Grunig (2006) stated, “the concept of relationship cultivation strategies is the heir to … two-way symmetrical model … Cultivation strategies identify specific ways in which symmetrical communication can be used to cultivate relationships” (p. 168). Taken together, it means that symmetrical maintenance strategies are positively related to perceived organization-employee relationship characteristics. Studies in the interpersonal communication literature and negotiation literature have also found associations between relationship maintenance strategies and components of perceived relationship characteristics (e.g., M. Allen, 1992; Canary & Stafford, 1994; Canary & Zelley, 2000; Guerrero, Eloy, & Wabnik, 1993; PunyanuntCarter, 2008; Putnam, 1990; Stafford, 2003; Stafford et al., 2000; Stafford & Canary, 1991). From the perspective of interpersonal communication, Canary and Stafford (1994) summarized that shared tasks, networking, and assurances of legitimacy (symmetrical relationship maintenance strategies) are positively tied to control mutuality; and that assurances of legitimacy are strongly and positively connected with commitment and
84
trust. Stafford and Canary (1991) found a positive association between individuals’ symmetrical maintenance strategies (openness, assurances of legitimacy, networking, and shared tasks) and relational satisfaction. In a similar vein, Stafford et al. (2000) found that symmetrical strategies such as assurances of legitimacy were primary predictors of control mutuality, satisfaction and commitment. Likewise, Punyanunt-Carter’s study of 198 father-daughter dyads reported that seven relationship maintenance strategies (openness, positivity, assurances of legitimacy, social networks or participation in mutual networks, conflict management, advice, and shared tasks) influenced participants’ perceptions of satisfaction. In addition, Guerrero et al. reported that openness (or disclosure), positivity, assurances (of legitimacy), participation in mutual networks, and shared tasks predict whether a couple stays or leaves a relationship (similar to relational commitment). Organizational theorists such as Putnam (1990) noted that integrative negotiation (a symmetrical strategy, as part of assurances of legitimacy in this study) can cultivate trust. Comparable patterns were found by M. Allen (1992) who examined links between communication, organizational commitment, and perceived organizational support and concluded that perceived support was a mediator between communication and organizational commitment. She defined perceived organizational support as employees’ perceptions of “the extent to which an organization values its employees’ contributions and cares about their well-being,” analogous to a few maintenance strategies explicated above, such as assurances of legitimacy (p. 359). With regard to congruence of perceived relationship characteristics, a related study by Shen and J.-N. Kim (2008) found that symmetrical communication strategies are
85
positively linked to organization-employee congruence on ethical values. Some symmetrical maintenance strategies discussed in foregoing sections are similar to the symmetrical communication strategies that Shen and J.-N. Kim measured, such as advice and disclosure. They defined congruence as compatibility between people’s and organization’s characteristics. To take one step further, one can then propose that symmetrical strategies can be positively associated with organization-employee congruence on their perceptions of relationship characteristics. On a similar note, Weigel and Ballard-Reisch (1999) who studied marriage relationships looked at the influence of relationship maintenance behaviors on joint couple-level perceptions of marital quality. Similar to this study, the authors argued that marriage relationship is more than the sum of the individual perceptions of husbands and wives. As opposed to the traditional view that examines couple relationship from the perspectives of either husbands or wives, Weigel and Ballard-Reisch posited that communication behaviors, such as communicative relationship maintenance behaviors by husbands and wives help bring about couples’ “couple marital quality.” They suggested that measures of spouses’ perceptions can be used to assess the shared couple reality. Their study of 129 married couples indicated that individuals’ use of maintenance behaviors (positivity, openness, assurances of legitimacy, networking, and shared tasks) were positively related to couples’ marital relationship quality (satisfaction, commitment, and love). In particular, wives’ use of maintenance behaviors had a stronger relationship with couples’ marital relationship quality than did the husbands’. Their conceptualization of joint couple relationship quality as shared couple reality is akin to congruence of perceived relationship characteristics in this study. Extending it to organization-employee
86
relationships, this suggests that an organization’s symmetrical (asymmetrical) relationship maintenance strategies can be positively (negatively) related to the congruence of perceived organization-employee relationship quality. To sum up the above discussion, I proposed the following hypothesis: H1: Symmetrical relationship maintenance strategies are positively correlated with congruence of perceived relationship characteristics (H1a), and asymmetrical strategies are negatively associated with congruence of perceived relationship characteristics (H1b). Congruence of Perceived Relationship Characteristics and Organizational Effectiveness Outcomes Organizational behavior and management researchers have extensively examined the effect of organizational commitment on contextual performance and turnover intention (e.g., N. Allen & Meyer, 1990; Brashear et al., 2002; S. Brown & Peterson, 1993; Chang, 1999; Z. Chen & Francesco, 2000; Gruen et al., 2000; Jaros, 1997; Morgan & S. Hunt, 1994; Mowday, Porter, & Steers, 1982; Russ & McNeilly,1995; Shore & Martin, 1989; Tett & Meyer, 1993; Williams & Hazer, 1986). Weiner (1982) stated that commitment acts as “an intervening process, mediating between certain antecedents and behavioral outcomes … [It] can be viewed as a motivational phenomenon” (p. 419). Gruen et al. found that commitment (normative and affective) partially mediates the effect of relationship-building efforts on members’ participation or involvement in their professional associations, analogous to a component of contextual performance. Brashear et al.’s research of 402 salespeople revealed a significant and negative relationship
87
between commitment and turnover intention. Russ and McNeilly established the predictive effect of satisfaction and organizational commitment on turnover intention. Shore and Martin’s (1989) survey of 72 professional staff in 15 departments of a large hospital and 71 bank tellers in 11 branches of a large bank investigated the effects of organizational commitment and job satisfaction on turnover intention and job performance (task performance, not contextual). Results indicated that commitment was more strongly and negatively associated with turnover intention than job satisfaction for the tellers, but not for the hospital staff. Satisfaction was more strongly related to supervisory ratings of performance than commitment for both samples. Psychologists Tett and Meyer (1993) used path analysis in a meta-analysis of 178 independent samples from 155 studies to examine the effects of satisfaction and organizational affective commitment on turnover intention and actual turnover behavior. They found that satisfaction and commitment each has independent effect on turnover intention. Turnover intention mediates the effect of the predictors on actual turnover behaviors. Their conceptualizations of satisfaction and organizational affective commitment coincide with the definitions in this study. They also relied on Meyer and N. Allen’s (1991) three forms of organizational commitment—affective, continuance, and normative commitment. Trust (e.g., Aryee, Budhwar, & Z. Chen, 2002; Brashear et al., 2003) was also shown to have substantial influence on turnover intention, task performance, and contextual performance. Based on an SEM analysis of 179 supervisor-subordinate dyads (179 subordinates and 28 supervisors) from an India-based company, Aryee et al. reported a significant and negative path between workers’ trust in the organization and
88
turnover intention, a significant and positive path between workers’ trust in their supervisors and contextual and task performance. Dimensions such as trust and power were also found to predict contextual performance. Campbell (1990), in a large study on the U.S. Army (N = 4,039), used general cognitive ability, achievement, dependability and adjustment to predict task performance and contextual performance. He found that dependability has a more substantial influence on contextual performance than on task performance. Dependability is a sub-dimension of trust. Likewise, Motowidlo and Van Scotter (1994) found that dependability and dominance have significantly higher correlations with contextual performance than with task performance. A meta-analysis by Organ and K. Ryan (1995) of attitudinal correlates of organizational citizenship behavior revealed that perceived fairness (a sub-dimension of trust), satisfaction, perceived leader supportiveness, and organizational commitment (dimensions of organization-employee relationships) predicted behaviors such as altruism and compliance, which were identified as components of contextual performances (e.g., Borman & Motowidlo, 1993, 1997; Goodman & Svyantek, 1999; Motowidlo et al. 1997). Likewise, scholars including Bateman, Organ (1983) and Smith et al. (1983) and Organ (1997) who were credited for coining the term organizational citizenship behaviors that overlaps with contextual performance, proposed the term because of their belief that employees’ job satisfaction could lead to such behaviors. In addition to the aforementioned studies that showed the predictive effect of dimensions of perceived relationship quality on contextual performance and turnover intention, congruence was found to be a predictor of contextual performance and turnover
89
intentions as well (e.g., Arthur, Bell, Doverspike, & Villado, 2006; Goodman & Svyantek, 1999; Kristof-Brown, Zimmerman, & Johnson, 2005; Saks & Ashforth, 2002; Vancouver & Schmitt, 1991; Van Vianen, 2000). Goodman and Svyantek reported that congruence of perceived organizational culture and the discrepancy between employees’ ideal and perceived actual organizational culture are positively associated with their contextual performance. Kristof-Brown et al.’s comprehensive meta-analysis of literature on personorganization congruence and other types of person-environment congruence indicated positive associations between congruence and contextual performance but negative relationships between congruence and turnover intention. Day and Bedeian’s (1995) study of 206 nursing service employees indicated that organization-employee congruence on conscientiousness positively influenced employees’ job performance, but had a negative impact on employees’ turnover. Saks and Ashforth’s longitudinal study on job search, congruence perceptions, and employment quality followed 113 students prior to their organizational entry and four months after entry. They found a negative association between person-job congruence (example item: “To what extent is the organization a good match for you?”) and intention to quit a job. Finally, the Attraction-SelectionAttrition framework (Schneider, 1983, 1987) posits that individuals whose characteristics are similar to those of an organization are more likely to be selected in the hiring process and choose to stay on the job. Extending this body of literature to congruence of organization-employee relationship characteristics, I hypothesized that it is positively related to employees’ level of contextual performance but negatively tied to their likelihood of turnover intention. Specifically, higher congruence of perceptions of their relationship characteristics by the
90
employees and the organization (as often represented by the management) will be linked to employees’ higher contextual performance (such as helping others, and voluntarily working extra hours) and lower turnover intention. Therefore, I propose the following hypotheses: H2: The level of congruence of perceived relationship characteristics is negatively associated with the likelihood of turnover intention. H3: The level of congruence of perceived relationship characteristics is positively associated with the level of contextual performance. These hypotheses are shown in a structural model (see Figure 2). Summary In summary, after arguing for the need of theory-building on internal relationship management, I extended the dialectical approach to relationship management and integrated existing relationship models in public relations to refine the general theory of relationship management proposed by Ledingham (2003). I critiqued literature from different disciplines on conceptualizations of organization-public relationships, relationship maintenance strategies, congruence of perceived organization-employee relationship characteristics, and behavioral outcomes of organizational effectiveness, and proposed a new three-stage model of organization-employee relationships. I also discussed issues related to measurement of organization-public relationships and suggested a new analytic framework. In order to test relationship management inside organizations, I have developed the following propositions, research questions, and hypotheses in this chapter:
91
Proposition 1: The dynamic and constant tension and change in organizationpublic relationships around negotiated interests and goals lead to benefits of varying degrees for the organization and its publics and better understanding of each other. Proposition 2: Symmetrical relationship maintenance strategies are composed by disclosure or openness, assurances of legitimacy, networking, and advice. Asymmetrical relationship maintenance strategies include distributive negotiation, avoiding, and compromising. Research Question 1: How do organizational members perceive relationship maintenance strategies used by their organizations? Proposition 3: Congruence of employees’ and organizations’ (represented by managers’ perceptions) perceptions of their relationship characteristics best captures the relational dynamics. These relationship characteristics are made up of five components: trust, distrust, control mutuality, satisfaction, and commitment. Research Question 2: How do non-management employees and managers perceive the organization-employee relationships? H1: Symmetrical relationship maintenance strategies are positively correlated with congruence of perceived relationship characteristics (H1a), and asymmetrical strategies are negatively associated with congruence of perceived relationship characteristics (H1b). H2: The level of congruence of perceived relationship characteristics is negatively associated with the likelihood of turnover intention. H3: The level of congruence of perceived relationship characteristics is positively associated with the level of contextual performance.
92
Symmetrical Relationship Maintenance Strategies
?
H2 H1a
?
Congruence of Perceived Relationship Characteristics
Turnover Intention ?
H3
H1b
Asymmetrical Relationship Maintenance Strategies
Contextual Performance
Figure 1. The structural model of internal relationship management theory. In this figure, Maintenance Strategies vary from high to low likelihood of using certain strategies (symmetrical or asymmetrical4). Congruence of Perceived Relationship Characteristics varies from high to low congruence, empirically represented by employees’ perceptions of relationship characteristics (person variable), organizations’ perceptions of relationship characteristics (organization variable), squares of employees’ perceptions of relationship characteristics (squared person variable), squares of organizations’ perceptions of relationship characteristics (squared organization variable), and product of person variable and organization variable. Turnover Intention varies from high to low likelihood of intention to leave an organization. Contextual Performance5 is participants’ evaluation of their own citizenship behaviors, ranging from high to low level of performance.
Symmetrical maintenance strategies are defined as strategies that are geared towards mutual or joint gains of both relational parties whereas the asymmetrical ones are focused on maximization of self-gains. 5 Contextual performance is defined as individuals’ behavior that “creates and maintains the psychological, social, and organizational environment” in which one’s work is performed (Hogan, Rybicki, Motowidlo, & Borman, 1998, p. 190). It complements task-related performance.
4
93
Examining these propositions, research questions, and hypotheses helped develop a theory of internal relationship management, moving the general theory of relationship management further. It not only provided a newer conceptualization of organizationpublic relationships and a fuller understanding of relationship management inside organizations, but also demonstrated the contributions public relations function makes to organizational effectiveness — the ultimate goal for which organizations build relationships with their publics.
94
Chapter III Methodology The survey method was used in this dissertation. This chapter discusses the choice of this method, sampling strategy, the data collection procedure, and data analysis techniques. Survey Fink (2006) defined surveys as “information collection methods used to describe, compare, or explain individual and societal knowledge, feelings, values, preferences and behavior” (p. 1). It is appropriate for both exploratory and explanatory purposes. Surveys can help “determine the incidence, distribution, and interrelations among …variables, and…the vital facts of people, and their beliefs, opinions, attitudes, motivations and behavior” (Kerlinger & Lee, 2000, p. 600). Respondents’ beliefs, opinions, attitudes, motivations, and behaviors are central to this study, such as perceptions of relationship maintenance strategies and relationship characteristics, and self-reported contextual performance behavior. Why Survey? Randomized experimental research designs are preferred if one wants to maximize internal validity and infer causality from the association between variables (Hoyle, Harris, & Judd, 2002). However, not all variables are manipulable (Hoyle et al.). As a result, survey method6 is chosen to assess people’s attitudes, beliefs, and behaviors, and draw causal inferences about the sources of such attitudes, beliefs, and behaviors (Weisberg, Krosnick, & Bowen, 1996). Statistical methods for multivariate data analysis,
A survey can be a true randomized experiment. Here the term survey method does not involve manipulation of independent variables and random assignment of participants into experimental conditions.
6
95
such as structural equation modeling (SEM), are helpful to test adequacy of one’s proposed models and to draw causal inferences. In the case of this study, manipulating independent variables such as the organizations’ relationship maintenance strategies and randomly assigning employees to different levels of the independent variables was not logistically realistic and possible. Thus the survey method was a better choice than randomized experiments. On the other hand, surveys have a number of weaknesses. First, survey research often gathers self-reported information. Social desirability biases can be a problem, which compromises validity (Weisberg et al., 1996). This occurs for example when sensitive topics are touched upon. Second, the typical format of close-ended questions in surveys makes it hard to contextualize respondents’ answers (Krosnick, 1999). Third, when data are collected face to face or over the telephone by researchers, interviewer effects create biased data. Specific problems include “reading errors, speech variations, improper probes, and unprogrammed feedback to the respondent” (Weisberg et al., p. 118). Lastly, surveys with low response rates have been regarded as non-representative, i.e., lacking external validity. However, research found that surveys with lower response rates can be more accurate than those with higher response rates (see Visser, Krosnick, Marquette, & Curtin, 1996). Strengths and Weaknesses of Online Surveys Surveys can be conducted in person, over the phone, via mail or email, or teleconference (Fink, 2006). Compared with mailed questionnaires, telephone surveys, and face-to-face surveys, the biggest advantages of electronic surveys are convenience and efficiency. Online surveys save more time and money than telephone surveys and
96
face-to-face surveys in terms of survey distribution, data collection, respondent reminding, and data entry, among others (Fink; Schmidt, 1997). Participants can choose to fill out questionnaires at their convenience and keep personal records (Neuman, 1994). This method allows researchers to obtain data instantly and be downloaded to a spreadsheet (Fink, 2006). In addition, it has the “ability to make complex skip pattern questions invisible to the respondent” (Fink, p. 9). Lastly, the dynamic features of online surveys can increase participant motivation, because participants can receive feedback right after filling out information (Schmidt). However, online surveys have a few disadvantages compared with other survey modes, including mailed questionnaires, telephone surveys, and face-to-face surveys. First and foremost, its advantage of convenience can become a disadvantage. The researcher can have little control over the conditions under which an electronic survey is completed, which may result in invalid and unreliable data. For example, questionnaires may not look the same on different monitors. Technology may be unstable and create difficulties for respondents to complete surveys. Respondents may have different levels of computer expertise (Fink, 2006). In a related vein, a common problem with online surveys is incomplete answers (Schmidt, 1997). Respondents may overlook certain questions or simply do not want to provide answers, which is hard to determine unless researchers force explanations in their questionnaire design. When incomplete answers only occur in certain demographics, analysis without taking into account this will lead to biased findings. Third, online surveys are only limited to respondents who know how to use a browser and have ready access to the Internet. Research has shown that typical Web users
97
have higher socio-economic and education status than the general population (Schmidt, 1997). Measures should be taken by researchers to control the impact of such biases on research findings. For example, researchers may screen participants to meet certain demographics criteria or use demographics variables as control variables (Schmidt). Nevertheless, considering the fact that face-to-face or telephone surveys might make participants hesitant to answer certain questions that they view as sensitive, I decided to choose electronic surveys for the ensured confidentiality and anonymity that can make participants feel more at ease. The questions on organization-employee relationships, turnover intentions, and contextual performance can be sensitive to many respondents. Also, the missing-data problem is present in mailed questionnaires. Online surveys technology enables researchers to remind respondents when they forget to answer questions. Combined with the efficiency and convenience of the method, I chose online surveys for data collection. Sampling The Appropriateness of Non-Probability Sampling Probability sampling is often recognized as superior to non-probability sampling; however non-probability sampling is commonly used because it is not as expensive and/or statistical generalizability is not necessary (Hoyle et al., 2002). For example, when one obtains a purposive sample of foods from a well-known cuisine, the purpose is to understand the variety of elements available in the population, but not to estimate population value—statistical generalizability being unnecessary. In addition, a probability sample is not necessary for making theoretical generalizations. As Shapiro (2003) argued, maximizing the ability to generalize means
98
more than probability sampling. A study should be examined for its contributions to theoretical development and its choice of theoretically driven variables. Provided that theoretical generalization was the focus of this study, the non-probability sampling method was deemed acceptable. Sample Size Estimation and Procedures Given that I used 60 items for the four constructs, the hypothesized model (see Figure 2) has degrees of freedom of 1,830, using the counting rule—([p*(p+1)/2] – total number of parameters to be estimated]). If I were to use composites to represent the indicators, then symmetrical relationship maintenance strategies had four composites, asymmetrical ones had three, congruence of perceived relationship characteristics had five, turnover intention had one, and contextual performance had five. Consequently, the degrees of freedom would become [18* (18+1)/2] -40 parameters to be estimated = 131. Based on Hancock’s (2006a) procedure of a priori power analysis for testing data-model fit as a whole, with a df =131, ?1 = .02, to have a level of power ? = .80 for the ? = .05level test associated with the root mean square error of approximation, the estimated sample size would be approximately 145. In other words, given that the true level of datamodel fit in the population is ?1 = .02, to have a .80 probability of rejecting ?0 ? .05 in favor of acceptable data-model fit using an ? = .05-level test, this study would need a minimum of 145 participants (see Hancock, 2006a, for a detailed discussion of arriving at this estimated sample size). However, Hancock further added a general rule: When ?1 = .02 and models have df ? 60, to achieve power of ? = .80 sample sizes of n = 300 appear to suffice for testing overall data-model fit … For models with
99
df ? 30 when ?1 = .02, sample sizes of around n = 500 appear to be adequate. (p. 103) Therefore, I decided on an estimated minimum sample size of 500. As elaborated in Chapter II, I used managers’ perceptions of relationship characteristics to represent the organization side of an organization-employee relationship, i.e., the organization variable. To ensure enough variance on the organization variable, I needed multiple managers within a single organization and multiple organizations. In terms of the actual number of organizations and number of participants (non-management employees plus managers), past research has suggested 30 organizations with 30 individuals in each organization or 150 organizations with 5 individuals in each organization (Hofmann, 1997) or an average of 10 employees in 113 bank branches (Ostroff, Shin, & Kinicki, 2005). Taking into consideration the minimum number of participants for the overall model, I successfully recruited on average 26.17 participants within a single organization (with a 1: 3 ratio between the managerial and non-management employees) and 30 organizations. I tried to recruit participants from different departments (within each organization) and have organizations of different sizes from different industries to ensure variance on the person variable (employees’ perceptions of relationship characteristics) as well as the organization variable (organizations’ perceptions of relationship characteristics). Pilot Studies Pilot Study 1 Sample and procedures. The objective of this pilot study was to test if distrust was a new dimension of perceived organization-employee relationship characteristics
100
distinct from trust. One hundred and forty-nine students from a large university in Northern China were recruited to complete paper-and-pencil questionnaires with 7-point Likert-type items on trust, distrust, and symmetrical communication strategies. Participants were between 18 and 22 years of age, and were predominantly female (81.9 %). Eighty were freshman, 61 were sophomore, and eight did not report their year in school. Five items on distrust were adapted and revised from Cho’s (2006) items. Six items on trust from Hon and J. Grunig (1999) and eight items on symmetrical communication strategies from L. Grunig et al. (2002) were used. Based on the meaning of these items, I wrote a Chinese version (not literal translation). The Chinese version of items was sent to two Chinese public relations professors and an English studies professor in China. Revisions were made based on their comments. The items were back translated into English to ensure their comparability. Results. The reliability of the trust, distrust, and symmetrical communication items were obtained (Cronbach ? = .863 for trust, .827 for distrust, and .600 for symmetrical communication). I conducted exploratory factor analysis using principal component analysis with Varimax rotation. Four factors with eigenvalues greater than 1 emerged, with trust items loading highly on one factor, distrust on another, and symmetrical communication on two factors (eigenvalue = 10.45, 65.337% variance explained). I then did exploratory factor analysis using principal component analysis with Varimax rotation on trust and distrust items only. Trust and distrust items loaded on separate factors (with the exception of one item of distrust “I am skeptical about whether the university will keep the students’ interests in mind when it makes decisions”)
101
(eigenvalue = 6.80, 61.786% variance explained). This finding showed preliminarily that distrust and trust may be different factors. I then conducted confirmatory factor analysis using the EQS 6.1 Program (Bentler, 2005) for trust and distrust as a two-factor oblique model (because conceptually these two factors are linked but distinct). To evaluate data-model fit, I used the oft-cited Hu and Bentler (1999) joint-criteria approach. According to the joint-criteria, a model is considered tenable when it achieves Non-Normed Fit Index (NNFI) ? .96, Comparative Fit Index (CFI) ? .96, and Standardized Root Mean Squared Residual (SRMR) ? .09 or Root Mean Squared Error of Approximation (RMSEA) ? .06 and SRMR ? .09. The model chi-square = 14.81 (df = 15, p = .47 > .05). NNFI = 1.00 > .96, CFI = 1.00 > .96, SRMR= .03 < .09. According to Hu and Bentler’s (1999) joint criteria of datamodel fit, the two-factor oblique model has excellent fit with the data. I tentatively concluded that distrust is a new dimension of perceived organization-employee relationship characteristics, linked to but distinct from trust. The English version of the distrust items developed in this pilot study was used in the dissertation study. Focusing on a different sample and in a different country, the dissertation study helped crossvalidate these items by examining its reliability and validity. I discuss measurement validity and reliability in Chapter IV. Pilot Study 2 Sample and procedures. The objective of this study was to have a final revision of the questionnaire items in terms of wording and clarity of instructions. I recruited 10 working professionals in both managerial and non-managerial positions from 10 different organizations to complete a draft questionnaire with 93 questions in March 2009. Two of
102
them filled out the questionnaire at home and emailed me revision suggestions and reported questionnaire completion time. A focus group was conducted with the remaining eight participants. Participants filled out the questionnaire and recorded the completion time. A two-hour discussion followed of each item’s meaning. Based on these results, I revised and shortened the questionnaire to 66 questions, including six questions on demographics (see Appendix A). Measures This section describes the final questionnaire (see Appendix A). Each question was in a five-point Likert format, with “totally disagree” being 1, “neutral being” 3, and “totally agree” being 5. I present items on the antecedent variables first, then on the mediating variable, and the behavioral outcomes of organizational effectiveness (see Figure 2 for the proposed model). As different reliability tests were conducted, the reliability values of all measures are reported in Chapter IV. Antecedents: Relationship Maintenance Strategies As is shown in Figure 2, symmetrical and asymmetrical relationship maintenance strategies are antecedents in the model. I discuss the measures for symmetrical and asymmetrical strategies herein. Symmetrical maintenance strategies. Twelve questions assessed four symmetrical maintenance strategies—disclosure, assurance of legitimacy, networking, and advice. Ki and Hon’s (2007a) items on disclosure, assurances of legitimacy, and networking were modified and used. Sample items included “The organization’s annual report is a valuable source of information for employees about what it has done (disclosure);” “The organization makes a genuine effort to provide personal responses to employees’
103
concerns (assurances of legitimacy);” “The organization builds coalitions with groups that help employees’ career growth (networking).” Advice items were adapted from Stafford (2000). An example was “The organization tells employees like me what to do about problems on the job” (advice). Asymmetrical maintenance strategies. Ten questions measured three asymmetrical maintenance strategies—distributive negotiation, avoiding, and compromising. Distributive negotiation items were revised and adapted from Beersma and De Dreu (2002), De Dreu, Evers, Beersma, Kluwer, and Nauta (2001) and Y. Huang (1997a). An example was “When there is a conflict between employees and the management, the management applies pressure so that employees may make concessions.” Avoiding items were revised and adapted from Hess (2003), De Dreu et al. and Y. Huang. A sample item was “When there is a conflict between employees and the management, the management gives as little attention as possible to it.” Items on compromising were adapted from De Dreu et al. A sample item was “When there is a conflict between employees and the management, the management tries to realize a middle-of-the-road solution with employees.” Mediator: Congruence of Perceived Organization-Employee Relationship Characteristics Congruence was measured by using both organizations’ and employees’ responses on the five dimensions of perceived organization-employee relationship characteristics: trust, distrust, control mutuality, commitment, and satisfaction. I present example measures of each dimension in this section.
104
Trust. I revised six items from Hon and J. Grunig (1999). An example was “I am willing to let the organization make decisions for employees like me.” Distrust. Four items developed from Pilot Study 1 were used. An example was “I feel that this organization will exploit employees’ vulnerability given the chance.” Control mutuality. Four items developed by Hon and J. Grunig (1999) were used. An example was “This organization believes the opinions of employees like me are legitimate.” Commitment. Seven items created by Allen and Meyer (1990) and one item by Hon and J. Grunig (1999) were revised and used. Example items were “I do not feel like ‘part of the family’ at my organization (reverse)” and “I feel that I have too few options to consider leaving this organization.” Satisfaction. Four items by Hon and J. Grunig (1999) were used. An example was “Most employees like me are happy in their interactions with this organization.” Dependent Variables Two variables were used to assess organizational effectiveness: turnover intention and contextual performance. I present sample items on each variable below. Turnover intention. Two items developed by O’Reilly, Chatman, and Caldwell (1991) and two items by Chen and Francesco (2000) were modified and used. A sample item was “I may leave this organization and work for another one in the next year.” Contextual performance. I pooled and modified eight items from Borman and Motowidlo (1993), Goodman and Svyantek (1999), and Smith et al. (1983). An example was “I often help other employees with their work when they have been absent.”
105
Data Collection Procedures Data collection was conducted in two phases from March 18 to June 3, 2009. First, I searched the Hoover’s Company Information Database for listed organizations with more than 200 employees and contact information of their human resources, public relations, and corporate communications personnel in Virginia, Maryland, West Virginia and the District of Columbia. I then randomly selected 150 organizations out of the total of 946 organizations, and called these organizations from March 18 to mid-April. Seven organizations agreed to participate. Only 2 organizations were included in the data analyses because there were fewer than 10 participants from each of the other 5 participating organizations. Second, I reached out to friends, acquaintances, and undergraduate students to recruit organizations between March and early June. Each of the 21 undergraduate students was paid 100 dollars for every organization they successfully recruited. They recruited organizations where they did or still were working at and where their parents and relatives were chief executives. I used a template email for each organization that I myself, my friends, and undergraduate students contacted (see Appendix B). Once the organization agreed to participate, they received a link in email. They were directed to Surveymonkey.com to read the informed consent form. Once they agreed, they would click on the “Yes, I would like to participate” button to start the questionnaire. But if they withdrew, they clicked on “No, thanks, I do not want to participate” and exited the Web site. I created a separate questionnaire (same questions, different links) for each organization to keep track of the number of responses per organization. Reminder emails
106
with the link were sent out each week until at least 25 participants were recruited from each organization. Data collected were confidential and anonymous, except that the organization names were known to the researcher. Their Web Internet Protocol (IP) addresses were blocked through Surveymonkey’s service to ensure anonymity. Finally, 48 organizations participated in the study, but 18 organizations were excluded from data analyses because there were fewer than 10 participants from each of these organizations. During the recruitment process, organizations were promised that they would have a summary report of the data if they wanted to, and participants were informed of a raffle drawing with the prize being four American Express gift cards ($50 worth each). To ensure anonymity, I included the following statement and the raffle link at the very end of the questionnaire. Thank you for your participation! To express our gratitude, we give away 4 gift cards (50 dollars value each). If you would like us to enter your name in the raffle for the gift cards, please click on (or Copy and Paste) the following link to enter a separate 1-question survey where you can put down your name and email. This way we can protect the anonymity and confidentiality of information you entered on this survey. A total of 263 participants entered their names and emails in the raffle. Even though some of these participants were from organizations that were excluded in the data analyses, I still included them in the raffle and four participants won the gift cards.
107
Data Analysis Data Aggregation As this study aimed to use managers’ perceptions to represent the organization side of employee-organization relationships, it was necessary to justify the aggregation of individual-level data in the first place. To do so, one should establish within-group agreement, or “greater than chance similarity” (Bliese, 2000, p. 350). I calculated rwg, ICC (1) and ICC (2). The index rwg compares an observed group variance to an expected random variance. ICC (1) assesses either non-independence, i.e., the extent to which variance in a dependent variable is because of group membership; or reliability of a measure—whether aggregation is justified (Kenny & Judd, 1986). When ICC (1) is calculated on independent variables, it is considered a measure of reliability. When it is calculated on dependent variables, it measures non-independence (Bliese). ICC (2) shows the reliability of group means (Bartko, 1976). Statistical Procedures for Data Analyses First, I recoded three four reverse coded items. As a few participants skipped questions, I then conducted missing value analysis (MVA) using the SPSS 16.0 Program. Next, I used the following statistical procedures to test the three propositions, two research questions, and three hypotheses. As the propositions and research questions pertain to construct validity and reliability, I performed exploratory factor analysis (EFA), Cronbach’s alpha reliability test and then confirmatory factor analysis (CFA) after removing items with low loadings for a given factor or significant cross-loadings; then conducted the Cronbach’s alpha reliability test. CFA compared different models to further examine these structures suggested by EFA.
108
To test the three hypotheses, I used multivariate regression to test H1 and polynomial regression procedures followed with surface response tests to examine H2 and H3. Distinct from a series of univariate regression analyses, multivariate regression examines the relationships among the independent variables and among the dependent variables. It allows for multivariate tests of the variance explained by all the equations jointly and provides estimates of the coefficients across equations (see Edwards, 1995 for more details of this procedure). In this study, I used the person variable and organization variable as the dependent variables jointly and the symmetrical and asymmetrical relationship maintenance strategies as the predictors. I examined the overall Wilks’ ?, a multivariate test of the two equations (i.e., for employees’ perceptions and for organizations’ perceptions) jointly. The coefficients’ direction and significance were examined next. Then, I checked whether each predictor had an equal but opposite effect on the person variable and the organization variable, a constraint that is implied but not tested in the difference-score approach. With respect to polynomial regression, I regressed each outcome variable on the mediator, represented by employees’ perceptions of relationship quality (person variable), managers’ perceptions of relationship quality (organization variable), squared person variable, squared organization variable, and the product of person variable and organization variable to specifically examine the links between the mediator and the two dependent variables (see Edwards, 1994; Edwards & Parry, 1993 for discussions of this procedure). The measures of perceived organization-employee relationships were centered on the midpoint of their scales to “reduce multicollinearity and allow for more meaningful interpretation of congruence relationships” (Ostroff et al., 2005, p. 603). Next,
109
the estimated response surfaces were examined to fully understand the relationships between the mediator and the outcome variables. Lastly, I used the four-step approach by Baron and Kenny (1986) to examine the mediation effect in the overall model (see Figure 2).
110
Chapter IV Results This study first used MVA to examine patterns of missing values, and conducted aggregation analysis to justify using managerial respondents’ data to represent the organization side of organization-public relationships. Next, descriptive statistics were obtained to understand the sampled participants and organizations. Also, reliability tests, exploratory factor analysis, and confirmatory factor analysis were performed to assess measurement validity and reliability, the research questions, and the last two propositions. Based on the discussion in Chapter II on the advantages of polynomial regression, surface response tests, and multivariate regression in examining both sides of organization-public relationships, I used these analytic procedures to test the three hypotheses. Lastly, I used the Baron and Kenny (1986) four-step approach to test the mediation effect, yielding insights into the first general proposition. In this chapter, I first discuss the preliminary analysis (MVA and aggregation analysis), then findings related to the research questions, last two propositions, as well as measurement reliability and validity, and lastly results on the hypotheses and overall model. Description of Participants Forty-eight organizations participated in this study, 18 of which were excluded from data analyses as there were fewer than 10 participants in each of these organizations. In total, 1293 participants agreed to participate, and 785 questionnaires with under 5% or fewer than 3 unanswered questions were deemed usable (see Hair, Black, Babin, Anderson, & Tatham, 2006 for discussion on handling missing values). The response rate was 61%, above the desirable benchmark of 50% suggested by Babbie (1992). Table 2
111
shows descriptions of the 30 participating organizations. Table 3 summarizes demographic information of participants. Table 2 Descriptions of Participating Organizations Org No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Industry/Type NGO, justice Health care Government NGO, environment Recreational services Market research Market research Printing Financial consulting Food Recreational services Food Government Recreational services Energy Media Manufacturing Participants no. 26 26* 25 21 28 20 34 28* 34 26 25 27 27 80* 28 23 27 Managerial 5 7 7 4 7 2 10 8 9 4 9 6 7 12 8 9 5 Non-Managerial 21 18 18 17 21 18 24 18 25 22 16 21 20 67 20 14 22
112
Table 2 (continued) Org No. 18 19 20 21 22 23 24 25 26 27 28 29 30 Industry/Type Education Aerospace Food Computer software Construction Contract research Publishing NGO, health Education Telecommunications Food Government Health care Participants no. 30 21 29 29 27 27* 28 16* 15 18 15 13 12* Managerial 6 6 5 6 5 6 5 8 6 9 4 3 7 Non-Managerial 24 15 24 23 22 20 23 7 9 9 11 10 4
Note. Total participants with asterisks have 1 or 2 missing values. Participants did not indicate their position in their respective organizations. Total missing = 7.
113
Table 3 Demographic Information of Participants (N = 785*) Variable Managers Frequency Male Female Sex Total Caucasian American African American Latin American Native American Pacific Islander Asian American Other Ethnicity Total Bachelor Master’s Doctorate Other Education Total Total* 79 114 193 (2 missing) 144 13 12 0 0 13 11 193 (2 missing) 90 57 11 32 190 (5 missing) 195 Percent (%) 40.5 58.5 99.0 73.8 6.7 6.2 0 0 6.7 5.6 99.0 46.2 29.2 5.6 16.4 97.4 24.8 Non-Managers Frequency 226 354 580 (3 missing) 387 65 29 3 4 56 32 576 (7 missing) 315 77 7 169 568 (15 missing) 583 Percent (%) 38.8 60.7 99.5 66.4 11.1 5.0 .5 .7 9.6 5.5 98.8 54.0 13.2 1.2 29.0 97.4 74.3
Note. 7 participants failed to report their position levels. Above demographic was based on 778 participants’ data, although the total N was 785.
114
As is shown in Table 2, 30 participating organizations were in a variety of industries and of different types, such as health care, recreational services, printing, financial consulting, energy, construction, education, and aerospace. Also, there were multiple managerial and non-managerial participants in each organization. Managerial employees numbered from 3 to 12, and non-managerial employees varied from 4 to 67. The ratio of managers to non-managers was 1:3. The average size of the organizations was 26.17. Such purposeful arrangements helped ensure enough variance on the organization variable (organizations’ perceptions of relationship characteristics) and person variable (employees’ perceptions of relationship characteristics) of congruence of perceived relationship characteristics. Table 2 and Table 3 indicated that the majority of respondents was female, Caucasian American, and holds a bachelor’s degree, regardless of their position levels. On average, respondents were 32 years old and had 4 years of tenure in their organizations. Most participants were younger than 45 (84.6%). A comparison of the descriptive statistics of managerial and non-managerial employees revealed that management employees on average were older and had a longer tenure at their organizations (Mage = 39, Mtenure = 7) than the non-management employees (Mage = 29, Mtenure = 3). Managers were more likely to be Caucasian American and have higher education levels than non-managers. Missing Values Analysis To minimize the impact of missing values, I first excluded cases that had more than 3 to 5 % unanswered questions. Still there were cases with missing data. To determine the extent and patterns of missing data, a rule of thumb is that missing data
115
under 10% for an individual observation can generally be ignored, unless the missing data occur in a nonrandom pattern (Hair et al., 2006). Also, the number of cases with no missing data ought to be sufficient for statistical tests to be performed if no substitution of missing values is planned (Hair et al.). Accordingly I used the SPSS 16.0 program to conduct an MVA on the final data set (N = 785) and found that the percent of missing values ranged from 0 to 3.1% (education being the variable with maximum missing data, i.e., 3.1%). Furthermore, I performed the Little’s Missing Completely at Random (MCAR) test and found that missing values occurred in a completely random fashion (?2 = 6482.34, df = 6308, p = .061 > 0.5). Even though the missing data occurred completely at random, they may cause serious problems. To begin with, most statistical procedures by default exclude cases with missing data from analysis, which leads to sample size reduction. Second, missing data can lead to misleading results and inferences (SPSS Inc., n.d.). The expectationmaximization (EM) method is the best way to deal with these problems. It generates estimates that are closest to the parameter values (see SPSS Inc., 2007 for demonstration example comparing the listwise, all-values estimate, and EM method). This method also introduces the least biases while providing the best representation of original distribution of values, compared with the other imputation methods, such as mean substitution, regression imputation, and all-available data (Hair et al., 2006). Therefore, I proceeded with the EM method of imputation.
116
Data Aggregation As discussed in Chapter II, managers’ data were to be aggregated to represent the organization side of organization-employee relationships. To justify aggregation, I calculated rwg, an estimator of within-group agreement. It ranges from 0 to 1, with a larger value indicating higher agreement. An rwg of 0.70 or above is considered acceptable (Klein & Kozlowski, 2000). For multi-item scales, James, Demaree, and Wolf (1984) proposed the following equation:
rwg ( J ) =
2 J [1- ( sxj / ? EU )] 2 2 J [1- ( sxj / ? EU )] + ( sxj / ? EU )] 2 2
2
where rwg ( J ) is the index of within-group agreement based on J parallel items on a
2 is the variable; sxj is the mean of the observed variance on the J parallel items; ? EU
2
variance on xj that would be expected if all cases on xj were due exclusively to random
2 measurement error. For scales using the 5-point Likert format such as in this study, ? EU
is equal to 2.0 (James, Demaree, & Wolf, 1993). The SPSS 16.0 program was used to calculate the rwg values. Because the aggregation involved only variables of congruence of perceived relationship characteristics, I present the rwg values for its five dimensions (trust, distrust, control mutuality, commitment, and satisfaction) in Table 4. The average rwg values were all above .80, and median rwg values were all above .85, sufficiently large to justify aggregation.
117
Table 4 Within-Group Agreement (rwg) Variable Trust Distrust Control Mutuality Commitment Satisfaction No. of items 6 4 4 8 4 Minimum rwg .60 .67 .41 .00* .50 Maximum rwg Average rwg 1.00 1.00 1.00 .98 1.00 .90 .86 .87 .85 .92 Median rwg .92 .85 .90 .89 .91
Note. A negative rwg value was recoded to 0 (James et al., 1984). In addition, I calculated ICC (1) and ICC (2). As previously stated, ICC (1) assesses non-independence when being calculated on dependent variables and measures reliability when being calculated on independent variables (Bliese, 2000). ICC (2) estimates reliability of group mean (Bartko, 1976). Both indices are recommended as criteria of aggregation (Bliese). Bartko suggested a formula for calculating ICC (1) from a one-way random-effects analysis of variance (ANOVA) model:
ICC (1) = MSB ? MSW MSB+ ( K ? 1) * MSW
Where MSB is the between-group mean square, MSW is the within-group mean square, and k is the group size. If group sizes differ, average group size for k is used. In this study, the average group size was 26.17. The range of ICC (1) calculated from the ANOVA model is -1 to +1. Negative values mean that the between-group variance is smaller than the within-group variance. According to Bliese, a large non-zero ICC (1) estimate suggests that group membership largely affects individual responses. If non-
118
independence is present, it should be accounted for to avoid biases caused in standard errors in further ordinary least squares analyses, such as ANOVA and regression. Similarly, from a one-way random-effects ANOVA model, ICC (2) can be calculated with the following formula (Bliese, 2000): ICC (2) = MSB ? MSW MSB
I obtained mean squares by conducting one-way ANOVA analyses with organization membership as the predictor and manager scores on the five dimensions of perceived relationship characteristics as the dependent variables. ICC (1) and ICC (2) values were calculated from the mean squares. The ICC (1) and ICC (2) estimates are presented in Table 5-I. Because I calculated ICC (1) on dependent variables, it assessed non-independence. It can be interpreted as the percent of variance in individuals accounted for by organization membership. The ICC (1) values in this study were low, ranging from .04 to .11. This indicated the influences of organization membership on managers’ responses, which is of small magnitude in this case. In contrast, the ICC (2) values were much greater. They showed that the group/organization means were relatively reliable, mostly above or near .70. In summary, based on the rwg, ICC (1), and ICC (2) values, I concluded that managerial employees’ data could be aggregated to represent their organizations’ perceptions of relationship characteristics.
119
Table 5-I ICC (1) and ICC (2) Results for Managers’ Data (n = 195) Variable Trust Distrust Control Mutuality Commitment Satisfaction No. of items 6 4 4 8 4 MSB 1.72 1.68 .44 1.50 1.41 MSW .42 .54 .22 .41 .36 ICC (1) .11 .07 .04 .09 .10 ICC (2) .76 .68 .49 .72 .74
In addition to the analysis on managers’ data, I calculated the ICC (1) values based on employee data to assess the extent to which employees’ data were nonindependent, which could result in biased standard errors in ordinary least squares analysis (G. Hancock, personal communication, July 16, 2009). The calculations were based on one-way ANOVA analyses with organization membership as the predictor and employee scores on all the dependent variables as the outcomes. All the ICC (1) estimates were low, ranging from .09 to .23. This indicated that the impact of organization membership on non-managerial employees’ responses was small, between .09 and 23. It was sufficient to conclude that the likelihood of having biased standard errors was small. The results are summarized in Table 5-II.
120
Table 5-II ICC (1) Results for Employees’ Data (n = 583) Variable Trust Distrust Control Mutuality Commitment Satisfaction Turnover Intention Contextual Performance No. of items 6 4 4 8 4 4 8 MSB 3.99 3.63 2.48 1.05 3.16 3.66 1.38 MSW .46 .56 .38 .30 .49 .81 .34 ICC (1) .23 .17 .18 .09 .17 .12 .11
Means, Standard Deviation, and Correlations I performed descriptive data analysis on both the managers’ and non-managers’ data. The managers’ data (n = 195) were aggregated in analyses involving the latent construct congruence of perceived relationship characteristics as the mediating variable. Descriptive statistics on the independent variables (symmetrical and asymmetrical relationship maintenance strategies) and the dependent variables (turnover intention and contextual performance) were based on non-managerial employees’ data (n = 583). The descriptive statistics yielded information on the two research questions and provided preliminary checks on content validity of the measures. I discuss these preliminary findings below. Table 6 and Table 7 summarize respectively the means, standard deviations, and correlations of independent and dependent variables. Table 8 and Table 9 provide the
121
means, standard deviations, and correlations of the dimensions of congruence of perceived relationship characteristics respectively based on the managerial respondents’ data (n = 195) as well as the organizational-level data (n = 30). Research Question 1: How do organizational members perceive relationship maintenance strategies used by their organizations? Descriptive statistics provided information on the first research question. With regard to independent variables (see Table 6), non-managerial employees perceived their organizations’ relationship maintenance strategies to be fairly symmetrical. The organizations reportedly had a relatively high level of disclosure, e.g., actively communicating with employees and encouraging employees to share thoughts and feelings about the organization (M = 3.50); showed considerable amount of assurances of legitimacy, e.g., taking employees’ concerns seriously (M = 3.62); made efforts to network with employees, e.g., involved in coalitions that benefit employees (M = 3.42); and provided advice to employees (M = 3.49). In addition, these employees appeared to believe that their organizations engaged less in asymmetrical relationship maintenance strategies, such as distributive negotiation (M = 2.70) and avoiding (M = 2.54). Yet they did feel that compromising was often used by organizations in conflict situations (M = 3.23). Taken together, non-managerial employees believed that their organizations employed more symmetrical relationship maintenance strategies than asymmetrical ones. Table 6 also demonstrated that all the independent variables were significantly correlated at the .01 level. The symmetrical relationship maintenance strategies were significantly and positively correlated with each other. On the other hand, two of the asymmetrical relationship maintenance strategies, i.e., distributive negotiation and
122
avoiding, were positively and significantly correlated with each other (r = .46, p < .01). Nevertheless, the third asymmetrical relationship maintenance strategy compromising correlated negatively and significantly with distributive negotiation (r = -.46, p < .01) and avoiding (r = -.32, p < .01). Combined with its higher mean value than distributive negotiation and avoiding, it seemed that non-managerial employees did not perceive compromising as much to be asymmetrical. Overall, the correlation between symmetrical and asymmetrical relationship maintenance strategies was -.20 (p < .01), preliminarily indicating that symmetrical and asymmetrical relationship maintenance strategies might be inter-related but distinct concepts. Table 6 Means, Standard Deviations, and Correlations of Independent Variables (n = 583) Variable 1. Disclosure 2. Assurances of 3.62 Legitimacy 3. Networking 4. Advice 5. Distributive Negotiation 6. Avoiding 7. Compromising **p < .01. 3.42 3.49 2.70 2.54 3.23 .85 .78 .87 .76 .71 .61** .51** .60** .53** 1.00 .53** 1.00 1.00 .46** -.46** 1.00 -.32** 1.00 .87 .75** 1.00 M 3.50 SD .77 1 1.00 2 3 4 5 6 7
-.35** -.40** -.26** -.16** -.38** -.41** -.28** -.26** .43** .52** .38** .33**
123
Research Question 2: How do non-management employees and managers perceive the organization-employee relationships? Congruence of perceived relationship characteristics from the non-managerial employees’ perspective. Descriptive statistics offered insights into the second research question. Table 7 showed that these non-managerial employees reported that they had good relationships with their organizations, judging from the relatively high means of trust (M = 3.67), control mutuality (M = 3.58), commitment (M = 3.11), and satisfaction (M = 3.77), as well as low mean level of distrust (M = 2.14). In addition, these five dimensions of the congruence of perceived relationship characteristics were all significantly correlated with each other at the .01 level. Coupled with the fact that distrust had significantly negative correlations with the other four dimensions, these significant correlations yielded preliminary support for the third proposition that relationship characteristics consist of these five dimensions. Table 7 Means, Standard Deviations, and Correlations of Dependent Variables (n = 583) Variable 1. Trust 2. Distrust 3. Control Mutuality 4. Commitment 5. Satisfaction 6. Turnover Intention 7. Contextual Performance M 3.67 2.14 3.58 3.11 3.77 2.98 3.79 SD .80 .83 .68 .57 .79 .97 .56 1 1.00 -.67 ** .72** .41** .80** -.51** .18** 1.00 -.66** -.28** -.66** .43** -.09* 1.00 .37** .74** 1.00 .45** 1.00 1.00 -.26** 1.00 2 3 4 5 6 7
-.47** -.48** -.55** .19** .27** .18**
124
*p < .05. **p < .01. Congruence of perceived relationship characteristics from the organizational perspective. I also examined the means, standard deviations, and correlations of congruence of perceived relationship characteristics from both the individual level (managerial employees’ responses) and the organizational level perspective (aggregated managers’ data). I averaged managerial employees’ responses (n = 195) in each organization to form the organizational-level data (n = 30). According to Table 8, managers reported high means of trust, control mutuality, commitment, and satisfaction, and low means of distrust (M = 2.08). Except distrust, the other four dimensions were all positively highly and significantly correlated with each other, at the .01 level. Distrust was significantly but negatively correlated with the other four dimensions. Similar patterns emerged from the organizations’ data in Table 9. The statistics suggested that organizations perceived good relationships with their employees. Table 8 Means, Standard Deviations, and Correlations of Congruence of Perceived Relationship Characteristics (n = 195) Variable 1. Trust 2. Distrust 3. Control Mutuality 4. Commitment 5. Satisfaction **p < .01. M 3.77 2.08 3.66 3.28 3.92 SD .78 .84 .76 .51 .72 1 1.00 -.79** .78** .43** .84** 1.00 -.78** -.35** -.78** 1.00 .34** .80** 1.00 .40** 1.00 2 3 4 5
125
Table 9 Organizational-Level Means, Standard Deviations, and Correlations of Congruence of Perceived Relationship Characteristics (n = 30) Variable 1. Trust 2. Distrust 3. Control Mutuality 4. Commitment 5. Satisfaction **p < .01. Turnover Intention and Contextual Performance from the Non-Managerial Employees’ Perspective Table 7 also presented descriptive information on turnover intention and contextual performance. Non-managerial employees reported moderate intention to leave their current organizations (M = 2.98) and considered themselves good performers on the job (M = 3.79), such as helping others’ work and contributing to organizations’ accomplishments. Also, these two dependent variables showed significant and negative correlation with each other (r = -.26, p < .01). Lastly, turnover intention had significantly negative correlations with four dimensions of congruence of perceived relationship characteristics (trust, control mutuality, commitment, and satisfaction), and significantly positive correlations with the dimension distrust. Similar patterns were found in contextual performance’s correlations with these five dimensions. M 3.78 2.07 3.66 3.30 3.92 SD .53 .52 .49 .27 .48 1 1.00 -.89** .92** .66** .96** 1.00 -.91** -.58** -.89** 1.00 .66** .95** 1.00 .68** 1.00 2 3 4 5
126
Reliability Tests and Factor Analyses After examining the data and descriptive statistics, I assessed reliability and validity of the measures with Cronbach’s alpha test, exploratory factor analysis (EFA), particularly principal component analysis (PCA), and confirmatory factor analysis (CFA). PCA was the extraction method used in EFA. Kaiser’s rule (eigenvalue > 1) was chosen for selection of number of components extracted. Varimax rotation was used to handle the issue of cross-loadings. EFA evaluates the dimensionality of measures and provides a preliminary set of measures. CFA assesses the degree to which data fit expected structure, yielding information on construct validity and construct reliability. I used the maximum likelihood estimation method in CFA. Reliability Coefficients for Initial Measures Table 10 summarizes alpha coefficients for the initial measures. Nunnally (1978) suggested .70 as an acceptable reliability coefficient threshold. Most of the initial measures had alpha values above .70. Commitment scale’s alpha coefficient was borderline (? = .68) for non-managerial data, and poor (? = .53) for managerial data. A possible explanation for this is that managers may be more homogenous than the nonmanagers on commitment. With everything else being equal, a more homogenous sample generally results in a lower reliability estimate than a more heterogeneous sample (Crocker & Algina, 1986, p. 144). The advice scale also has relatively low reliability values (? = .61 for managerial data, .62 for non-managerial data). This was probably because there were only two items for the variable.
127
Table 10 Cronbach’s Alpha of Initial Measures Variable Disclosure Assurances of Legitimacy Networking Advice Distributive Negotiation Avoiding Compromising Trust Distrust Control Mutuality Commitment Satisfaction Turnover Intention Contextual Performance Managers .75 .84 .87 .61 .90 .83 .83 .91 .87 .83 .53 .92 .81 .80 Non-Managers .78 .85 .89 .62 .88 .81 .81 .92 .85 .79 .68 .93 .83 .80 N of Items 4 3 3 2 3 4 3 6 4 4 8 4 4 8
EFA and Reliability Results For EFA, I retained items that loaded higher than .50 on a single factor (Hair et al.). Some items loaded highly (? .50) on more than one factor. To deal with the issue of cross-loadings, I used Varimax rotation so as to arrive at a simpler structure (Hair et al.). If an item still loaded on multiple factors after the rotation, it was deleted. The analyses
128
of the five dimensions of congruence of perceived relationship characteristics were based on all data (N = 785). The analyses on other variables were based on non-managerial employees’ responses (n = 583). Table 11, 12, and 13 present the factor loadings and alpha values of the revised scales. I discuss the results of different variables below. Relationship maintenance strategies. Proposition 2: Symmetrical relationship maintenance strategies are composed by disclosure or openness, assurances of legitimacy, networking, and advice. Asymmetrical relationship maintenance strategies include distributive negotiation, avoiding, and compromising. Research Question 1: How do organizational members perceive relationship maintenance strategies used by their organizations? The second proposition and first research question explored the kinds of relationship maintenance strategies used by organizations. I proposed four main symmetrical relationship maintenance strategies and three asymmetrical ones. I used 12 items to measure the four different kinds of symmetrical relationship maintenance strategies. Four items measured disclosure and three items assessed each of assurances of legitimacy and networking. Advice had two items. The EFA results showed that that disclosure, assurances of legitimacy, networking, and advice emerged as one factor—symmetrical relationship maintenance strategies. One item on networking (“the coalitions that the organization forms with other groups benefit the organization’s employees”) was deleted due to significant cross-loading. I then examined the reliability of symmetrical relationship maintenance strategies with all the remaining 11 items. It had excellent reliability (? = .90). The loadings of each item are presented in Table 11.
129
With regard to asymmetrical relationship maintenance strategies, 10 items were used to measure it. Three items were respectively used to assess distributive negotiation and compromising. Four items measured avoiding. The initial extraction in EFA resulted in multiple measures having cross-loadings. After the Varimax rotation, all the items loaded highly on a single factor. As a result, no item was deleted. The reliability coefficients of the three kinds of strategies were the same as the initial measures (see Table 11). All had highly reliable measures (> .80). Distributive negotiation, avoiding, and compromising stood out as three separate factors, instead of one factor of asymmetrical relationship maintenance strategies. It is possible that the latent construct asymmetrical relationship maintenance strategies is a second-order factor, and distributive negotiation, avoiding, and compromising are three first-order factors that are caused by asymmetrical relationship maintenance strategies. The EFA results also suggested that symmetrical and asymmetrical relationship maintenance strategies were related but conceptually distinct latent factors. These structures from EFA were further tested in CFA, and are discussed later. The factor loadings of each item obtained in EFA are shown in Table 11.
130
Table 11 Items, Factor Loadings, and Reliability for Relationship Maintenance Strategies (n = 583) Factor Construct Item Loading The organization’s annual report is a valuable source of information for employees about what it has done. (Disclosure 1) The organization shares enough information with employees about the organization’s governance. (Disclosure 2) The organization’s employee meetings are a valuable way for employees to communicate their opinions to the organization. (Disclosure 3) The organization encourages employees to disclose thoughts and feelings about the organization. (Disclosure 4) The organization makes a genuine effort to provide personal .80 responses to employees’ concerns. (Assurances 1) Symmetrical Relationship Maintenance Strategies The organization’s policy development process allows employees adequate opportunity to raise an issue and propose a solution. (Assurances 2) When employees raise concerns, the organization takes these concerns seriously. (Assurances 3) .76 .80 .76 .76 .71 .57 Alpha
131
Table 11 (continued) Construct Item The organization builds coalitions with groups that help employees’ career growth. (Networking 1) The activities that the organization is involved in with other groups are helpful to its employees. Symmetrical (Networking 3) Relationship The organization tells employees like me what Maintenance to do about problems on the job. (Advice 1) Strategies The organization shares opinions on things going on in my career. (Advice 2) Eigenvalues % of Variance Explained 50.96 .68 5.61 .90 .51 .71 .74 Factor Loading Alpha
132
Table 11 (continued) Construct Item Factor Loading When there is a conflict between employees and the management, the management will not give up, but keep on persuading until the employees concede. the management applies pressure so that employees may make concessions. Distributive Negotiation the management strongly adheres to its point of view until the employees agree. When there is a conflict between employees and the management, the management gives as little attention as possible to it. the management avoids addressing it. the management tries to avoid a confrontation. the management tries to avoid expressing differences of Avoiding opinion with employees involved. .77 .81 .70 .73 .84 .82 .88 .83 .90 Alpha
133
Table 11 (continued) Construct Item Factor Loading When there is a conflict between employees and the management, the management tries to realize a middle-of-the-road solution with employees. the management emphasizes a compromise solution. the management insists that both employees and the Compromising management give in a little. Eigenvalues % of Variance 73.47 Explained .85 .76 7.35 .81 .86 Alpha
Congruence of perceived relationship characteristics. Proposition 3: Congruence of employees’ and organizations’ (represented by managers’ perceptions) perceptions of their relationship characteristics best captures the relational dynamics. These relationship characteristics are made up of five components: trust, distrust, control mutuality, satisfaction, and commitment. Research Question 2: How do non-management employees and managers perceive the organization-employee relationships? The third proposition and second research question proposed that perceived relationship characteristics have five dimensions. I used six items for trust, four items for
134
distrust, four items for control mutuality, eight items for commitment, and four items for satisfaction. Because both employees and organizations were involved here, the EFA was based on all data (N = 785). A Varimax rotation was performed to eliminate multiple cross-loadings. No item was deleted. Factor loadings are provided in Table 12. The EFA results provided preliminary support for the proposition that trust, distrust, control mutuality and satisfaction are distinct latent factors. Commitment items emerged as three separate factors, i.e., the three sub-dimensions of affective, continuance, and normative commitment. Similar to the case of asymmetrical relationship maintenance strategies, it was likely that commitment was a higher-order factor that caused these three first-order factors. These structures from EFA were further tested in CFA and will be discussed later. I examined the reliability coefficients for these seven factors. The coefficients for trust (? = .92), distrust (? = .85), control mutuality (? = .79), and satisfaction (? = .93) were high. I also obtained separate reliability coefficients for the three sub-dimensions of commitment. Affective commitment had a reliability of .82, continuance commitment .61, and normative commitment .57. Except affective commitment, the other two subdimensions’ reliability were minimally acceptable. It was probably because each subdimension only has two or three items. Generally, the shorter a scale becomes, the smaller alpha is (Crocker & Algina, 1986). Take together, the measures of trust, distrust, control mutuality, commitment, and satisfaction were reliable. The reliability coefficients are also included in Table 12.
135
Table 12 Items, Factor Loadings, and Reliability for Relationship Characteristics (n = 785) Construct Item Factor Loading I believe that the organization treats employees fairly and justly. Whenever this organization makes an important decision, I know it will be concerned about its employees. The organization can be relied on to keep its promises to the employees. I am willing to let the organization make decisions for employees like me. I feel very confident about the organization’s capabilities. The organization has the ability to accomplish what it says Trust it will do. Eigenvalues % Variance 70.62 Explained .80 4.24 .92 .80 .87 .87 .86 .84 Alpha
136
Table 12 (continued) Construct Item Factor Loading I am skeptical about whether the organization will keep the employees’ interests in mind when it makes decisions. I feel that this organization will exploit employees’ vulnerability given the chance. I feel that this organization will engage in damaging and harmful behavior to employees to pursue its own interest. I feel that the way this organization is run is irresponsible Distrust and unreliable. Eigenvalues % of Variance Explained 71.25 .86 2.85 .85 .87 .88 .76 Alpha
137
Table 12 (continued) Construct Item Factor Loading This organization and employees like me are attentive to what each other say. This organization believes the opinions of employees like me are legitimate. In dealing with employees like me, this organization has a tendency to throw its weight around. (R) Control Mutuality Eigenvalues % of Variance Explained 63.59 This organization really listens to what employees like me have to say. .86 2.544 .79 .67 .87 .77 Alpha
138
Table 12 (continued) Construct Item Factor Loading I do not feel like “part of the family” at my organization. (R) This organization has a great deal of personal meaning for me. Affective Commitment There is a long-lasting bond between this organization and employees like me. I feel that I have too few options to consider leaving this organization. Continuance Commitment Too much in my career would be disrupted if I decided to leave the organization now. I think that people these days move from organization to organization too often. Jumping from organization to organization does not seem at all unethical to me. (R) One of the major reasons I continue to work for this Normative organization is that I believe that loyalty is important and Commitment therefore feel a sense of moral obligation to remain. Eigenvalues % of Variance Explained 67.84 .58 5.43 .57 .83 .70 .82 .61 .82 .91 .82 .90 .72 Alpha
139
Table 12 (continued) Construct Item Factor Loading I am happy with this organization. Both the organization and employees like me benefit from the relationship. Most employees like me are happy in their interactions with this organization. Generally speaking, I am pleased with the Satisfaction relationship this organization has established with me. Eigenvalues % of Variance Explained Note. (R) means that the items were reverse-coded. Turnover intention and contextual performance. As I discussed in Chapter II, organizations are supposed to incorporate the values and demands of their strategic constituencies into their long-term and short-term goals in order to be effective. Turnover intention and contextual performance are indicators of the extent to which organizations do so. I used four items to measure turnover intention, and eight items to evaluate contextual performance. Statistical analyses were based on non-managerial employees’ data (n = 583). EFA showed the appropriateness of these measures and dimensionality of these two dependent variables. 82.13 .92 3.29 .93 .88 .90 .93 Alpha
140
A Varimax rotation was performed on contextual performance items to eliminate significant cross-loadings. The last two items of contextual performance were deleted because they failed to load highly on a single factor. The items were: 1) “I tend to follow proper procedures and avoid unauthorized shortcuts;” 2) “I have put in extra hours to get work done on time.” All three scales were very reliable (? > .75). Table 13 presents the factor loadings as well as reliability coefficients of the revised scales. The EFA extraction indicated that turnover intention emerged as one factor, while contextual performance items yielded two factors. The first three items of contextual performance grouped as one factor, whereas the next three items loaded highly on another. The first three items assessed how much employees helped co-workers and volunteered to do extra work. The other three items dealt more with the contributions employees made toward their organizations. Organ (1997) respectively labeled them as altruism and conscientiousness. Altruism behaviors are geared towards individuals, whereas conscientiousness actions are focused on organizations. They both are components of contextual performance. Therefore, I named these two factors as altruism and conscientiousness. Similar to the three factors of asymmetrical relationship maintenance strategies and three factors of commitment, these two factors were likely governed by one underlying higher-order latent factor contextual performance. CFA results on this are to be discussed in the following section.
141
Table 13 Items, Factor Loadings, and Reliability for Turnover Intention and Contextual Performance (n = 583) Construct Item Factor Loading I would prefer another more ideal job than the one .85 I now work in. I have seriously thought about changing .78 Alpha
organizations since beginning to work here. I may leave this organization and work for another one in the next year. Turnover Intention Eigenvalues % of Variance Explained 66.31 I plan to stay in this organization to develop my career for at least three years. (R) .73 .83 .88
2.65
142
Table 13 (continued) Construct Item Factor Loading I often help other employees with their work when they have been absent. Altruism (Contextual Performance) I help other employees when their work load increases. I often volunteer to do things that are not formally required by the job. I have made suggestions to improve my organization. I have informed the management of potentially Conscientiousness (Contextual Performance) Eigenvalues % of Variance Explained Note. (R) means that the item was reverse-coded. Summary. In conclusion, exploratory statistical tests of the variables resulted in four factors of relationship maintenance strategies, three of which were likely first-order factors of asymmetrical relationship maintenance strategies. Seven factors were extracted for perceived relationship characteristics, three of which were likely first-order factors of commitment. The other four factors were trust, distrust, control mutuality, and 57.92 unproductive policies and practices. .86 I am willing to speak up when policy does not contribute to goal achievement of the organization. .85 4.63 .82 .76 .76 .78 .80 Alpha
.84
143
satisfaction. Turnover intention items emerged as one factor, whereas contextual performance grouped as two factors (altruism and conscientiousness). One item of networking and two items of contextual performance were deleted. Most items had high internal consistency. Continuance commitment and normative commitment had acceptable internal consistency. However, EFA is exploratory in nature. To confirm these structures, particularly to test whether some of the extracted factors were first-order factors (i.e., commitment factors, asymmetrical factors, and contextual performance factors), I conducted CFA. CFA Results I used the EQS 6.1 Program for CFA (Bentler, 2005). Following the two-step process, I first specified the initial measurement models. Then, using the Lagrange Multiplier test in EQS, I checked for model specification errors, such as error covariances and cross-loadings where a measured variable serves as an indicator of more than one latent factor. Respecifications arrived at the final measurement model (Hancock & Mueller, 2007). To evaluate data-model fit, I used the oft-cited Hu and Bentler (1999) joint-criteria approach. According to the joint-criteria, a model is considered tenable when it achieves NNFI ? .96, CFI ? .96, and SRMR ? .09 or RMSEA ? .06 and SRMR ? .09. Based on the EFA results, I compared different models for each latent construct. Regarding comparison among hierarchically nested models, I performed chi-square difference tests. Concerning comparison among models that were not nested within each other, I compared model AIC values. The models with smaller AIC values were considered a better representation of the data (Hancock & Mueller).
144
Relationship Maintenance Strategies Symmetrical relationship maintenance strategies. For symmetrical relationship maintenance strategies, I compared a one-factor model, a four-factor oblique model, and a second-order factor model. I did not include a four-factor orthogonal model because it is locally under-identified with advice and networking having only two measures respectively. The one-factor model was the most parsimonious model. Along with the second-order factor, it was hierarchically nested within the four-factor oblique model. Therefore I used nested model chi-square difference tests for nested model comparison. Table 14 summarizes the model comparison results and goodness-of-fit indices. The null model’s information as a baseline for computing goodness-of-fit indices is also presented. The statistics in Table 14 showed that the one-factor model had significantly datamodel fit than the null model: ??2 = 2729.122 with ?df = 11, p < .01. Next, I compared the one-factor model with the four-factor oblique model. The latter demonstrated significantly better data-model fit: ??2 = 213.509 with ?df = 6, p < .01. Also, its fit indices met the Hu and Bentler (1999) criteria: NNFI = .958 (approaching .96), CFI > .96, and SRMR = .031 (< .09). In addition, its RMSEA was very close to .06. Its model AIC was much smaller than that of the one-factor model. Thus I concluded that the four-factor oblique model was a more viable representation of the relations underlying the data.
145
Table 14 Symmetrical Relationship Maintenance Strategies CFA Results (n = 583) Model Null Model One-Factor Model 338.737 (Model A) Four-Factor A vs. B Oblique Model (Model B) Second-Order B vs. C Factor Model (Model C) 150.589 40 < .01 .950 .963 .037 .069 70.589 2 25.361 < .01 125.228 38 < .01 .958 .971 .031 .063 49.228 6 213.509 < .01 44 < .01 .878 .902 .053 .107 250.737
?2
3067.859
df 55
p
NNFI
CFI
SRMR
RMSEA
AIC 2957.859
?df
??2
p
146
Then, I compared the four-factor oblique model with the second-order factor model: ??2 = 25.361 with ?df = 2, p < .01. The nested model chi-square difference test showed that the four-factor oblique model was significantly better. Furthermore, as stated above, its fit indices met the Hu and Bentler criteria. Its model AIC was also much smaller than that of the second-order factor model. Based on the comparisons, I concluded that the four-factor oblique model was the most viable structure for the data with respect to symmetrical relationship maintenance strategies. The CFA results provided evidence to Proposition 2 that symmetrical relationship maintenance strategies were composed by disclosure or openness, assurances of legitimacy, networking, and advice. I provide the loadings for the measures on the four factors, correlations among the factors, and error variances in Figure 3. All the loadings, correlations, and path coefficients were significant at least at the .05 level. It was not surprising that these four factors were all significantly correlated, because they were all measuring the underlying concept of symmetrical relationship maintenance strategies.
147
V1 .533 Disclosure .687
.
.
846
E1
V2
.727
E2
758 V3
.653
E3
.
762 V4 .648 E4
.
926 V5
.
575
.
751 Assurance
.818 .809
.
E5
.
720 785
V6
.
587
E6
V7
.
.
620
E7
731 .835 Networking .805 .745 .558 Advice .805 V11
. .
.
700
V8
.
550
E8
V9
594
E9
V10
830
E10
.
594 E11
Figure 3. Final four-factor oblique model of symmetrical relationship maintenance strategies.
148
Asymmetrical relationship maintenance strategies. For asymmetrical relationship maintenance strategies, I compared a one-factor model, a second-order factor model, a three-factor orthogonal model, and a three-factor oblique model. The first three models were all hierarchically nested within the three-factor oblique model. I used both the nested model chi-square difference test and model AIC to compare these models. Table 15 summarizes the results. Table 15 Asymmetrical Relationship Maintenance Strategies CFA Results (n = 583) Model Null Model One-Factor Model Three-Factor Orthogonal Model Three-Factor Oblique Model Second-Order Factor Model 204.953 32 < .01 .919 .942 .061 .096 140.953 204.954 32 < .01 .919 .942 .061 .096 140.954 590.411 35 < .01 .762 .815 .254 .165 520.411
?2
3039.310 1147.60
df 45
p
NNFI
CFI
SRMR
RMSEA
AIC 2949.310
35 < .01
.522
.628
.131
.234
1077.603
The statistics in the above table showed that the one-factor model had significantly better data-model fit than the null model: ??2 = 1891.71 with ?df = 10, p < .01. Next, I compared the one-factor model with the three-factor oblique model. Because the two models were not nested within each other, it was incorrect to do the model chi-square difference test. I compared their model AIC. The three-factor
149
orthogonal model had a much smaller AIC value (520.411) than that of the one-factor model (1077.603), which suggested that the three-factor orthogonal model was better than the one-factor model. Then, I compared the three-factor orthogonal model with the three-factor oblique model. As the three-factor orthogonal model was nested within the three-factor oblique model, I conducted the chi-square difference test: ??2 = 385.457 with
?df = 3, p < .01. The test showed that the three-factor oblique model had significantly
better data-model fit than the three-factor orthogonal model. Also, the fit indices of the three-factor oblique model were close to the Hu and Bentler (1999) criteria: NNFI = .919 (approaching .96), CFI = .942 (borderline), and SRMR = .061 (< .09). Lastly, I compared the three-factor oblique model with the second-order factor. It was worth noting that these two models were equivalent models (see Hershberger, 2006 for more a detailed discussion on equivalent models). Given that equivalent models had the identical fit indices and chi-square values among others, one criterion for selecting the best model among equivalent models was the information complexity criterion (ICOMP). ICOMP = model misfit + complexity. The model with lower correlations among the parameter estimates would have smaller discrepancy between the implied and observed covariance matrix (i.e., small misfit) and smaller complexity, hence a better model for the data (Hershberger). I obtained correlations among the parameter estimates for both the three-factor oblique model and the second-order factor model. The second-order factor model had lower correlations among its parameter estimates than the three-factor oblique model, indicating that it was a better model for the data. Furthermore, the second-order factor model’s fit indices had acceptable fit based on the Hu and Bentler (1999) joint criteria: NNFI = .919 (approaching .96), CFI = .942 (borderline), and SRMR = .061
150
(< .09). As a result, I concluded that the second-order factor model was a more viable structure for the data. To sum up, the CFA results supported Proposition 2 that asymmetrical relationship maintenance strategies consisted of distributive negotiation strategies, avoiding, and compromising. I present the loadings for the measures on the three factors and error variances in Figure 4. All the loadings and path coefficients were significant at the .05 level. The three first-order factors were all highly significantly correlated with the underlying factor of asymmetrical relationship maintenance strategies, ranging from .663 to .819.
151
Asymmetrical Relationship Maintenance Strategies
.819 D1 .574 D2 .677 Distributive Negotiation .789 .873 .893
.736
-.663 D3 .748
Avoiding
Compromising
.859 V2 .512 E2
.875 .566
.475
.698
.952
.669
V1 .614 E1
V3 .487 E3 .451
V4
V5 .484
V6
V7
V8 .716 E8
V9 .305 E9
V10 .743 E10
.824 .880 E6 E7
E4
E5
Figure 4. Final second-order factor model of asymmetrical relationship maintenance strategies. Turnover Intention and Contextual Performance Turnover intention. As the EFA results yielded a one-factor structure for turnover intention, I tested the one-factor model in CFA. The results are summarized in Table 16. The one-factor model appeared to fit the data significantly better than the null model: ??2 = 880.653 with ?df = 4, p < .01. However, it did not have a very good data-model fit according to the Hu and Bentler (1999) criteria: NNFI = .679 (< .96 criterion), CFI = .893 (< .96 criterion), and SRMR
Employment is a relationship between two parties, usually based on a contract, one being the employer and the other being the employee.
ABSTRACT
Title of Document:
ORGANIZATION-EMPLOYEE RELATIONSHIPS MODEL: A TWO-SIDED STORY Hongmei Shen, Doctor of Philosophy, 2009
Directed By:
Professor Elizabeth L. Toth Department of Communication
The purpose of this study was to develop a theory of internal relationship management, and to propose a new way of measuring organization-public relationships by simultaneously examining the organizations’ as well as their employees’ perceptions of the quality of their relationships. It sought to contribute to theory-building on the process of relationship management from its maintenance through its quality to the consequences. An online survey was used to collect data. Usable questionnaires totaled 785 from 30 organizations. Data analytic methods included missing value analysis, exploratory factor analysis, confirmatory factor analysis, multivariate regression, polynomial regression, surface response tests, mediation tests, and reliability tests.
The proposed measures of symmetrical relationship maintenance strategies, asymmetrical relationship strategies, organization-employee relationship characteristics, turnover intention, and contextual performance were found to be valid and reliable. The major findings included: first, the more organizations used symmetrical relationship maintenance strategies with their employees, the more likely both employees and the organizations reported greater trust, control mutuality, commitment, satisfaction, and less distrust in the relationship; and vice versa for asymmetrical strategies. Second, employees would have higher turnover intention when both employees and their organizations perceived higher distrust and lower trust, control mutuality, commitment, and satisfaction. Also, when employees were more optimistic than their organizations about their relationships, employees were more likely to leave the organization. Third, employees’ contextual performance would rise as both these employees and their organizations reported greater level of commitment and satisfaction. However, employees’ level of contextual performance would drop when incongruence increased. Lastly, mediation tests showed that the effects of symmetrical and asymmetrical relationship maintenance strategies on turnover intention and contextual performance were partially mediated by congruence of perceived relationship characteristics, excluding the dimension of distrust regarding the effect of relationship maintenance strategies on contextual performance. This study contributed to public relations theory by 1) clarifying and refining the conceptualizations and operationalizations of relationship maintenance strategies, congruence of perceived relationship characteristics, and organizational effectiveness, 2)
proposing a new way to evaluate two sides of organization-public relationships, and 3) empirically testing a relationship-building model within organizations to develop a theory of internal relationship management.
ORGANIZATION-EMPLOYEE RELATIONSHIPS MODEL: A TWO-SIDED STORY
By Hongmei Shen
Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park, in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2009
Advisory Committee: Professor Elizabeth L. Toth, Chair Professor Meina Liu Professor Xiaoli Nan Professor Leah Waks Professor Cheri Ostroff
© Copyright by Hongmei Shen 2009
Dedication
I dedicate this dissertation to my family in China and my husband Shan Zhu. I am grateful for their love and support that made this work possible.
ii
Acknowledgements
Academic research is a road less travelled by. One needs courage and support to fend off self-doubts on the way to excellence. Only a novice on the road, I owe immense debt to many people who have encouraged and supported me. Special thanks go to my advisor, Dr. Elizabeth L. Toth for her continued support, encouragement, and guidance throughout the process. Thank you for holding me to higher standards. Thank you for guiding me into the academic world. Thanks also are due to my committee members: Dr. Meina Liu, Dr. Xiaoli Nan, Dr. Leah Waks, and Dr. Cheri Ostroff. Dr. Liu, Dr. Nan, and Dr. Waks shared their wisdom and knowledge whenever I asked. I am particularly grateful for Dr. Ostroff who has shared her expertise in quantitative methods and knowledge in organizational psychology and provided timely feedback on my data analysis even at midnight, without which I could not have completed this dissertation. I also thank Dr. Edward L. Fink who has been my mentor in quantitative methodology, teaching, and life. I thank Dr. Linda Aldoory whose academic rigor has pushed me forward. I am grateful for Dr. Gregory R. Hancock who patiently discussed with me possible ways to analyze my data. I thank my friend Dr. Jeong-Nam Kim who has spent so much time helping with my data analysis. Dr. Dale Hample, Dr. Brecken Chinn Swartz, and Mr. Richard J. Toth selflessly assisted my data collection. I am indebted to my friends at Maryland: my sisters Huang Jiang, Ai Zhang and Xiaodong Hou whose friendship made the journey more enjoyable, my buddies Xiaoying Xie and Bing Han who shared numerous insights on the phone in my research, Lucinda
iii
Austin and Lindsay Becker who offered tremendous help in my data collection, Shuo Yao, Sabine Chai, Nance McCown, Ying Wei, Yi Luo, Jingwei Yu, Teng Zhang, Lin Zhu, Lin Julie Zhu, Ling Na, Paula Weissman, Hyunhee Kim, Jarim Kim, Ahnlee Jang, Katie Place, Renata Schloss, Brian Smith, Jen Vardeman, Sanja Sipek, Jill Underhill, Erica Lamm, Sejal Patel, and David Payne. I thank my family in China: my parents Changchun Shen and Jiaxiu Du, my parents-in-law, Guangheng Zhu and Mingzhe Feng, my sister Qiurong Shen and my brothers Jianping Shen and Dong Xiang. Last, but not least, I owe the deepest gratitude to my husband Shan Zhu for his love, sacrifice, and support.
iv
Table of Contents
List of Tables ................................................................................................................... viii List of Figures .................................................................................................................... xi Chapter I.............................................................................................................................. 1 Overview of Study .............................................................................................................. 1 Purpose............................................................................................................................ 1 Theoretical Framework................................................................................................... 4 The Dialectical Approach ........................................................................................... 4 Extant Relationship Models........................................................................................ 5 Relationship Maintenance Strategies.......................................................................... 7 Congruence of Perceived Organization-Employee Relationship Characteristics....... 9 Organizational Effectiveness Outcomes ................................................................... 11 Method and Research Design ....................................................................................... 13 Significance of the Study .............................................................................................. 14 Chapter II .......................................................................................................................... 16 Conceptualization ............................................................................................................. 16 Defining Organization-Public Relationships ................................................................ 17 Why Organization-Employee Relationships?............................................................... 20 Research on Relationship Management........................................................................ 21 Application of the Relational Perspective................................................................. 22 Relationship Maintenance Strategies........................................................................ 23 Dimensions, Types, and Models of Organization-Public Relationships .................. 24 Conclusion ................................................................................................................ 31 Toward A Theory on Organization-Employee Relationships ...................................... 33 Systems Theory......................................................................................................... 33 Resource Dependence Theory .................................................................................. 35 Social Exchange Theory ........................................................................................... 36 The Dialectical Approach ......................................................................................... 37 Relationship Maintenance Strategies............................................................................ 42 Definition and a Terminological Issue...................................................................... 42 Relationship Maintenance Strategies in the Literature ............................................. 44 Symmetrical Maintenance Strategies........................................................................ 44 Congruence of Perceived Relationship Characteristics ................................................ 51 Trust .......................................................................................................................... 53 Distrust...................................................................................................................... 54 Commitment ............................................................................................................. 58 Satisfaction................................................................................................................ 60 Control Mutuality...................................................................................................... 61 Congruence of Perceived Relationship Characteristics ............................................ 62 Measurement Issues Related to Congruence of Perceived Relationship Characteristics........................................................................................................... 66
v
Organizational Effectiveness ........................................................................................ 74 Organizational Effectiveness: Definition and Approaches....................................... 74 Contextual Performance............................................................................................ 78 Turnover Intention .................................................................................................... 81 Model and Hypotheses: Linking the Variables............................................................. 82 Research on Relationship Maintenance Strategies as Antecedents .......................... 82 Summary ....................................................................................................................... 91 Chapter III......................................................................................................................... 95 Methodology ..................................................................................................................... 95 Survey ........................................................................................................................... 95 Why Survey?............................................................................................................. 95 Strengths and Weaknesses of Online Surveys .......................................................... 96 Sampling ....................................................................................................................... 98 The Appropriateness of Non-Probability Sampling ................................................. 98 Sample Size Estimation and Procedures................................................................... 99 Pilot Studies ................................................................................................................ 100 Pilot Study 1............................................................................................................ 100 Pilot Study 2............................................................................................................ 102 Measures ..................................................................................................................... 103 Antecedents: Relationship Maintenance Strategies ................................................ 103 Mediator: Congruence of Perceived Organization-Employee Relationship Characteristics......................................................................................................... 104 Dependent Variables............................................................................................... 105 Data Collection Procedures......................................................................................... 106 Data Analysis .............................................................................................................. 108 Data Aggregation .................................................................................................... 108 Statistical Procedures for Data Analyses ................................................................ 108 Chapter IV....................................................................................................................... 111 Results............................................................................................................................. 111 Description of Participants.......................................................................................... 111 Missing Values Analysis............................................................................................. 115 Data Aggregation ........................................................................................................ 117 Means, Standard Deviation, and Correlations ............................................................ 121 Research Question 1: How do organizational members perceive relationship maintenance strategies used by their organizations? .............................................. 122 Research Question 2: How do non-management employees and managers perceive the organization-employee relationships?............................................................... 124 Turnover Intention and Contextual Performance from the Non-Managerial Employees’ Perspective .......................................................................................... 126 Reliability Tests and Factor Analyses......................................................................... 127 Reliability Coefficients for Initial Measures........................................................... 127 EFA and Reliability Results.................................................................................... 128 CFA Results ............................................................................................................ 144 Relationship Maintenance Strategies...................................................................... 145 Turnover Intention and Contextual Performance ................................................... 152 Perceived Relationship Characteristics................................................................... 158
vi
Construct Reliability and Variance Explained........................................................ 169 Summary ................................................................................................................. 171 Hypotheses Testing..................................................................................................... 172 Multicollinearity Test.............................................................................................. 172 Multivariate Regression for Hypothesis 1 .............................................................. 174 Polynomial Regression Steps.................................................................................. 181 Overall Effect of Congruence of Perceived Relationship Characteristics on Dependent Variables............................................................................................... 183 Congruence of Perceived Relationship Characteristics and Turnover Intention .... 183 Congruence of Perceived Relationship Characteristics and Contextual Performance ................................................................................................................................. 186 Testing the Mediation Effect ...................................................................................... 208 Summary ..................................................................................................................... 214 Chapter V ........................................................................................................................ 216 Conclusions and Implications ......................................................................................... 216 Summary of Results.................................................................................................... 216 Data Aggregation .................................................................................................... 216 Descriptive Statistics Results for Research Questions............................................ 217 Reliability and Validity of Measuring Instrument .................................................. 218 Hypotheses Testing................................................................................................. 222 Mediation Effect Test Results................................................................................. 227 Theoretical Implications ............................................................................................. 228 Refined Conceptualization...................................................................................... 229 Relationship Building within Organizations........................................................... 233 Linking Public Relations to Organizational Effectiveness ..................................... 236 Relationship as a Two-Sided Story: Methodological Implications ............................ 237 Practical Implications.................................................................................................. 239 Limitations .................................................................................................................. 241 Future Research .......................................................................................................... 242 Appendix A. Questionnaire ............................................................................................ 246 Appendix B. Contact Email Template ............................................................................ 255 References....................................................................................................................... 256
vii
List of Tables
Table 1 Trust and Distrust Prototypical Relationship Conditions .................................... 56 Table 2 Descriptions of Participating Organizations ...................................................... 112 Table 3 Demographic Information of Participants (N = 785*)....................................... 114 Table 4 Within-Group Agreement (rwg) ........................................................................ 118 Table 5-I ICC (1) and ICC (2) Results for Managers’ Data (n = 195) ........................... 120 Table 5-II ICC (1) Results for Employees’ Data (n = 583) ............................................ 121 Table 6 Means, Standard Deviations, and Correlations of Independent Variables (n = 583) ............................................................................................................................. 123 Table 7 Means, Standard Deviations, and Correlations of Dependent Variables (n = 583) ............................................................................................................................. 124 Table 8 Means, Standard Deviations, and Correlations of Congruence of Perceived Relationship Characteristics (n = 195)................................................................ 125 Table 9 Organizational-Level Means, Standard Deviations, and Correlations of Congruence of Perceived Relationship Characteristics (n = 30) ........................ 126 Table 10 Cronbach’s Alpha of Initial Measures ............................................................. 128 Table 11 Items, Factor Loadings, and Reliability for Relationship Maintenance Strategies (n = 583).............................................................................................................. 131 Table 12 Items, Factor Loadings, and Reliability for Relationship Characteristics (n = 785) ............................................................................................................................. 136 Table 13 Items, Factor Loadings, and Reliability for Turnover Intention and Contextual Performance (n = 583) ........................................................................................ 142 Table 14 Symmetrical Relationship Maintenance Strategies CFA Results (n = 583).... 146
viii
Table 15 Asymmetrical Relationship Maintenance Strategies CFA Results (n = 583).. 149 Table 16 Turnover Intention CFA Results (n = 583)...................................................... 154 Table 17 Contextual Performance CFA Results (n = 583) ............................................. 156 Table 18 Trust Distrust CFA Results (N = 785) ............................................................. 159 Table 19 Perceived Relationship Characteristics CFA Results (N = 785) ..................... 162 Table 20 Correlations between Items with Added Error Covariance for Relationship Characteristics Measurement Model (N = 785).................................................. 165 Table 21 Comparison of Correlations between Relationship Characteristics Dimensions in the Initial Five-Factor Oblique Model vs. the Final Five-Factor Oblique Model with Error Covariances (N = 785) ...................................................................... 168 Table 22 Variance Explained and Coefficient H for All Measurement Models............. 169 Table 23 VIF of Independent Variables (n =583)........................................................... 174 Table 24 Multivariate Results of Symmetrical Relationship Maintenance Strategies Predicting Congruence of Perceived Organization-Employee Relationships (n =583)................................................................................................................... 178 Table 25 Multivariate Results of Asymmetrical Relationship Maintenance Strategies Predicting Congruence of Perceived Organization-Employee Relationships (n =583)................................................................................................................... 181 Table 26 Polynomial Regression Results for Congruence of Perceived Relationship Characteristics and Turnover Intention (n =583)................................................ 186 Table 27 Polynomial Regression Results for Congruence of Perceived Relationship Characteristics and Contextual Performance (n =583) ....................................... 188
ix
Table 28 Polynomial Regression Results for Congruence of Perceived Relationship Characteristics and Altruism (n =583)................................................................ 196 Table 29 Polynomial Regression Results for Congruence of Perceived Relationship Characteristics and Conscientiousness (n =583) ................................................ 203 Table 30 Mediation Effect Results for Relationship Maintenance Strategies and Turnover Intention (n =583) ............................................................................................... 211 Table 31 Mediation Effect Results for Relationship Maintenance Strategies and Contextual Performance (n = 583)...................................................................... 214
x
List of Figures
Figure 1. The structural model of internal relationship management theory...................... 4 Figure 2. The coorientation model adapted from Broom and Dozier (1990) ................... 68 Figure 3. Final four-factor oblique model of symmetrical relationship maintenance strategies. ............................................................................................................ 148 Figure 4. Final second-order factor model of asymmetrical relationship maintenance strategies. ............................................................................................................ 152 Figure 5. Final one-factor model of turnover intention. ................................................. 154 Figure 6. Final two-factor oblique model of contextual performance. ........................... 158 Figure 7. Final five-factor oblique model of perceived relationship characteristics. ..... 167 Figure 8. Congruence of perceived trust on contextual performance............................. 190 Figure 9. Congruence of perceived control mutuality on contextual performance. ....... 191 Figure 10. Congruence of perceived commitment on contextual performance.............. 192 Figure 11. Congruence of perceived satisfaction on contextual performance................ 193 Figure 12. Congruence of perceived trust on altruism.................................................... 198 Figure 13. Congruence of perceived commitment on altruism....................................... 200 Figure 14. Congruence of perceived satisfaction on altruism......................................... 201 Figure 15. Congruence of perceived trust on conscientiousness. ................................... 205 Figure 16. Congruence of perceived control mutuality on conscientiousness................ 206 Figure 17. Congruence of perceived commitment on conscientiousness. ...................... 207 Figure 18. Congruence of perceived satisfaction on conscientiousness. ........................ 207 Figure 19. Final model of internal relationship management. ........................................ 228 Figure 20. Final model of internal relationship management. ........................................ 228
xi
Chapter I Overview of Study Purpose Since the 1970s, organizations have demanded that public relations practitioners demonstrate the value of communication programs, as public relations is an intangible management function (L. Grunig, J. Grunig, & Dozier, 2002). Practitioners have to justify why organizations should spend millions of dollars in their work (J. Grunig & Y. Huang, 2000; Lindenmann, 1997, 2003). Public relations practitioners often face downsizing when organizations hit financial crises because of the difficulty to demonstrate public relations’ value (L. Grunig et al.). As the former President and CEO of the Institute for Public Relations Jack Felton said, “For years we have been told that we can never expect to get proper credit for what we do in public relations until we can find an effective way to measure our effectiveness” (Lindenmann, 1997, p. 1). To address this question, the strategic management approach posited that the public relations function adds values to organizations by scanning the environment, identifying stakeholders, and segmenting different publics from stakeholder categories (e.g., Dozier, L. Grunig, & J. Grunig, 1995; J. Grunig, 2006; L. Grunig et al., 2002). Public relations also adds value by using two-way symmetrical communication to cultivate long-term quality relationships with strategic publics (e.g., J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999; Y. Huang, 1997; Hung, 2001, 2002, 2004, 2007; H.-S. Kim, 2005; Ledingham, 2003; Ledingham & Bruning, 1998, 2000; Ni, 2006; Yang, 2005).
1
These strategic management scholars particularly contended that the value of public relations lies in relationships. Since Ferguson’s (1984) call for a relational perspective of public relations, relationship management has emerged as a new paradigm in public relations research (Ledingham, 2003). Research on organization-public relationships has identified different types of relationships (e.g., Bruning & Ledingham, 1999; Hung, 2004, 2005, 2007), models of organization-public relationships (e.g., Broom et al., 2000; J. Grunig & Y. Huang, 2000; Toth, 2000), dimensions of relational outcomes (e.g., Bruning & Ledingham, 1999, 2000; Hon & J. Grunig, 1999; Y. Huang, 2001; Y. Kim, 2001), strategies to maintain relationships (e.g., J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999; Hung, 2007; Ki & Hon, 2007a), and the effect of relationships on reputation (J. Grunig & Hung, 2002), publics’ attitudes, behavioral intention (Ki & Hon, 2007b), and actual behavior (Bruning, 2002). It has been applied to community relations, media relations, employee relations, public affairs, issues management, and crisis communication (e.g., Bridges & Nelson, 2000; Coombs, 2000; Ledingham & Bruning, 1998; L. Wilson, 2000). One problem with this body of literature is that most organization-public relationship research views relationship quality as the ultimate end product and purpose of public relations programs. With its basic premises tied to organizational effectiveness, L. Grunig et al.’s (2002) research has already demonstrated how excellent public relations can contribute to organizational effectiveness. Still to be examined is how public relations relationship maintenance programs that help build relationships with publics can contribute to organizational effectiveness.
2
Second, despite the growing interests in organization-public relationships, few scholars have focused on the relationships an organization has with its employees, a most important public to organizational survival (Cutlip, Center, & Broom, 1985; H.-S. Kim, 2003, 2005, 2007; Ni, 2006; Wright, 1995). A primary purpose of this study therefore was to develop a theory of internal relationship management. To that end, I tested a model consisting of relational antecedents (symmetrical and asymmetrical relationship maintenance strategies), congruence of perceived organization-employee relationship characteristics1 and organizational effectiveness outcomes. In so doing, I sought to enhance our understanding of relationship building within organizations, from its maintenance through its characteristics to its consequences. The third problem with current research on organization-public relationships is its asymmetrical nature. Although relationships involve more than one party, very limited research (Christensen, 2005; Kelly, Thompson, & Waters, 2006; Seltzer, 2005; Seltzer & Mitrook, 2009; Shin & Cameron, 2005; Waters, 2007) has been devoted to the measurement of both sides of a relationship. A second purpose of this dissertation thus is to propose a new way of measuring organization-public relationships by looking at both the organization’s and employees’ perceptions of the quality of their relationships. To solve these problems, this study tested a model of internal relationship management (Figure 1) by evaluating publics’ as well as organizations’ perceptions of
Many studies, including Hon and J. Grunig (1999), called dimensions of relationship characteristics as “relational outcomes” or “outcomes of relationships,” which is confusing because these are attributes of the quality of a relationship, rather than results of the relationship. Simply put, they are not effects and the relationship is not the cause. Also, some (e.g., Ki & Hon, 2007b) used “indicators” and “outcomes” interchangeably. To avoid confusion, this study therefore will not use the terms of “relational outcomes” or “outcomes of relationships” and the like.
1
3
relationship quality. I discuss the underlying theory and the conceptualization of each component of the model in the following section.
?
H2 H1a
?
Symmetrical Relationship Maintenance Strategies
Congruence of Perceived Relationship Characteristics
Turnover Intention ?
H3
H1b
Asymmetrical Relationship Maintenance Strategies
Contextual Performance
Figure 1. The structural model of internal relationship management theory. Theoretical Framework The Dialectical Approach Relying on Y. Huang’s (1998) definition, I defined organization-employee relationships as the extent to which an organization and its employee publics trust one another, agree each party has legitimate power to influence, report satisfaction with one another, and commit to one another. The main theoretical framework used in this study was the dialectical approach that consists of four key assumptions, namely, contradiction, change, praxis, and totality (Baxter & Montgomery, 1996). Contradiction is “the dynamic interplay between unified oppositions”, in which oppositions are “tendencies or features of a phenomenon” (Baxter & Montgomery, p. 8). Change arises from contradiction. It is “the interplay of stability and flux” (Baxter & Montgomery, p. 10). Praxis means that individuals are both
4
proactive in making decisions and choices and reactive to their partners’ decisions and choices. Totality refers to the notion that “social phenomena are defined by the relations among their characteristics, not by the characteristics themselves” (Montgomery, 1993, p. 206). Extending this approach to organization-public relationships, I viewed organizations and their publics (e.g., employees) as two “unified oppositions” that interact with and influence each other’s decisions and choices. Their different agendas inevitably lead to the dynamic interplay between them. I further developed a general proposition of relationship management—the dynamic and constant tension and change in organization-public relationships around negotiated interests and goals lead to benefits of varying degrees for the organization and its publics and better understanding of each other. Extant Relationship Models Based on the above proposition and models in the literature (Broom et al., 1997, 2000; Bruning & Ledingham, 1999; J. Grunig & Y. Huang, 2000; Hung, 2007; Toth, 2000), I proposed to test an organization-employee relationship model. Particularly, the two three-stage models proposed by Broom et al. (1997, 2000), and J. Grunig and Y. Huang (2000) inspired the initial conceptualization of this study. Adopting the systems theory, social exchange and resource dependence theory, Broom et al.’s (1997, 2000) relationship model includes antecedents, concept and consequences. Antecedents are the “perceptions, motives, needs, behaviors, and so forth” (e.g., perceptions of uncertain environment, needs for resources, and collective perceptions and expectations) that are “causes in the formation of relationships” (Broom
5
et al., 1997, p. 94). In light of the notion of open systems, Broom et al. (2000) noted that these antecedents are “the sources of change, pressure, or tension on the system derived from the environment” (p. 16), though they did not clarify the meaning of “environment.” Concept represents the attributes of the exchanges or transfers of information, energy or resources in a relationship. Consequences are outputs of such exchanges or transfers (e.g., goal achievement, dependency/loss of autonomy, and routine and institutionalized behavior). Although it is subject to the limitations of the systems theory, social exchange theory, and resource dependence theory, Broom et al.’s model suggests that organizations and publics build relationships to achieve certain outcomes, such as goal achievement, akin to the reasoning of my study. J. Grunig and Y. Huang’s (2000) model comprises situational antecedents, relationship maintenance strategies, and relationship outcomes. The situational antecedents indicate when and how relationships form. Pointing out the inadequacy of the resource dependence theory and social exchange theory in accounting for all the change pressures from the environment, J. Grunig and Y. Huang argued that relationship antecedents are situational. The term situational emphasizes that “publics come and go and change as situations change” (J. Grunig & Y. Huang, p. 35). Once a relationship is forged between an organization and a public, or multiple organizations and multiple publics, organizations develop communication strategies to maintain the relationship. Relationship maintenance strategies include symmetrical as well as asymmetrical strategies. The symmetrical ones are geared towards mutual or joint gains whereas the asymmetrical ones are focused on maximization of self-gains. Examples of symmetrical maintenance strategies are shared tasks and integrative
6
negotiation. Instances of asymmetrical ones are contending and distributive negotiation. Relationship outcomes resultant from the maintenance strategies are control mutuality, commitment, satisfaction, trust, and goal attainment (J. Grunig & Y. Huang, 2000). This model overcomes the inadequacy of the theoretical framework of Broom et al’s (1997, 2000) in explaining all the tensions and pressures from the environment. Nevertheless, it remains unclear what the “environment” means and constitutes in J. Grunig and Y. Huang’s model (2000). In contrast with Broom et al., J. Grunig and Y. Huang regard relationship quality as the end product, which is analogous to assuming that a clothing factory buys silk and thread only to make clothes instead of further selling such clothes for profit. Integrating these two models and using the dialectical approach to internal relationship management, I proposed a three-stage model of organization-employee relationships (see Figure 1), including two antecedents (symmetrical and asymmetrical relationship maintenance strategies), the congruence of perceived relationship characteristics (trust, distrust, satisfaction, control mutuality, and commitment) as the mediator, and organizational effectiveness outcomes (contextual performance and turnover intention). Relationship Maintenance Strategies Consistent with the dialectical approach, I defined relationship maintenance strategies as planned and intentional organizational behavioral attempts or efforts “to keep a relationship in a satisfactory condition” and “to keep a relationship in repair” (Dindia, 2003, pp. 3-4). Researchers debated on whether it is necessary to differentiate these strategies from routine behaviors. I concurred that intent does distinguish strategic
7
from routine relationship maintenance behaviors, but argued that such distinctions are practically difficult and unnecessary. Relationship maintenance strategies encompass symmetrical and asymmetrical ones (Hon & J. Grunig, 1999; J. Grunig & Y. Huang, 2000). Symmetrical maintenance strategies are disclosure or openness, assurances of legitimacy, networking, and advice. The asymmetrical ones include distributive negotiation, avoiding, and compromising/accommodating. They are respectively connected with the two-way symmetrical model and two-way asymmetrical model in public relations. Only a limited number of studies have concentrated on relationship maintenance strategies in public relations research (Ki & Shin, 2005). These studies and research in other areas of study such as interpersonal communication and conflict management have linked relationship maintenance strategies to dimensions of perceived organizationemployee relationship characteristics (trust, distrust, satisfaction, control mutuality, and commitment). Furthermore, a study by Weigel and Ballard-Reisch (1999) of 129 married couples provided evidence that individuals’ use of certain symmetrical maintenance strategies (positivity, openness, assurances of legitimacy, networking, and shared tasks) were positively associated with the couples’ joint marital relationship quality (satisfaction, commitment, and love), where the notion of joint marital relationship quality is similar to my conceptualization of congruence of perceived relationship characteristics. Therefore, by extending these pieces of work, I hypothesized relationship maintenance strategies as antecedents of congruence of perceived organization-employee relationship characteristics in my study; specifically anticipating a positive relationship between symmetrical relationship maintenance strategies and congruence of perceived
8
organization-employee relationship characteristics and a negative relationship between asymmetrical relationship maintenance strategies and congruence of perceived organization-employee relationship characteristics. Congruence of Perceived Organization-Employee Relationship Characteristics Existing public relations literature acknowledges the multidimensionality of the concept of organization-public relationships, proposing four such dimensions—trust, satisfaction, control mutuality, and commitment (e.g., J. Grunig & Hung, 2002; Hon & J. Grunig, 1999; Y. Huang, 2001; Ki & Hon, 2007b, 2007c; Ki & Shin, 2005; H.-S. Kim, 2005, 2007; Y. Kim, 2001; Yang & J. Grunig, 2005; Yang, 2007). Various problems exist with the oft-cited conceptualization of these dimensions by Hon and J. Grunig (1999): 1) Is trust a perception or a behavioral intention? 2) Is trust the opposite of distrust? 3) Is normative commitment an additional component of commitment? 4) Is satisfaction towards the job or the organization or both? 5) Are these four dimensions the only dimensions to measure relationship characteristics? 6) Why has public relations research used only publics’ perceptions to represent organization-public relationship characteristics? I addressed these problems by clarifying the conceptualization of the aforementioned dimensions, elucidating the differences and similarities between trust and distrust, and proposing the notion of congruence of perceived relationship characteristics from the dialectical perspective. For example, I added the new dimension of distrust from research on business management (e.g., Lewicki, McAllister, & Bies, 1998) as an additional component of relationship characteristics because it has been shown as a concept distinct from trust. Trust pertains to integrity, dependability, and competence.
9
Commitment encompasses the components of continuance and affective commitment. I also suggested normative commitment as another component of commitment (see Gruen, Summers, & Acito, 2000). Lastly, I laid out a few key problems with the existing coorientation method in public relations research to assess both the organization’s and its publics’ perceptions of their relationships, and provided an alternative approach. The concept of congruence has been examined for at least three decades by organizational psychologists (e.g., Chatman, 1989; Ostroff & Schulte, 2007; Schneider, 1983, 1987; Vancouver & Schmitt, 1991). Extending their conceptualization of congruence as compatibility between individual and organizational characteristics (e.g., values, beliefs, attitudes, and perceptions), I argued from the dialectical perspective that it is imperative to understand the interaction between organizations and their publics (e.g., employees) in order to best capture the essence of organization-public relationships. Congruence of perceived organization-public relationship characteristics is not exactly the same as consensus or similarity, but reflects the dynamic interaction and process of interdependence between unified oppositions, namely, forces such as an organization and its publics. It changes and adapts over time. However, at a given point in time, compatibility and/or consensus may emerge. Based on the person-organization fit literature in organizational psychology, I operationalized congruence as five variables: the perceptions of employees (one opposition) as the person variable, the perceptions of the organization (another opposition) as the organization variable, the squared person variable, the squared organization variable, and the product of the person and organization variable (Edwards, 1994, 1995; Ostroff, Atwater, & Feinberg, 2004). Considering that this was the first study that adopted such an approach to studying
10
organization-public relationship characteristics, I emphasized its exploratory nature in examining the ways in which non-management employees and managers perceive their organization-employee relationships although I used research from other disciplines to support my reasoning. Organizational Effectiveness Outcomes One pivotal development in the relationship paradigm was the identification of the linkage between organization-public relationship quality and public attitudes, perceptions, and behavior (Ledingham, 2003). Yet scant attention has been devoted to the behavioral consequences of organization-public relationship quality (Ki & Hon, 2007b). To add to this line of research, I focused on organizational effectiveness outcomes. Organizational effectiveness lies at the heart of most organizational research (Robbins, 1990). L. Grunig et al.’s (2002) landmark study of public relations excellence also began with the quest of organizational effectiveness. They suggested that the public relations function contributes to organizational effectiveness by building quality, longterm relationships with strategic publics. One such strategic public is the employee public. Thus, building upon this body of research on public relations excellence, I proposed that the public relations function may demonstrate its contribution to organizational effectiveness through building quality relationships with employees. By considering the goal attainment approach, the systems approach, the strategic constituencies approach, and the competing values approach to studying organizational effectiveness (Robbins, 1990), I suggested an integration of these approaches and thus presenting my definition of organizational effectiveness—the degree to which an
11
organization incorporates the values and demands of its strategic constituencies into its long-term (ends) and short-term (means) goals that it achieves. Employees are one of an organization’s strategic constituencies. As Campbell (1977) demonstrated, employees’ turnover and productivity are two key criteria of organizational effectiveness. Furthermore, these two criteria have been widely studied in research on human behavior in organizations (e.g., Borman & Motowidlo, 1993, 1997; Goodman & Svyantek, 1999; J. Hogan, Rybicki, Motowidlo, & Borman, 1998; Motowidlo, 2000; Motowidlo & Van Scotter, 1994; Van Scotter, 2000). Lastly, relationship management researchers such as Ledingham (2003) have uncovered evidence linking organization-public relationship quality to publics’ loyalty, similar to the intention to stay with an organization by employees, namely, turnover intention. Therefore, I chose turnover intention and contextual performance as two behavioral outcomes of organizational effectiveness in this study. The term contextual performance was developed to tap non-task-related performance by employees, which is often not part of their formal job responsibilities (Borman & Motowidlo, 1993). It overlaps with organizational citizenship behavior (Bateman & Organ, 1983). Examples of contextual performance include assistance to coworkers and initiatives to support organizational objectives (Borman & Motowidlo, 1997). On the other hand, turnover intention has been conceived of as deliberate willingness to leave an organization (Tett & Meyer, 1993). I operationalized it as intention to leave within a particular time interval. Based on and extending research from organizational behavior and business management, I anticipated that the level of congruence of perceived organization-
12
employee relationship characteristics is linked to the above two organizational effectiveness outcomes (e.g., Allen & Meyer, 1990; Z. Chen & Francesco, 2000; Goodman & Svyantek, 1999; Morgan & S. Hunt, 1994; Saks & Ashforth, 2002; Schneider, 1983, 1987; Vancouver & Schmitt, 1991; Van Vianen, 2000). Method and Research Design The online survey method was used in this study due to its efficiency, convenience, and ensured confidentiality and anonymity. A combination of random sampling and non-probability sampling was employed in the selection of participating organizations. As theoretical generalizability other than statistical generalizability was the focus, non-probability sampling was appropriate (Shapiro, 2002). Undergraduate students in the Department of Communication were paid to recruit organizations during the non-probability sampling stage. The same template email was used for organization recruitment, containing information on the purpose of the study, the investigators, expected participation time, and potential harm and benefits, among other content that has been approved by the Human Subjects Review Board at the University of Maryland. Each organization was provided a separate link to the same questionnaire. Template reminder emails were sent each week following the indication of willingness to participate, until at least 25 participants were recruited from a particular organization. Confidentiality and anonymity were ensured, with the exception that the organization names were known to the researcher and the respective student recruiter for an organization. Forty-eight organizations participated in the study, but 18 of them were excluded from data analyses because of the small number of responses (< 10) from each
13
organization. The organizations were in different industries and of diverse sizes to ensure sufficient variance between organizations. A total of 1293 participants agreed to participate. The final sample size was 785, consisting of 195 managers, 583 nonmanagerial employees, and seven with no indication of position level. Willing participants (n = 263) were entered in a raffle to reward their participation. Significance of the Study This study contributed to the continued search for the value of public relations. As many have contended, the value of public relations lies in relationship management. But what comes into relationships? How do employees form relationships with their organizations? What do organizations do to maintain relationships with their employees? What do we build relationships for? How can relationship management contribute to the overall organization effectiveness? This study provided answers to these questions, important to research and practice alike. Second, despite the burgeoning body of literature on relationship management, it remains a fragmented area fraught with research that often does not explicitly state its conceptualization of relationship quality. Conceptual clarity is necessary for theory development (Dindia, 2003). Taking the dialectical approach and relying on congruence research in organizational psychology, this study clarified the conceptualization of organization-public relationships, its operationalization and measurement, and its antecedent and outcomes, thus potentially advancing relationship research to a higher theoretical level. Third, this study refined existing measurement of organization-public relationships by distinguishing trust from trustworthiness and distrust, adding new
14
dimensions to the current widely used measuring instrument by Hon and J. Grunig (1999), problematizing the coorientation method, and proposing a new way to evaluate two sides of the relationship. In so doing, it moved relationship management research to a new operational level. Fourth, this study provided insights into the process of how organizations build relationships with one of their most important publics—employees—by developing a theory of internal relationship management. The study demonstrated the ways in which organizations’ relationship maintenance strategies influence the congruence of organizations’ and their employees’ perceptions of relationship characteristics, which eventually impacts organizational effectiveness outcomes—employees’ contextual performance and turnover intention. Few studies on relationship management have presented a comprehensive view of this process.
15
Chapter II Conceptualization J. Grunig and T. Hunt (1984) defined public relations as the “management of communication between an organization and its publics” (p. 6), which represents the strategic management paradigm of public relations. As part of this paradigm, the excellence study sought to determine how public relations makes an organization more effective, how much this contribution is worth economically, and the characteristics of this public relations function (L. Grunig et al., 2002). This study that spanned 17 years and surveyed three countries found that public relations can make an organization more effective “when it identifies the strategic publics that develop because of the consequences that organizations and publics have on each other” and “when it uses symmetrical communication programs to develop and maintain quality long-term relationships with these strategic publics” (L. Grunig et al., 2002, p. 548). In other words, the value of public relations to organizations lies in the relationships it builds with strategic publics. Built upon the excellence study and extant literature on relationship management, this study posited that the public relations function uses symmetrical relationship maintenance strategies to develop quality relationships between organizations and their employees, which consequently leads to organizational effectiveness, as assessed by employees’ turnover intention and contextual performance. In this section, I first provide conceptualization of organization-public relationships, an explanation for my focus on organization-employee relationships, and a review and critique of current research on relationship management. Next, I offer conceptualization of focal constructs: relationship maintenance strategies (symmetrical
16
and asymmetrical), perceived relationship characteristics, congruence of perceived relationship characteristics, organizational effectiveness, contextual performance, and turnover intention; and reviews of relevant literature. I conclude with review of literatures linking the focal constructs and introduction of the conceptual model. Defining Organization-Public Relationships Public relations as a scholarly discipline has struggled for an identity that distinguishes it from marketing, advertising, and journalism (Bruning & Ledingham, 1999; Cheney & Christensen, 2001; L. Grunig et al., 2002). Since more than two decades ago when Ferguson (1984) first advocated a relational perspective of public relations, a multitude of research has examined organization-public relationships. Nearly two decades later, Ledingham (2003) observed “pivotal developments which spurred emergence of the relational perspective as a framework for public relations study, teaching, and practice” (p. 286). He reasoned that relationship management has become a general theory that “unifies … [the public relations] discipline” (Ledingham, 2003, p. 192). He further concluded, “The relational theory of public relations also serves as the foundation for a relationship-grounded research paradigm” (Ledingham, 2003, p. 193). Under the umbrella of this relational theory are a wide range of relational dimensions, types, and models identified by researchers. Despite the celebration of the emerging research on relationship management, it remains a fragmented paradigm according to Ledingham and Bruning (2000) and Ledingham (2003). To begin with there still is no agreed-upon definition of “relationship” (Broom et al., 1997; Ki & Shin, 2005). Ledingham and Bruning (1998a) referred to an organization-public relationship as a state that “exists between an
17
organization and its key publics, in which the actions of either can impact the economic, social, cultural or political well being of the other” (p. 62). Broom et al. (2000) conceptualized it as characteristics of relationship-related activities—“the patterns of interaction, transaction, exchange, and linkage between an organization and its publics” (p. 18). Thomlison (2000) stated that a relationship is “a set of expectations two parties have for each other’s behavior based on their interaction patterns” (p. 178). Y. Huang (1998) defined organization-public relationships from the perspective of its characteristics, referring to relationships as “the degree that the organization and its publics trust one another, agree [each party] has rightful power to influence, experience satisfaction with each other, and commit oneself to one another” (p. 12). She regarded organization-public relationships as relationships between a corporate person and another corporate person or between a corporate person and a group of individuals (Y. Huang, 1997a). Hung (2005) developed her conceptualization of relationships from the perspective of systems theory, resource dependence theory and exchange theory. She stated, “[organization-public relationships] arise when organizations and their strategic publics are interdependent, and this interdependence results in consequences to each other that organizations need to manage constantly” (Hung, p. 396). Most of these definitions have one thing in common—they all defined a concept by laying out its characteristics, exemplified by sets of expectations, or specific properties such as trust, power, satisfaction, commitment (Y. Huang, 2001), or interdependence (Hung, 2005). In a similar vein, Wood’s (1995) exhaustive review of more than 700 articles and books on relational communication led her to conclude with four dimensions of characteristics of relationships—investment, commitment, trust, and comfort with
18
relational dialectics. Another interpersonal communication researcher Wilmot (1980) also used two dimensions of characteristics to define the nature of dyadic relationships: love/hate, and dominant/submissive. Broom et al.’s (2000) extensive review further revealed other dimensions of characteristics of relationships that interpersonal communication scholars commonly use. In addition, Ledingham and Bruning (1998a) cited research from marketing and social psychology that proposed various dimensions of relationship characteristics. Relying on these bodies of literature from different disciplines, this study thus proposed to conceptualize organization-public relationships from the perspective of its characteristics, and used Y. Huang’s (1998) definition of organization-public relationships as a tentative one: “the degree that the organization and its publics trust one another, agree [each party] has rightful power to influence, experience satisfaction with each other, and commit oneself to one another” (p. 12). In addition to the lack of a clear definition, researchers disagreed on whether an organization-public relationship is a subjective reality or an objective entity. Some scholars (Duck, 1973, 1986; J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999; Y. Huang, 1997a, 2001; Ki & Hon, 2007b; Ledingham & Bruning, 1998) conceived of it in a subjective way. Duck (1973) considered relationships inseparable from relational partners’ cognition and values. Others (Broom et al., 1997, 2000; Katz & Kahn, 1967; Oliver, 1990; Rogers, 1998) viewed it in an objective manner, independent of relational partners. Rogers stated, “The relationship lies in the connection [of relational partners], not in one or the other but in the between” (p. 77). She further suggested, “Relationships are socially performed, constructed, maintained, and altered in the reciprocal actions of
19
the relational members….Relationships do not have feelings, make imputations, engage in sense making, and the like; only the [relational] members have these capacities” (p. 78). Still others (Andersen, 1993; Cappella, 1991) saw a relationship as a combination of subjective perceptions and objective qualities independent of relational partners. Andersen argued, “Relationships are the combined product and producers of both the interpersonal interactions and the cognitive activity of the interactants” (p. 2). Most empirical research on organization-public relationships in the public relations literature to date has measured organization-public relationships as perceptions. Ki and Hon (2007b) reasoned that it is no surprise to treat relationships as relational parties’ perceptions. Broom and Dozier (1990) proposed a coorientational approach to measure both an organization’s and its publics’ perceptions. Hon and J. Grunig (1999) also asked for perceptions of an organization’s key publics. Only Broom et al. (2000) proposed assessing organization-public relationships as independent of relational members. They used a third party to observe organizational members’ interaction with each other. These observations are perceptions and judgments of the third party. Therefore, this study defined organization-public relationships as relational members’ perceptions of their relationship characteristics. Why Organization-Employee Relationships? Cutlip et al. (1985) pointed out that organization-employee relationships are the most important among all the relationships organizations have with their publics. Holtzhausen (2002) also noted that an important aspect of public relations managers’ job responsibilities is to maintain quality relationships with internal publics—employees. J. Grunig (1992) and L. Grunig et al. (2002) discussed in great detail the significance of
20
internal communication for organizational excellence. Rhee (2007) made a similar point citing literature from internal marketing. She wrote, “According to the proponents of internal branding, a corporation must first align its employees with its brand before it can make brand promises to its customers” (p. 8). Likewise, Guaspari (2002) claimed that internal marketing can retain employees whose increased loyalty then helps organizations build better relationships with external publics. It appears well understood that organizations cannot be what they are without support from their employees. Notwithstanding the significance of organization-employee relationships, Sallot, Lyon, Acosta-Alzuru, and K. Jones (2003) assessed the articles found in three major public relations journals from inception through 2000, and found that only two percent of the total number of published studies examined organization-public relationships. Among the two percent, even fewer focused on organization-employee relationships. Public relations scholars have continued to express concerns that researchers and practitioners alike have neglected employees as a type of public, assuming that their loyalty and commitment can always be counted on (D’Aprix, 1984; Hung, 2002; H.-S. Kim, 2003, 2005; Ni, 2006; Wright, 1995). It is clear that little is known of organization-employee relationships—a key area of research in the field of public relations, with the exceptions such as Rhee (2004), Ni (2006), and H.-S. Kim (2005). This study sought to provide more insights into this area. Research on Relationship Management Ledingham (2003) identified four pivotal developments of current research on organization-public relationships, which were: (1) the “recognition of the central role of relationships in public relations” exemplified by Ferguson’s (1984) article; (2)
21
reconceptualization of public relations as a strategic management function; (3) “identification of components and types of organization-public relationships, their linkages to public attitudes, perceptions, knowledge and behavior, and relationship measurement strategies”; and (4) “construction of organization-public relationship models that accommodate relationship antecedents, process, and consequences” (pp. 182183). These pivotal developments included articles and book chapters on definitions of organization-public relationships reviewed in preceding sections; application of the relational perspective to different contexts; maintenance strategies; and dimensions, types, and models of organization-public relationships. Application of the Relational Perspective The relational perspective has been used by organization-public relationships researchers in contexts such as public affairs, community relations, employee relations, media relations, manufacturer-retailer relations, university-student relations, issues management, and crisis communication (e.g., Bridges & Neilson, 2000; Coombs, 2000; Jo, 2003; Ki & Hon, 2007b; Ledingham & Bruning, 1998b; Ni, 2006; L. Wilson, 2000). For instance, Coombs (2000) integrated stakeholder theory and attribution theory as an organizing framework for a relational approach to crisis management. He suggested the ways in which existing stakeholder-organization relationships influence crisis managers’ selection of crisis response strategies. L. Wilson (2000) used a case study of Novell, Inc. to examine the linkage between perceptions of dimensions of organization-public relationships and a behavioral outcome loyalty. She argued that the company’s local community publics and employee publics expressed their loyalty because they perceived openness, trust, commitment and investment in the company’s actions.
22
Relationship Maintenance Strategies Relationship maintenance was commonly defined as “to keep a relationship in a satisfactory condition” and “to keep a relationship in repair” (Dindia, 2003, pp. 3-4). Existing public relations literature derived two types of relationship maintenance strategies—symmetrical and asymmetrical ones—from the interpersonal communication literature (e.g., Y. Huang, 1997a; J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999; Hung, 2007). The symmetrical strategies encompass disclosure/openness, assurances of legitimacy, participation in mutual networks, shared tasks, integrative negotiation, cooperation/collaboration, being unconditionally constructive, and win-win or no deal. The asymmetrical ones include distributive negotiation, avoiding, contending, compromising, and accommodating (J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999). There is a paucity of research focusing on relationship maintenance strategies (Hung, 2007; Ki & Hon, 2007a; Ki & Shin, 2005). Only a handful of studies examined relationship maintenance strategies through qualitative interviews, case studies, and content analysis of company Web sites (Bortree, 2003; Hong & Kiousis, 2007; Hung, 2003, 2004; Hung & Y. Chen, 2007; Kelleher & Miller, 2006; Ki & Hon, 2003, 2006, 2007a; Wigley, 2003). These studies verified several strategies theorized by Hon and J. Grunig (1999) and J. Grunig and Y. Huang (2000). For example, Ki and Hon (2006) discovered in an analysis of Fortune 500 companies’ Web sites that openness and access are two most commonly employed strategies by these organizations. It appears that the long list of strategies proposed by Hon and J. Grunig (1999) was often accepted as given. However, several questions remain, such as: Do all of the
23
strategies exist in all kinds of relationships organizations have with their various publics or does it depend on the type of publics? Do the symmetrical strategies have the same effect on quality of organization-public relationships as the asymmetrical ones? As researchers posited that relationship maintenance strategies are predictive of quality of organization-public relationships (J. Grunig & Y. Huang, 2000; Hon & J. Grunig), it is imperative to further understand the process of relationship maintenance. A first step can be looking into the above questions. Dimensions, Types, and Models of Organization-Public Relationships Dimensions. According to Ki and Shin’s (2005) analysis of published articles on organization-public relationships, dimensions of relationship quality were a primary focus of existing research on relationship management. Ferguson (1984) first proposed some dichotomous dimensions of quality of organization-public relationships, such as dynamic versus static, open versus closed, and satisfactory versus unsatisfactory. L. Grunig et al. (1992) theorized six dimensions: reciprocity, trust, mutual legitimacy, openness, mutual satisfaction, and mutual understanding. Y. Huang (1997a) combed through public relations models and conflict resolution strategies and developed four dimensions of relational quality in her dissertation: trust, control mutuality, relational commitment, and relational satisfaction. Ledingham and Bruning (1998a) presented five dimensions: trust, openness, involvement, investment, and commitment. Based on Y. Huang’s work, Hon and J. Grunig’s (1999) report on how to measure relationship quality in public relations outlined six dimensions: trust, commitment, satisfaction, and control mutuality, communal and exchange relationships. Len-Rios (2001) used these six dimensions in two online surveys to look at online organization-public relationships. J.
24
Grunig and Y. Huang (2000) contended that trust, commitment, satisfaction, and control mutuality (excluding the two types of relationships) seem to best represent the essence of organization-public relationships. Y. Huang (2001) produced a cross-cultural, multi-item scale of quality of organization-public relationships and added face and favor as a fifth dimension in addition to the four dimensions originally in her dissertation. Y. Kim (2001) suggested four relational dimensions: trust, commitment, local and community involvement, and reputation. Among the different versions of dimensions of the quality of organization-public relationships, J. Grunig and Y. Huang’s (2000) four dimensions were widely adopted in different studies. Ki and Shin’s (2005) review of 38 published articles in a dozen major academic journals from 1990 to 2005 revealed that satisfaction, commitment, trust, mutual understanding, control mutuality, and benefit were primary indicators of relationships reported by researchers. A key problem with the extant research on dimensions of the quality of organization-public relationships is that despite the different versions of dimensions and measures of these dimensions, only a few studies have measured the reliability and validity of the proposed scales (Y. Huang, 2001; Ki & Hon, 2007c; Y. Kim, 2001). Secondly, relationship quality has been treated as a one-sided concept, i.e., quality as perceived by publics, which is ironic considering the name organization-public relationships. Research is needed to address these problems. Types. Fewer studies have focused on the types of organization-public relationships (Hon & J. Grunig, 1999; Hung, 2002, 2005, 2007). Hon and J. Grunig distinguished communal relationships from exchange relationships. In communal
25
relationships, relational parties offer benefits to one another out of concerns for their welfare, expecting nothing in return. In exchange relationships, parties provide benefits to one another to return the favor they received in the past. The level of trust, control mutuality, satisfaction, and commitment that was developed in exchange relationships are not comparable to that in communal relationships. Most relationships begin as exchange relationships and then mature to communal ones (Hon & J. Grunig). Hung (2005, 2007) proposed six more types of relationships: exploitative relationships, manipulative relationships, symbiotic relationships, contractual relationships, convenantal relationships, and mutual communal relationships. Exploitative relationships happen when one relational party takes advantage of the other or does not fulfill his or her obligation when expected to. Manipulative relationships arise when an organization uses asymmetrical or pseudo-symmetrical communication on publics to serve its own interests after learning about the actual needs and interests of publics. Symbiotic relationships develop when organizations and publics work towards mutual goals (such as survival), with the knowledge of their interdependence in the environment and awareness of the consequences of their behavior on one another. Contractual relationships occur when relational parties come to an agreement on what each should do in the relationship, analogous to signing a contract. Such relationships are not necessarily equal; power is a salient factor. Convenantal relationships are characterized by both parties committed to a common good through open exchanges of communication (e.g., suggestions and criticisms). The norm of reciprocity is applicable in such relationships. An example of convenantal relationships is between a teacher and students. Mutual communal relationships are those in which relational parties are concerned about the
26
welfare of the other. This focus on psychological intention differentiates mutual communal relationships from convenantal relationships as the latter emphasize open exchanges. Hung (2005) argued that Hon and J. Grunig’s (1999) communal relationships should be called “one-sided communal relationships,” which are more sophisticated than mutual communal, because of the expectation of nothing in return (reciprocity becoming unnecessary). Hung (2005) cited her personal communication with J. Grunig and explained that J. Grunig developed a continuum of types of relationships ranging from concern for oneself (exploitative relationships) to concern for others (communal relationships). As the continuum moves towards concern for others, there exist contractual relationships, exchange relationships, and convenantal relationships. Hung (2005, 2007) modified this continuum: with exploitative relationships on the end of “concern for self interests,” followed by manipulative relationships, contractual relationships, symbiotic relationships, exchange relationships, convenantal relationships, and mutual communal relationships, and one-sided communal relationships on the end of “concern for others’ interests.” She called exchange relationships, convenantal relationships, and mutual communal relationships the “win-win zone” for relational parties. Models. There are several models of organization-public relationships in the public relations literature (Broom et al., 1997, 2000; Bruning & Ledingham, 1999; J. Grunig & Y. Huang, 2000; Hung, 2007; Toth, 2000). Broom et al. (1997, 2000) were among the first to develop such a model. They suggested a three-stage model of relationship management with antecedents, concept and consequences. Posited as causes of the formation of relationships, antecedents include social and cultural norms,
27
collective perceptions and expectations, needs for resources, perceptions of uncertain environment, and legal/voluntary necessity. Representing the process of relationship formation, concept is comprised of properties of exchanges, transactions, communication, and other interconnected activities. As outputs of the relationships, consequences are goal achievement, autonomy, routine and institutionalized behavior. Broom et al. noted that in the long run consequences eventually will result in changes in the antecedents, which then influence the concept. Concept ultimately leads to changes in consequences. Relying on the strategic management theory and the excellence theory, J. Grunig and Y. Huang (2000) developed a parallel three-stage model consisting of situational antecedents, maintenance strategies, and relationship outcomes. The situational antecedents describe how relationships come into being, typically when organizations and publics have consequences on each other, such as organization(s) affecting a public/publics and vice versa. The middle part of the model is maintenance strategies or specific activities to maintain a relationship, which are categorized as symmetrical and asymmetrical. These strategies are derived from the interpersonal communication literature, particularly work by Stafford and Canary (1991), as well as the conflict resolution literature that is applicable to public relations, based on Y. Huang’s (1997a) and Plowman’s (1995) research. The symmetrical strategies are posited as more effective than the asymmetrical ones, analogous to the symmetrical and asymmetrical model in the excellence theory by L. Grunig et al. (2002). Examples of the symmetrical strategies include disclosure (openness), shared tasks (helping to solve problems of interest to the other party), and integrative negotiation (maximizing joint gains); whereas the asymmetrical ones consist of distributive negotiation (maximizing self-gain), avoiding,
28
and contending (trying to convince the other to accept one’s position). Relationship outcomes resulting from such maintenance strategies are control mutuality, commitment, satisfaction, trust and goal attainment. Toth (2000) presented and contrasted a “pure personal influence” type of public relations practice in which interpersonal communication is utilized to “dominate individuals, to accept either the organization’s or public’s position” and a “pure interpersonal influence” type of practice in which interpersonal communication is used to “influence to find mutual definitions, mutuality of understanding, agreement, [and] consensus” (p. 214). The former is “closed and static in attributes” whereas the latter is “open and dynamic in attributes” (Toth, p. 214). She proposed the adoption of the “pure interpersonal influence” model in relationship management. Bruning and Ledingham (2000) developed a five-step process of relationship management: scan, map, act, rollout, and track (SMART). Akin to J. Grunig and Y. Huang’s (2000) notion of environment scanning, scan entails “surveil[ling] the environment to better understand the current state of organization-public relationships and the communication patterns that exist in those relationships, as well as to determine the current state of key public member knowledge, opinions, attitudes, and behaviors” (p. 92). Map means working out a strategic plan with strategies, tactics, objectives, and goals. Act includes conducting a field test of the strategic plan to make modifications as needed. Rollout is implementing the modified strategic plan on key publics. The last step track refers to tracking the influence of the organization’s activities on key publics’ perceptions and behaviors through the use of both quantitative and qualitative methods.
29
Incorporating both the social exchange theory and the dialectical perspective, Hung (2007) developed a relationship management model that views relationship as an ongoing dynamic process, in which each stage of relationship development is subject to changes of the previous stage. Her model consists of relationship objectives, cultivation, development and change strategies, and relationship outcomes. Hung (2007) posited that the types of relationships (under relationship objectives) determine the types of cultivation, development and change strategies, i.e., symmetrical versus asymmetrical, the choice of which affects relationship outcomes. Relationship objectives have four components. The first component is the type of relationships an organization intends to have with its publics resulting from its acknowledgement of interdependence with publics and subsequent intention and motivation to survive in the environment. She listed eight types of possible relationships based on Hung (2005)—one-sided communal, mutual communal, convenantal, exchange, manipulative, contractual, symbiotic, and exploitative relationships. The second component is the quality of relationships that the organization has with its publics in the previous stage (trust, control mutuality, commitment, and satisfaction). The third and fourth components respectively are goal attainment and reputation perceived by the publics. Cultivation, development and change strategies consist of the symmetrical and asymmetrical categories. The symmetrical ones are openness/disclosure, positivity, legitimacy, networking, sharing tasks, keeping promises, cooperation, being unconditionally constructive, and win-win or no deal. The asymmetrical ones include distributive negotiation, avoiding, compromising, contending, and accommodating.
30
Relationship outcomes comprise the quality of relationships (trust, control mutuality, commitment, and satisfaction), the extent of goal attainment, reputation perceived by publics, and types of relationships that may differ from the organization’s initial expectation and intention. They are the same components of relationship objectives. Conclusion Ledingham (2003) contended that the above reviewed four pivotal developments of current research on how organization-public relationships contribute to a theory of relationship management. He summarized the findings of extant research on relationship management as the following: 1. In the relational perspective, relationship presuppositions act as a framing mechanism for theory building, teaching, and practice. The use of communication output as the measure of programmatic accountability is superceded by measurement of relational and behavioral outcomes. 2. The appropriate unit of measurement of public relations impact is the organization–public relationship. 3. Analysis of organization–public relationships is grounded in interpersonal relationship building. Moreover, many interpersonal relationship principles— complete with guidelines for initiating, maintaining, and improving relationships—serve as a foundation for the exploration of organization–public relationships. 4. Organization–public relationships involve an ongoing interchange of needs, expectations, and fulfillment.
31
5. Ratings of relationship dimensions can define the state of an organization– public relationship, which, in turn, can act as a predictor of public perceptions and choice behavior. 6. Organization–public relationship types include symbolic and behavioral, as well as personal, professional, and community-related. 7. Organization–public relationships change over time. 8. The outcome of effective relationship management is mutual understanding and benefit. 9. Successful organization–public relationships develop around common interests and shared solutions to common problems. 10. Relationship state reflects perceptions of needs and expectations fulfillment. 11. Mutual benefit strategies can generate economic, societal, and political gain both for organizations and publics. Based on these findings, Ledingham (2003) argued that a theory of relationship management has emerged, which is that “Effectively managing organizational[sic]-public relationships around common interests and shared goals, over time, result in mutual understanding and benefit for interacting organizations and publics” (p. 190). In other words, the value of public relations exists in effective management of relationships. Ledingham’s (2003) piece was commendable with regard to providing a useful framework for research on relationship management. Nevertheless, it was rather vague what he meant by “common interests and shared goals”, and “mutual understanding and benefit” and what “effective management” entailed. Furthermore, although Ledingham explained that this theory of relationship management is a general theory that
32
accommodates middle-range theories, such as using the situational theory of publics by J. Grunig (1997) to identify and categorize publics, it was not clear what specific middlerange theories can be considered as components of the general theory of relationship management. Toward A Theory on Organization-Employee Relationships As is shown above, despite the burgeoning research on relationship management, more theory-building endeavors are necessary. Because of the interdisciplinary nature of the research on relationship management, it is important to first lay out the theoretical foundations. Extant research has relied on the systems theory, resource dependence theory, social exchange theory, and relational dialectics theory as their frameworks (Broom et al., 1997, 2000; J. Grunig & Y. Huang, 2000; Y. Huang, 1997; Hung, 2007). Systems Theory A system is “a set of interacting units with relationships among them” (Miller, 1978, p. 16). Organization theorists using the systems theory by Bertalanffy see organizations as open systems, characterized by “a continuous cycle of input, internal transformation (throughout), output, and feedback (whereby one element of experience influences the next)” (G. Morgan, 1997, p. 40). They emphasize the environment in which organizations exist, define organizations in terms of interrelated subsystems, and seek “to establish congruencies or ‘alignments’ between different systems and to identify and eliminate potential dysfunctions” (G. Morgan, p. 39). For instance, Lee and Jablin (1995) contended that work relationships are subsystems nested within larger organizational systems. Likewise, Waldron (2003) argued:
33
Any given work relationship is nested within a complex system of vertical and horizontal networks. The communication of co-worker peers is influenced in part by their individual relationships with those in power and their perception of the supervisory relationships their peers enjoy. (p. 165) Two relevant concepts from the systems approach are interdependence and homeostasis (Broom et al., 1997, 2000; Vogl-Bauer, 2003; Waldron, 2003). Interdependence means that the environment and the organizations are seen interacting with and dependent upon each other. The essence of such interactions is mutual adaptation (Broom et al., 2000, p. 15). The notion of interdependence has been widely used in research on relationship management (Broom et al., 1997, 2000; L. Grunig et al., 1992; Hung, 2005; Yang, 2005). Hung defined organization-public relationships as interdependence, assuming “when an organization realizes the interdependence with its publics, it either competes or collaborates with its publics in acquiring the resources for its survival” (p. 396). Homeostasis is “self-regulation and the ability to maintain a steady state” (G. Morgan, 1997, p. 40). Systems use homeostatic control processes to sustain themselves. With regard to relationship management, relationship maintenance strategies are examples of such homeostatic control processes by systems to stay in a steady state (Dindia & Canary, 1993; Lee & Jablin, 1995). Nonetheless, the problem with homeostasis is its association with stability, negating the fact that social relationships involve both stability and change (Baxter & Montgomery, 1996). A complete acceptance of stability makes it unlikely for relationships to grow or adapt to new circumstances.
34
Resource Dependence Theory Resource dependence theory was one of the most influential theories of organization-environment relations by Pfeffer and Salancik (Hatch, 1997). The assumption is that organizations need resources from their environment for survival, growth and achievement of goals, such as raw materials, labor, capital, and technology (Hatch, 1997; Broom et al., 2000). This is consistent with the notion of interdependence of the systems theory. The basic argument of the theory is that managers’ knowledge of the power/dependence relationships between their own organization and the network it is in can help the organization anticipate sources of influence from the environment and work out ways to offset such influence (Hatch, p. 78). Often managers are required to sort the resources they need according to their criticality and scarcity. Both critical and scarce resources should have the highest priority (Hatch). Furthermore, managers can help organizations develop sustained competitive advantage by creating or obtaining resources that are difficult to be imitated by competitors (Barney, 1991). One example of managing labor and knowledge dependence is organizational strategies to recruit executives from competitors (Hatch). This theory has been applied to understanding inter-organizational relationships. Van de Ven (1976) posited that organizations form relationships with one another because they depend on each other for resources. Specifically, an organization may form a relationship with another because the resources it needs are scarce, or it needs certain specialized skills or services, or because it competes with another organization for clients, resources, etc (Van de Ven & Walker, 1984).
35
Employing this theory to organization-employee relationships, one can see that organizations need labor and knowledge from employees for their survival and growth. Organizations’ strategies to maintain their relationships with employees are efforts to counter the dependence on one resource—employees, and to develop unique competitive advantage over competitors. Social Exchange Theory Originated in psychology and sociology, social exchange theory was used for relational communication (Liska & Cronkhite, 1995; Roloff, 1981). The theory holds that social relationships involve exchanges of relationships, such as status, information, services, goods, money, love, companionship, friendship, and social acceptance (Devito, 1996, Hinde, 1997). People develop relationships to maximize profits. Profit is rewards minus costs. Usually individuals in a relationship have certain expectations of the kind of profits to be obtained from the relationship, referred to as comparison level. The theory suggests that whether a person will stay in a relationship depends on his or her comparison level for alternatives. Individuals compare current relationship profits with anticipated profits from alternative relationship choices. In essence, individuals tend to be less satisfied if they have many relationship options available, unless their expectation of the current relationship is met or exceeded (Thomlison, 2000). When applied to relationship maintenance, the activities to cultivate a relationship can be conceived of as costs. If the comparison level as well as the comparison level for alternatives can be determined, organizations then may predict the amount of rewards necessary to maintain a relationship (Thomlison, 2000). According to Thomlison, public
36
relations practitioners should be constantly monitoring the comparison level and the comparison level for alternatives of their publics. This theory is similar to resource dependence theory in its emphasis on exchanging resources. Nevertheless, with regard to explaining the formation of relationships, J. Grunig and Y. Huang (2000) contended that neither the resource dependence theory nor the social exchange theory can account for all the change pressures from the environment. They stated: These [change] pressures may have little or nothing to do with resources or with exchanges. Instead, they may come from publics, activist groups, government, or media who simply want to change the behavior of the organization in specific situations, thus interfering with the behavior the organization wants to implement. (p. 35) Further, J. Grunig and Y. Huang (2000) claimed, “[T]he antecedents of relationships are situational, just as publics are situational … publics come and go and change as situations change” (p. 35). The Dialectical Approach The systems theory, resource dependence theory and social exchange theory all emphasize the notion of interdependence, a defining nature of organization-public relationships (including organization-employee relationships). However, the notion of environment is too ambiguous in the systems theory. Moreover, the implication of stability by the concept of homeostasis in the systems theory neglects the fact that social relationships are characterized by both stability and change. The problem with resource dependence theory and social exchange theory is that employees are not just resources.
37
Employees’ decisions to enter or exit the relationships with their organizations may not be simply a calculation of rewards minus costs. Therefore, I relied on the dialectical approach that embraces the notion of interdependence but also better captures the quality of organization-public (including employees) relationships. Applying Marx’s concept of dialectic and Bakhtin’s extension of Marxist dialectic—dialogism, Baxter, Montgomery, and Rawlins proposed the dialectical approach to relationship studies (Baxter, 1988; Baxter & Montgomery, 1996; Montgomery, 1993; Rawlins, 1989, 1994). These researchers view the dialectical approach as a perspective, other than a theory of prediction and causal explanation (Baxter & Montgomery). It does not describe an array of axioms or propositions; rather, it provides a set of assumptions—contradiction, change, praxis, and totality (Baxter & Montgomery). The dialectical approach maintains that change, tension, and instability are ever present qualities of social relationships. Relational partners spiral between opposing yet interdependent communicative tendencies, in contrast with the traditional linear model that expects individuals to seek a certain state of satisfaction. That is, “partners are constantly being pulled together and simultaneously being pushed apart” (Dainton, Zelley, & Langan, 2003, p. 83). The dialectical approach consists of four key assumptions: contradiction, change, praxis, and totality (Baxter & Montgomery, 1996). Without the traditional negative connotation, contradiction is defined as “the dynamic interplay between unified oppositions”, in which oppositions are “tendencies or features of a phenomenon” (Baxter & Montgomery, p. 8). A contradiction is formed when two opposing yet interdependent
38
forces or tendencies mutually negate each other. One cannot understand one force without understanding the other (Dainton et al., 2003). Dialectical theorists commonly use contradiction as the unit of analysis in their research. Extending this to organizationpublic relationships, oppositions can be regarded as opposing yet interdependent relational parties that have shared as well as opposing interests. A contradiction then is the dynamic interaction between these opposing relational parties who negate one another at the same time that they are interdependent on one another. It resembles L. Grunig et al.’s (2002) idea of symmetry—not consensus, but “the give-and-take of persuasion and collaboration that organizations and publics use when they must interact with each other” (p. 551). This concept also fits squarely with the notion of congruence of organizationpublic relationships, which captures the interaction between an organization and its publics. The second assumption change arises from contradiction. Because of the tug-ofwar nature, it is impossible to fulfill both of the opposing yet interdependent forces or tendencies. Change then is always present; it forms a dialectical unity with stability (Baxter & Montgomery, 1996). In other words, “dialectical change is the interplay of stability and flux” (Baxter & Montgomery, p. 10). The change process is like a spiral that contains elements of cyclical change (recurrence) as well as linear change (non-repeating changes). Praxis means that “people are at once actors and objects of their own actions” (Baxter & Montgomery, 1996, p. 12). This concept suggests that people are both proactive in making decisions and choices and reactive to partners’ decisions and choices. Such choices and decisions lead to different patterns of dialectical change. For example,
39
subordinate employees and managers who act as representatives of an organization may perceive too little trust and too much distrust in their relationships. Consequently, they may choose to ignore the tension, or try to increase trust and decrease distrust, or simply accept it as inevitable, or redefine the nature of their relationships. They are both proactive and reactive. Furthermore, their present decisions and choices are bound to constrain their future interactions. Totality refers to the idea that “social phenomena are defined by the relations among their characteristics, not by the characteristics themselves” (Montgomery, 1993, p. 206). Put simply, social phenomena are assumed to be understood only in relation to other phenomena. Consequently, relationships “are created by individuals’ interactions with other elements in society, but they also create individuals and societies” (Hung, 2007, pp. 452-453). The dialectical approach is consistent with the organization-public relationships theory summarized by Ledingham (2003). Ledingham stated that “organization-public relationships involve an ongoing interchange of needs, expectations, and fulfillment” and “organization-public relationships change over time” (p. 188), which is commensurate with the concept of change and contradiction. Also, the notion of interdependence lies at the core of the dialectical approach, and constitutes the heart of relationships in the relationship management literature. Furthermore, the dialectical approach requires us to understand organization-public relationships from both the organization’s and the public’s perspective, and to analyze the dynamic interaction between them. The dialectical approach also allows us to revisit the organization-public relationships theory—“Effectively managing organization[sic]-public relationships
40
around common interests and shared goals, over time, results in mutual understanding and benefit for interacting organizations and publics” (Ledingham, 2003, p. 190). Contradiction spurs change. Organizations and publics may not always have the same goals or interests. They often have opposite goals or interests. However, the interdependent yet opposing tension in their relationships give rise to interaction and change. During the process, they can find common grounds. Both the organization and its publics consciously make their own decisions but also react to each other’s decisions. This constant tension and change in their relationships lead to benefits of varying degrees for the organization and publics as well as better understanding of each other—they may not agree with each other. Thus, the above statement can be changed to a major proposition about managing relationships and organizational effectiveness: Proposition 1: The dynamic and constant tension and change in organizationpublic relationships around negotiated interests and goals lead to benefits of varying degrees for the organization and its publics and better understanding of each other. In this light, relationship maintenance strategies can be part of the tension and change. The organization and its publics’ perceptions of their relationships show their “understanding of each other.” In the context of organization-employee relationships, such benefits include organizational effectiveness. As Ledingham (2003) stated, “models are an illustration of theories in action” (p. 186). Based on the models in the literature (Broom et al., 1997, 2000; Bruning & Ledingham, 1999; J. Grunig & Y. Huang, 2000; Hung, 2007; Toth, 2000) and the above discussion, I proposed a three-stage model of organization-employee relationships to shed new light on theory-building of relationship management.
41
In the next sections I offer conceptualization of focal constructs: relationship maintenance strategies (symmetrical and asymmetrical), perceived relationship characteristics, congruence of perceived relationship characteristics, organizational effectiveness, contextual performance, and turnover intention; and reviews of relevant literature. I conclude with review of literatures linking the focal constructs and introduction of the conceptual model. Relationship Maintenance Strategies Definition and a Terminological Issue As Canary and Dainton (2003) stated, “Most sane people know that [interpersonal] relationships require work. That is, partners need to spend time and effort to maintain functional, satisfying relationships. Without such efforts, relationships tend to deteriorate” (p. xiii). This also applies to organization-public relationships. There are four common definitions of relationship maintenance strategies in the interpersonal communication literature: 1) “to keep a relationship in existence;” 2) “to keep a relationship in a specified state or condition;” 3) “to keep a relationship in a satisfactory condition;” and 4) “to keep a relationship in repair” (Dindia, 2003, pp. 3-4). The first definition does not concern the quality of a relationship. A dissatisfying relationship that is kept in existence is not worth maintaining. The second definition is at odds with the dialectical approach that emphasizes the constant change and adaption to opposing tensions in a relationship. In contrast, the third and fourth definitions allude to the endeavors by an organization to make its publics satisfied and prevent the relationships from decaying. Therefore, this study combined the last two definitions to refer to relationship maintenance strategies.
42
Consistent with the dialectical approach, an organization constantly and strategically needs to keep its relationships with the publics satisfying, trusting, and so on; and use preventive and corrective maintenance to prevent something from going awry with the relationship. These planned organizational behavioral attempts or efforts to keep a relationship in a satisfying condition and in repair are called relationship maintenance strategies2. A related terminological issue is whether these strategies include routine behaviors. Dainton and Stafford (1993) distinguished these two by labeling strategic maintenance as planned and conscious whereas routine behaviors as not intentional and automatic. Slightly differently, Dindia (2003) admitted that the dichotomy between routine and strategic relational maintenance behaviors may not be so clear-cut, even though she advocated a delineation of the two. Canary (2001, personal communication cited in Stafford, 2003) further departed from Dainton and Stafford’s position and contended that it was unnecessary to separate routine and strategic, stating: For me, a strategy is defined as an approach someone takes. In other words, strategic communication is implicitly learned and often mindlessly enacted. This is a broad definition of “strategic” that encompasses a lot of behavior. Strategic approaches are often routinized but become more cognitively processed when the routine plan does not work. (p. 54) It is true that in reality strategic and routine relational maintenance behaviors are often intertwined. Some maintenance behaviors may be strategic initially but become
Montgomery (1993) rejected the term relational maintenance and advocated relational sustainment in that the former portrays relationships as static to her. Similarly, J. Grunig (2006) replaced maintenance with cultivation strategies. I argue that maintenance does not necessarily imply static. Organizations must constantly monitor their maintenance strategies to keep the relationships in a satisfying condition and in repair. This means the same as sustainment and cultivation.
2
43
routine over time. Some organizations’ or people’s strategic maintenance behaviors may be perceived as routine by others, and vice versa. Stafford (2003) thus suggested using the more encompassing term maintenance behaviors in lieu of maintenance strategies to avoid controversy. I adhered to Dindia’s position that theoretically speaking, intent differentiates strategic from routine relational maintenance behaviors, and agreed that in practice this distinction appears difficult and not as meaningful and necessary. Relationship Maintenance Strategies in the Literature Prior research has proposed many relationship maintenance strategies. For example, drawn from the literature of interpersonal communication (e.g., Stafford & Canary, 1991) and conflict resolution and negotiation, Hon and J. Grunig (1999) proposed a set of relationship maintenance strategies. J. Grunig and Y. Huang (2000) modified those strategies and categorized them into two types: symmetrical and asymmetrical. The symmetrical strategies are disclosure or openness, assurances of legitimacy, participation in mutual networks, shared tasks, integrative negotiation, cooperation/collaboration, being unconditionally constructive and win-win or no deal. The asymmetrical ones encompass distributive negotiation, avoiding, contending, compromising, and accommodating. Hon and J. Grunig pointed out that the symmetrical (mutually beneficial) strategies are more effective in maintaining relationships than the asymmetrical (self-gain interested only) ones, saying “the most productive relationships in the long run are those that benefit both parties in the relationship rather than those designed to benefit the organizations only” (p. 11). Symmetrical Maintenance Strategies
44
Disclosure or openness. Canary and Stafford (1994) defined disclosure or openness as “direct discussion about the nature of the relationship and setting aside times for talks about the relationship” (p. 12). Hon and J. Grunig (1999) conceived of this strategy as being open about one’s thoughts and feelings. For example, a relational party may encourage the other to disclose thoughts and feelings to him or her, try to discuss the quality of relationship with the other, and remind the other of past relationship decisions they made (Stafford, 2003). In the context of organization-public relationships, both organizations and publics are supposed to be open and frank about what they think about certain situations and decisions, and share with each other concerns they hold (L. Grunig et al., 2002). J. Grunig and Y. Huang (2000) suggested that public relations managers might measure publics’ disclosure by “counting suggestions, complaints, inquiries, and other contacts that members of publics, the media, government, or leaders of activist groups make with the organization” (p. 41). J. Grunig and Y. Huang (2000) considered disclosure or openness logically connected with the two-way symmetrical model of public relations, and positive relationship outcomes. Although Hung (2000) pointed out that openness or disclosure may not guarantee quality relationships, Bok (1989) noted that sharing information can lead to better equity of power in a relationship because access to information is power and often the more powerful can access more information than the powerless. Networking. In interpersonal relationships, networking means spending time with mutual friends, focusing on common friends and affiliations, and expressing willingness to do things with partners’ friends and family (Stafford, 2003). From the perspective of building organization-public relationships, also referred to as participation in mutual
45
networks, this strategy asks organizations to build “networks or coalitions with the same groups that their publics do, such as environmentalists, unions, or community groups” (Hon & J. Grunig, 1999, p. 15). Likewise, publics may try to join networks or coalitions to which organizations belong. J. Grunig and Y. Huang (2000) proposed to assess networking through counting organizations’ contacts with networks of activist groups. Assurances of legitimacy. Assurances of love and commitment are common in interpersonal relationships. For example, married couples stress their commitment to each other and show themselves to be faithful to each other; relational partners indicate that the relationship has a future (Stafford, 2003).With regard to organization-public relationships, assurances of legitimacy are attempts by relational parties to assure each other that their needs, concerns, and problems are legitimate and should be addressed (Hon & J. Grunig, 1999). This is also in line with the two-way symmetrical model (L. Grunig et al., 2002). Hon and J. Grunig (1999) noted that demonstration of commitment to maintaining a relationship can help assure legitimacy. They provided an example of a university demonstrating assurances by developing a new land grant system. The director of external relations for this university’s agricultural extension office explained that university programs catered to the needs of publics and tried to deliver in the way publics wanted. Similarly, L. Grunig’s (1992) research on activist groups confirmed that assurances of legitimacy by organizations are necessary when dealing with activists. Shared tasks. This strategy typically means sharing household responsibilities in interpersonal relationships (Canary & Stafford, 1992, 2001; Stafford, 2003). Messman, Canary, and Hause (2000) in their examination of non-romantic relationships identified
46
shared activity as a maintenance strategy, consisting of sharing special rituals, specific routine activities, and time with the other relational parties. Extending this notion of shared tasks to organization-public relationships, Hon and J. Grunig (1999) conceptualized shared tasks as “organizations’ and publics’ sharing in solving joint or separate problems” (p. 15). They included examples of tasks for organizations and publics such as “reducing pollution, providing employment, making a profit, and staying in business” that interest either the publics or the organizations or both (p. 15). As the organizations’ devotion to solving these problems can be recognized as the organizations’ attempts to ensure the publics that their concerns and problems are legitimate and should be addressed, assurances of legitimacy can subsume shared tasks. Advice. Another maintenance strategy in the interpersonal communication context that can be used in organization-public relationships is advice (Stafford, Dainton, & Haas, 2000)—telling a relational party what he or she needs to do about his or her problems or giving opinions on things going on in his or her life. Given that Hon and J. Grunig’s (1999) and J. Grunig and Y. Huang’s (2000) were based on the earlier work of Stafford and Canary (1991), advice can be added to the collection of maintenance strategies. For example, an organization’s management, as the representative of the organization, can advise about subordinate employees’ careers and/or personal lives when relevant to their career. In addition to these interpersonal strategies, Hon and J. Grunig (1999) and J. Grunig and Y. Huang (2000) suggested a list of symmetrical strategies originated from the conflict resolution and negotiation literature.
47
Integrative negotiation. Putnam (1990) described integrative negotiation as attempts to use “open information exchange and joint decision making” to accomplish joint gains or mutually beneficial outcomes (p. 3). Hon and J. Grunig (1999) considered integrative negotiation a symmetrical strategy because common/mutual interests and winwin solutions are sought out. Nevertheless, “open information exchange” overlaps with disclosure or openness, and “joint decision making” can be subsumed by assurances of legitimacy. Therefore this strategy was not included in this study. Cooperation/collaboration. Cooperation or collaboration involves both the organization and the public trying to reconcile their interests toward building a mutually beneficial relationship (J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999). Such attempts are similar to disclosure or openness, which also aims to reconcile different interests through honest discussions. This strategy accordingly can be incorporated into disclosure or openness. Being unconditionally constructive. This strategy means that “the organization does whatever it thinks is best for the relationship, even if it means giving up some of its positions and even if the public does not reciprocate” (Hon & J. Grunig, 1999, p. 17). It has been often used by organizations when faced with resistant activists who do not compromise or even refuse to negotiate (J. Grunig & Y. Huang, 2000). Win-win or no deal. This strategy is used when the organization and the public fail to find a mutually beneficial solution and agree to disagree — no deal (Hon & J. Grunig, 1999). It is viewed as symmetrical because agreeing to disagree leaves open the potential that the organization and public may still be able to work together to find a winwin solution next time (Covey, 1989; J. Grunig & Y. Huang, 2000; Hon & J. Grunig).
48
This strategy as well as being unconditionally constructive examines how an organization works with its publics for a mutually beneficial solution, which overlaps with disclosure or openness and assurances of legitimacy. Hence, this study excluded these two strategies. Asymmetrical Maintenance Strategies In addition to these symmetrical maintenance strategies, organizations and publics also use asymmetrical ones: distributive negotiation, avoiding, contending, compromising, and accommodating. Distributive negotiation. This strategy aims for self-gains, involving attempts to maximize gains and minimize losses for oneself (J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999). Specific tactics of this strategy include imposing one’s position onto the other relational party without concern for the other, threatening, dominating, and arguing (Hon & J. Grunig; Morill & Thomas, 1992; Putnam & C. Wilson, 1982). This win-lose or self-gain perspective indicates the asymmetry of this strategy. Dual concern strategies. Plowman identified four dual concern strategies, namely, avoiding, contending, compromising and accommodating. They were called “mixed motive or collaborative advocacy” by Hon and J. Grunig (1999, p. 16). An organization and its publics do take each other’s interests into consideration when employing these strategies, but emphasize self interests over others’ interests; thus they are still asymmetrical in nature. Avoiding was defined as “the organization leaves the conflict either physically or psychologically” (Hon & J. Grunig, p. 16). Hess (2003) put forward similar strategies in the interpersonal communication context—avoidance (“[t]rying not to be in the presence of the other person”), ignoring (“[a]cting as if the other person is not there”), and
49
inattention (“[g]iving as little attention as possible to the other person”) (p. 115). Put in the organization-employee relationships context, avoiding/ignoring/inattention can be classified as one strategy that either the organization or public or both use to distance from the other. Contending refers to the extent to which an organization tries to convince the public to accept its position or vice versa (J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999). The notion of contending is analogous to some aforementioned distributive negotiation tactics, such as threatening, dominating, and arguing. Therefore this study used distributive negotiation only. Compromising and accommodating were respectively conceptualized as “the organization meets the public part way between its preferred positions, but neither is completely satisfied with the outcome” and “the organization yields, at least in part, on its position and lowers its aspirations,” which were virtually the same (J. Grunig & Y. Huang, p. 40). Therefore, I will labeled them as one strategy compromising. In sum, there are two kinds of relationship maintenance strategies in the public relations literature: symmetrical and asymmetrical ones. The symmetrical strategies consist of disclosure or openness, assurances of legitimacy, networking, and advice. The asymmetrical ones encompass distributive negotiation, avoiding, and compromising. J. Grunig (2006) considered relationship maintenance strategies “the heir to the models of pubic relations and the two-way symmetrical model, in particular” (p. 168). Truly the symmetrical relationship maintenance strategies are consistent with the two-way symmetrical model, underscoring collaboration, two-way communication, the use of both mediated and interpersonal forms of communication, and nearly always ethical practice.
50
In contrast, the asymmetrical relationship maintenance strategies are logically connected with the two-way asymmetrical model, characterized by advocacy, one-way communication, the use of either mediated or interpersonal forms of communication, and sometimes ethical or unethical practice. Accordingly, I end the above discussion with the second research proposition and first research question on relationship maintenance strategies: Proposition 2: Symmetrical relationship maintenance strategies are composed by disclosure or openness, assurances of legitimacy, networking, and advice. Asymmetrical relationship maintenance strategies include distributive negotiation, avoiding, and compromising. Research Question 1: How do organizational members perceive relationship maintenance strategies used by their organizations? Congruence of Perceived Relationship Characteristics Using the congruence between two constructs or two sides of a construct (e.g., non-management employees vs. managers’ perceptions) is not uncommon in organizational psychology (e.g., Ostroff et al., 2004; Saks & Ashforth; London & Wohlers, 1991; Shore & Bleicken, 1991). As noted earlier, I intended to measure both sides of organization-employee relationships, thus using the construct of congruence of perceived relationship characteristics. It refers to the degree of congruence between employees’ and their organizations’ perceptions of their relationship characteristics. Such characteristics describe the quality of organization-employee relationships. I first review the dimensions of relationship quality and then discuss issues related to congruence.
51
Prior research on relationship management has shown that the quality of an organization-public relationship is multidimensional. Extensive scholarly work has been devoted to identifying these dimensions. Ferguson (1984) was among the first researchers to suggest a few dichotomous dimensions of organization-public relationships, including dynamic versus static, open versus closed, and satisfactory versus unsatisfactory. L. Grunig et al. (1992) proposed six dimensions of perceived relationship state: reciprocity, trust, mutual legitimacy, openness, mutual satisfaction, and mutual understanding. These pioneering scholars nonetheless did not specify how to measure these dimensions or test these theoretical dimensions. Hon and J. Grunig (1999) were among the first to empirically test these dimensions. They derived six dimensions from the interpersonal communication and psychology literature: trust, commitment, satisfaction, control mutuality, communal relationships, and exchange relationships. The University of Maryland research team invited 200 people from randomly chosen email addresses to take an online survey of their perceptions of relationships with five large organizations. The sub-scales (for each dimension) had good reliability (? > .80 on average). Communal relationships and exchange relationships were dropped later by J. Grunig and Y. Huang (2000). Ever since, the remaining four dimensions (trust, commitment, satisfaction, and control mutuality) have been widely used in a variety of studies (e.g., J. Grunig & Hung, 2002; Y. Huang, 2001; Ki & Hon, 2007b, 2007c; Ki & Shin, 2005; H.-S. Kim, 2005, 2007; Y. Kim, 2001; Yang & J. Grunig, 2005; Yang, 2007). Y. Huang added a fifth dimension to Hon and J. Grunig’s four-dimensional measuring instrument—face and favor, which is presumably unique in the Taiwanese culture. This study used these four
52
dimensions (trust, commitment, satisfaction, and control mutuality) to assess characteristics of organization-employee relationships and proposed a fifth dimension-distrust. I present a detailed explanation of these terms below. Trust Trust as a dimension of relationship characteristics in organizational settings has been widely acknowledged (see Mayer, Davis, & Schoorman, 1995; Schoorman, Mayer, & Davis, 2007 for more details). Hon and J. Grunig (1999) defined trust as “one party’s level of confidence in and willingness to open oneself to the other party” (p. 5). Similarly, Moorman, Deshpande and Zaltman (1993) viewed trust as “a willingness to rely on an exchange partner in whom one has confidence” (p. 82). On a different note, the well-cited marketing scholars R. Morgan and S. Hunt (1994) opposed to having “willingness” in the conceptualization of trust because willingness implies behavioral intention. They asserted, “Just as behavioral intention is best viewed as an outcome of attitude and not as part of its definition … ‘willingness to rely’ should be viewed as an outcome … of trust and not as a part of how one defines it” (p. 24). As Hon and J. Grunig (1999) considered relationship characteristics as perceptions rather than behavioral intentions, this study removed the “willingness” component from Hon and J. Grunig’s (1999) definition and defined trust as one party’s level of confidence in the other party. According to Hon and J. Grunig (1999), trust has three sub-dimensions: integrity, dependability, and competence. Integrity shows one party’s belief of consistency and fairness in the other’s behaviors. An example of integrity is “This organization treats people like me fairly and justly” (Hon & J. Grunig, p. 28). Dependability describes the consistency between one’s words and behaviors. An example of dependability is “This
53
organization can be relied on to keep its promises” (Hon & J. Grunig, p. 28). Competence indicates the extent to which one party believes the other has the ability to do what it says it will do. An example of competence is “This organization has the ability to accomplish what it says it will do” (Hon & J. Grunig, p. 28). Cho (2006), Mayer et al. (1995) and Schoorman et al. (2007) contended that ability (equivalent to competence), integrity, and benevolence are dimensions of a person’s perceptions of another’s trustworthiness and thus antecedents to the amount of trust this person can have in another. Benevolence was referred to as “the extent to which a party is believed to want to do good for the trusting party, aside from an egocentric profit motive” (Schoorman et al., p. 345). Likewise, Cho (2006, 2007) treated benevolence, integrity, and competence as dimensions of trustworthiness, antecedents to customers’ perceptions of trust. The question central to trust then is: are the three sub-dimensions proposed by Hon & J. Grunig (1999) sub-dimensions of trust or actually measures of a party’s trustworthiness? Many researchers on relationship management tend to take Hon and J. Grunig’s scale for granted without clarifying the conceptualization. This study examined the validity of trust in Chapter III to address this question. Distrust In relation to the dimension of trust, there are some new developments in the management literature. Based on Luhmann’s work (1979), Lewicki et al. (1998) were among the first (e.g., Kramer, 1994; Sitkin & Roth, 1993; McKnight & Chervany, 2001) to propose distrust as distinct from trust. One would think that distrust is simply the opposite of trust. To the contrary, Lewicki et al. argued that trust and distrust are
54
“separate but linked dimensions” (p. 439). Trust is not simply the absence of trust. They defined trust “in terms of confident positive expectations regarding another’s conduct” and distrust “in terms of confident negative expectations regarding another’s conduct” (p. 439). More specifically, “another’s conduct” includes “words, actions, and decisions;” such “confident positive expectations” are “a belief in, a propensity to attribute virtuous intentions to, and a willingness to act on the basis of another’s conduct;” (Lewicki et al., p. 439). This is quite similar to Hon and J. Grunig’s definition of trust. Lewicki et al. elaborated that the “negative expectations” involved in distrust are “a fear of, a propensity to attribute sinister intentions to, and a desire to buffer oneself from the effects of another’s conduct” (p. 439). It is possible for relational parties in organizations to have both trust and distrust in one another, analogous to harboring both love and hatred. In this light, low distrust cannot be the same as high trust and high distrust not the same as low trust. To better illustrate their conceptualization, Lewicki et al. produced a 2 by 2 table of high and low trust and high and low distrust (p. 445). I reproduced their table (see Table 1). Based on the two-dimensional framework, Lewicki et al. (1998) proposed four prototypical relationship conditions: low trust/low distrust, high trust/low distrust, low trust/high distrust, and high trust/high distrust. Under the condition of low trust/low distrust, the individual has no reason to be confident or vigilant. One example is casual acquaintances. Under the condition of high trust/low distrust, one is confident and does not suspect the other. The relationship is characterized by increased interdependence, where relational parties try to pursue new common initiatives and opportunities. Under the condition of low trust/high distrust, the person is not confident in but very watchful of
55
his or her relational partner. Relational parties attribute harmful motives to the other, attempt to exploit potential vulnerabilities of the other, and dedicate considerable resources to monitor the other’s behavior. Under the condition of high trust/high distrust, one is confident in another in some aspects but suspicious of another in other aspects. Partners have separate and shared objectives. They trust but also verify the each other’s words, actions, and decisions. They pursue new opportunities but also monitor risks involved. Table 1 Trust and Distrust Prototypical Relationship Conditions High Trust Characterized by hope, faith, confidence, assurance, initiative High-value congruence; Interdependence promoted; Opportunities pursued; New initiatives Trust but verify; Relationships highly segmented & bounded; Opportunities pursued & down-side risks/vulnerabilities continually monitored Low Trust Characterized by no hope, no faith, no confidence, passivity, hesitance Low Distrust Characterized by no fear, absence of skepticism, absence of cynicism, low monitoring, no vigilance Casual acquaintances; Limited interdependence; Bounded, armslength transactions; Professional courtesy Undesirable eventualities expected & feared; Harmful motives assumed; Interdependence managed; Preemption; best offense is a good defense; paranoia High Distrust Characterized by fear, skepticism, cynicism, wariness and watchfulness, vigilance
56
Lewicki et al.’s (1998) proposition is intuitively appealing, particularly considering the relationship an employee has with his or her organization (represented by the management). In the age of corporate scandals, even the most satisfying employee’s relationship with his or her organization may be high trust/high distrust, other than high trust/low distrust. From an organizational point of view, it is important to investigate trust as well as distrust because the strategies to build and maintain trust do not necessarily reduce or eliminate distrust; and outcomes as a result of trust as opposed to distrust may differ. A few researchers have begun to empirically test the distinction between trust and distrust (e.g., Adler, 2007; Benamati & Serva, 2007; Cho, 2006, 2007; H. Huang & Dastmalchian, 2006; Lai & Mollering, 2007). For example, Cho (2006) used Lewicki et al.’s (1998) conceptualization of trust and distrust to test antecedents to and consequences of trust and distrust. She posited that core business operations and relationship investments influence a customer’s judgments of an e-vendor’s trustworthiness (i.e., competence and benevolence), which in turn influence differently consumers’ perceptions of trust and distrust toward the e-vendor. Trust and distrust were expected to have different effects on behavioral intentions (i.e., self-disclosure and relationship commitment). Using newly developed measures of distrust, Cho (2006) found that benevolence can be considered as a “trust-producing factor” that has a more significantly positive effect on trust than its negative influence on distrust, but competence is rather “a distrust-avoidance factor” that has a more significantly negative impact on distrust than a positive effect on trust (p. 32).
57
Further, she concluded that a high level of distrust is instrumental in affecting selfdisclosure and high trust more vital to consumer commitment. Cho (2007) investigated the differing effects of trust and distrust on information privacy concerns within the context of obtaining loan services over the Internet. The Web study of 662 responses indicated that distrust has a greater positive effect on information privacy concerns than the negative impact of trust, moderated by customers’ risk proneness (risk-seeking or risk-averse) and trust propensity. Similar to Cho’s (2006) previous study, findings established that distrust is a concept distinct from trust. In lieu of embracing the new concept of distrust, Schoorman et al. (2007) charged that extant theorizing as well as empirical evidence on distrust is not convincing enough to demonstrate that the concept of distrust is actually theoretically or empirically distinct from trust. They criticized researchers (e.g., McAllister, Pang, Tan, & Ruan, 2006) who argued for this distinction but only reverse scored trust items to measure distrust. Nevertheless, a lack of effort in developing and testing valid measures of distrust does not necessarily “prove” that distrust is not a separate construct. Cho’s (2006, 2007) studies provided evidence of the distinctness of distrust. This study accordingly investigated whether distrust is a new dimension of characteristics of organizationemployee relationships. Commitment Hon and J. Grunig (1999) defined commitment as “the extent to which the public [or the organization] feels that the relationship is worth spending time and energy to maintain” (p. 3). This differs from others’ definitions of relational commitment as a behavioral intention (e.g., E. Anderson & Weitz, 1992; Cho, 2006; Michaels, Acock, &
58
Edwards, 1986; Wulf, Odekerken-Schroder, & Iacobucci, 2001). Michaels et al. (1986) stated that commitment is “one’s desire and intent to maintain, rather than terminate, a relationship” (p. 162). In a similar vein, Cho (2006) referred to it as “a lasting or enduring intention to build and maintain a long-term relationship” (p. 28). Likewise, Canary and Zelley (2000) used this conceptualization—“the extent to which a person wants to remain in the relationship and has feelings of attachment” (p. 308). These definitions of commitment as behavioral intention are conceptually analogous to turnover intentions, which are a criterion of organizational effectiveness (Robbins, 1990), distinct from relationship characteristics. Furthermore, because this study viewed relational characteristics as perceptions, I chose Hon and J. Grunig’s definition rather than others’ that focused on behavioral intention. Marketing researchers and public relations scholars alike have recognized commitment as a multidimensional construct (e.g., J. Brown, Lusch, & Nicholson 1995; Gruen et al., 2000; Gundlach, Achrol, & Mentzer, 1995; Hon & J. Grunig, 1999; Hung, 2001; K. Kim & Frazier,1997; Kumar, Scheer, & Steenkamp 1995). Hon and J. Grunig proposed two sub-dimensions—continuance and affective commitment. They derived these sub-dimensions from Meyer and N. Allen’s (1984) work in which continuance commitment indicates the degree to which a relational party (e.g., an employee) is psychologically bonded to an organization out of concerns for the perceived costs associated with leaving the organization; and affective commitment means emotional attachment. In addition to these two sub-dimensions, Gruen et al. proposed another subdimension of commitment that members have towards their professional associations— normative commitment, which describes the extent to which a member is
59
“psychologically bonded to the organization on the basis of the perceived moral obligation to maintain the relationship with the organization” (p. 37). Their study of member-association relationships confirmed the validity of normative commitment as a sub-dimension of commitment. Such obligation-based commitment makes intuitive sense; for example, an employee of a charity organization or activist organization may feel a sense of moral obligation to stay in a relationship with his or her organization if this employee joined the organization to promote charity or a cause. Although Gruen et al. only focused on member-association relationship, it is intriguing to investigate whether normative commitment can be extended to other contexts. This study explored such a possibility. It is noteworthy that Hon and J. Grunig’s (1999) definition of continuance commitment lacks clarity because they stated that it “refers to a certain line of action,” implying a behavior, which contradicts their conceptualization of commitment and other indicators as perceptions of relationships (p. 25). Such lack of clarity was present in their measures of commitment; items such as “I have no desire to have a relationship with this organization” pertain to behavioral intentions. Accordingly, refinement and addition to the existing measures of commitment were included in this study. Satisfaction Hon and J. Grunig (1999) viewed satisfaction as “the extent to which one party feels favorably toward the other because positive expectations about the relationship are reinforced” (p. 20). From a social exchange perspective, both parties are satisfied with the relationship when they achieve a balance of rewards and costs (Stafford & Canary, 1991). J. Grunig and Y. Huang (2000) observed that satisfaction involves affection and emotion
60
whereas trust and control mutuality consist of cognitive dimensions. Initially used to evaluate intimate relationships (Hendrick, 1988), romantic relationships (Stafford, 2003), and work relationships (Waldron, 2003), satisfaction has been widely used to assess organization-public relationships (e.g., J. Grunig & Y. Huang, 2000; Y. Huang, 2001; Hung, 2001; H. Kim, 2005; Ledingham, 2003; Yang, 2007). Organizational psychologists distinguish job satisfaction and satisfaction with the organization. Job satisfaction indicates an employee’s “subjective evaluation of the degree to which his or her requirements are met by the work environment” (Bretz & Judge, 1994, p. 32). It often has to do with factors such as complexity of the job. Public relations scholars (e.g., D’Aprix, 1996; J. Grunig, 1992; L. Grunig et al.) maintained that the satisfaction with the organization overall is a better indicator of the effect of the organization’s communication efforts because employees can be satisfied with their jobs but not with the organization. In the event of low satisfaction with the organization but high job satisfaction, it will be misleading to use measures of job satisfaction to evaluate satisfaction with the organization. Accordingly, the concept satisfaction in this study does not involve job satisfaction. Control Mutuality Control mutuality represents “the degree to which parities agree on who has rightful power to influence one another” (Hon & J. Grunig, 1999, p. 19). Stafford and Canary (1991) more specifically suggested that such “influence” is over “relational goals and behavioral routines” (p. 224). Similar ideas abound in the literature, such as Ferguson’s (1984) “distribution of power in the relationship,” Aldrich’s (1979) notion of reciprocity, and Kelley’s (1979) bilateral control over the relational partner, Bruning and
61
Ledingham’s (1999) “mutual legitimacy” and Bortree and Waters’ (2007) “balanced power.” Power imbalance seems inevitable in interpersonal, inter-organizational and organization-public relationships. It exists in three forms (J. Grunig & Y. Huang, 2000). First, one party is completely powerless and consequently both parties try not to antagonize each other because of fear of escalated tension. Second, when the power difference is minimum, relational parties engage in intensive power struggle that does not end well for either one. Third, when the power difference is large, a third party can be introduced to help with the less powerful and balance out the difference. Research has shown that excessive attempts to exert control are negatively associated with perceived competence of communicators (Bochner, Kaminski, & Fitzpatrick, 1977) and increase levels of activism (L. Grunig, 1992). On the other hand, power does not have to be completely equally distributed to maintain a stable relationship (L. Grunig et al., 1992). Hon and J. Grunig (1999) noted that as long as organizations and publics exert some control over each other, the relationship will be quite stable and positive. Research has linked attempts to control a relationship unilaterally to relational dissatisfaction and misunderstanding (Canary, Weger, & Stafford, 1991; Courtright, Millar, & Rogers-Millar, 1979). Congruence of Perceived Relationship Characteristics Defining congruence. As I discussed earlier, this study adopted the dialectical approach to relationships, a core assumption of which is contradiction. One needs to understand the interaction between organizations and their publics in order to best capture
62
the essence of organization-public relationships, hence the notion of congruence of perceived relationship characteristics. Congruence, also known as fit, in organizational psychology originated from the idea that individual behavior can only be understood as an interaction between individuals and their organizational context (Ostroff & Schulte, 2007). Based on this interactionist perspective, scholars proposed the notion of person-environment fit or congruence that suggests matching individual characteristics with their corresponding environmental characteristics (e.g., Schneider, 1983, 1987). This matching between persons and organizations occurs when individuals’ personality matches the organizational climate (Christiansen, Villanova, & Mikulay, 1997; A. Ryan & Schmitt, 1996; Tom, 1971), or when their values agree with the organizational culture (e.g., Chatman, 1989; O’Reilly, Chatman, & Caldwell, 1991), or when their goals are consistent with those of organizations (e.g., Vancouver & Schmitt, 1991), or when their needs are congruent with characteristics of the environment (Pervin, 1968), to name just a few. This similarity-based person-organization fit or congruence is built upon Schneider’s (1987) Attraction-Selection-Attrition framework, which posited that similar people tend to be attracted to, selected by, and retained by organizations. However, the caveat is that the resultant homogeneity from similarity-based fit can lead to lack of creativity, diversity, and inability to change (Schneider, Goldstein, & Smith, 1995). In light of the dialectical approach to relationships, I argue that the interactionist perspective on in organizational psychology is consistent with the dialectical assumptions. A dialectical approach to relationship management views organizations and their publics, such as employee publics, as opposing yet interdependent forces (contradiction) whose
63
agenda differ considerably at times. Their contradiction serves as the agent of change. Due to the constant negotiation of interests and goals, change occurs in their relationships. During this process of negotiation, organizations resort to an array of relationship maintenance strategies, which result in different perceptions in employees. Their proactiveness and reactiveness in the decisions and choices organizations and employees make (praxis) are shown in both sides’ perceived relationship quality and employees’ behavior at work. Understanding organization-public relationships as a process of relating exhibits the applicability of totality to relationship management research. Integrating both the dialectical approach and the interaectionist perspective and applying it to organization-public relationships, it is clear that “relationships” is a “relating” process that highlights the interaction between publics (e.g., employees) and their corresponding organizations. That is, the concept of congruence of perceived organization-public relationship characteristics is the key to understanding how organizations and publics (e.g., employees) “relate” to each other. Congruence of perceived organization-public relationship characteristics does not refer to similarity or consensus, but represents the dynamic interaction and process of interdependence between unified opposing relational parties. As Broom et al. (2000) noted, “Relationships are the dynamic results of the exchanges and reciprocity that manifest themselves as the relationships develop and evolve, yet they can be described at a given point in time” (p. 17). At a given point in time, the interaction may result in compatibility and consensus. But it will change (change from the dialectical approach) as the two opposing forces in the relationship continue to negotiate and adapt to each other’s reactions (praxis and totality from the dialectical approach).
64
The matching or fitting that happens between the organization and its employees in this study is give-and-take of perceptions. Here employees’ perceptions are akin to the person variable (individual characteristics), and the organization’s perceptions are the organization variable (aggregate assessment of person’s characteristics to represent organizational characteristics) in the aforementioned person-organization fit research. The reasoning for using aggregate personal characteristics (e.g., perceptions) to represent organizational characteristics is that organizations are made up of people and people’s perceptions constitute the environment (e.g., R. Hogan & Roberts, 2000; Schneider, 1987). According to Edwards (1994, 1995), congruence can be operationally expressed by five variables: the perceptions of employees (one opposition in the dialectical approach) as the person variable, the perceptions of the organization (another opposition in the dialectical approach) as the organization variable, the squared person variable, the squared organization variable, and the product of the person and organization variable. Level of analysis. Person-environment fit researchers assume that individuals can be fit to different levels of the environment, such as supervisor, job, group, and organization (Ostroff & Schulte, 2007). Outcomes of fit are primarily at the individual level, including job satisfaction, employee turnover and employee performance (Ostroff & Schulte). It is necessary for researchers to indicate what they are fitting to and at. Fitting to is related to the level of the environment, whereas fitting at means the level of analysis for the outcomes of fit. One can fit employee characteristics to those of their organizations and analyze the individual-level consequences. Conversely, one can fit employee characteristics to those of co-workers (dyadic) while examining effectiveness outcomes at the group or organizational level. This study was fitting employees’
65
perceptions of organization-employee relationship characteristics to their organizations’ perceptions (represented by aggregation of managers’ perceptions), and focusing on effectiveness outcomes at the individual level. Measurement Issues Related to Congruence of Perceived Relationship Characteristics Extant literature primarily assesses relationships from a one-sided perspective (Broom et al., 1997; Christensen, 2005; J. Grunig & Y. Huang, 2000; Hon & J. Grunig, 1999; Kelly et al., 2006; Seltzer, 2005; Seltzer & Mitrook, 2009; Shin & Cameron, 2005; Waters, 2007). Researchers typically ask publics’ perceptions of their relationships with a given organization. Leading scholars on organization-public relationships such as Hon and J. Grunig and J. Grunig and Y. Huang have long acknowledged the necessity of examining both parties in a relationship. Waters posed the question: “[W]hy does our literature encourage symmetrical communication yet our research is still asymmetrical” (p. 25)? Seltzer used an interpersonal example to illustrate, Albert may like Betty, [and] he may think that she likes him, and … that they have a healthy relationship, but if Betty doesn’t like Albert, then their relationship is not healthy, regardless of Albert’s perceptions. Simply put, the saying that it takes two to tango is directly applicable to public relations relationship measurement. (p. 13) Measuring both relational parties in a relationship is not news in the interpersonal communication literature (e.g., Wilmot, 1980). However, few scholars have made such attempts in public relations research. More recently, a handful of studies employing the coorientation method surfaced in the public relations literature (Christensen, 2005; Kelly et al., 2006; Seltzer, 2005; Seltzer & Mitrook, 2009; Shin & Cameron, 2005; Waters,
66
2007). The coorientation method was initially introduced by J. Grunig and T. Hunt (1984) and Broom and Dozier (1990) into public relations. It consists of four parts: 1) the organization’s view of a relationship; 2) the public’s view of a relationship; 3) the organization’s estimate of the public’s view; 4) and the public’s estimate of the organization’s view. Agreement shows the extent to which the organization’s view matches that of the public’s. Accuracy represents the degree to which the organization’s or the public’s estimate of each other’s view was correct. Perceived agreement is the degree of congruence between the organization’s (or the public’s) view and its perception of the public’s view (or the organization’s). Based on this conceptualization, Broom and Dozier suggested using difference scores or correlation coefficients between two parties’ mean ratings on each questionnaire item. The empirical studies using this method calculate mean scores for each relational dimension for both the organization and the public, then use t-tests, analysis of variance (ANOVA), or multivariate analysis of variance (MANOVA) to see if the mean differences are significant. Significant mean differences represent different levels of accuracy, agreement, and perceived agreement, which are categorized into four states: (a) consensus, (b) dissensus, (c) false consensus, and (d) false conflict (see Figure 1).
67
Organization’s Views of Relationship
Agreement
Public’s Views of Relationship
Perceived Agreement
Accuracy
Perceived Agreement
Organization’s Estimate of Public’s Views of Relationship
Public’s Estimate of Organization’s Views of Relationship
Figure 2. The coorientation model adapted from Broom and Dozier (1990).
68
Though conceptually intriguing, the coorientation method is not without problems. Organizational psychologists Edwards and Parry (1993) and Edwards (1994, 1995) were the first ones to discuss the substantial problems associated with using difference scores as dependent variables as well as independent variables. I first discuss the problems with using difference scores in examining organization-public relationships as a dependent variable herein. Problems with difference scores representing relationships as a dependent variable. As a dependent variable, the use of difference scores to represent agreement, accuracy and congruence between perceived organization-public relationships introduces four major problems (Edwards, 1995). First, as it often happens, the component measures (i.e., two sides’ views and estimates) are positively correlated, which results in difference scores less reliable than their component measures. When one uses components of the difference as separate dependent variables, the reliability tends to be higher. In the case of a decreased reliability by using the difference scores, the standard error of estimate will go up and statistical power will drop (Edwards, 1995; Pedhazur, 1982). Second, the difference-score approach of the coorientation method essentially collapses views of two sides into a single index (differences of mean scores). It “conceal
69
Third, difference scores confound the influence of the independent variables on the components (i.e., each side’s views and estimates). Edwards (1995) illustrated this problem using regression equations (pp. 308-309): Y1 – Y2 = b0 + b1X + e (Equation 1), in which Y1 and Y2 are measures of the two sides, X is an independent variable, b1 is the unstandardized regression coefficient, and e is random disturbance. One can express the difference scores of the two sides independently in two equations: Y1 = b10 + b11X + e1 (Equation 2); and Y2 = b20 + b21X + e2 (Equation 3), from which the first equation can be rewritten as: Y1 – Y2 = (b10 - b20) + (b11 - b21) X + (e1 - e2) (Equation 4) It is clear that estimating equations 2 and 3, one can obtain b1 in equation 1. Yet, estimations based on equation 1 alone cannot yield information on the signs or magnitudes of b11 and b21. When interpreting data, one cannot draw conclusions about the effects of the independent variables on each side of relational parties. It is possible that X has an equal but opposite effect on Y1 and Y2 when b11 = - b21, or an effect on Y1 but not on Y2 when b11 = b1, b21= 0, etc. But relying on equation 1 alone cannot offer such information. Lastly, Edwards (1995) pointed out that using difference scores as a dependent variable “transforms an inherently multivariate model into a univariate model” (p. 310). This again can be shown from the above equations. By jointly estimating equation 2 and 3, one can obtain information on the independent effects of X on Y1 and Y2, the joint effects of X on Y1 and Y2 (using multivariate tests of significance), and correlations
70
between residuals (e1 and e2). Such tests are not possible in a univariate model shown in equation 1. Problems with difference scores representing relationships as an independent variable. Similarly, when algebraic difference indices are used to represent relationships as an independent variable, four major problems emerged (Edwards, 1994). First, algebraic difference indices are difficult to interpret (Edwards). They would appear to represent equal but opposite contributions of each component, i.e., each of the two sides of a relationship3, which could only be assumed in the event that both sides had the same variance. Yet, unequal variances are often the case when multiple subordinating employees are compared with the same managerial employee. Second, algebraic difference indices conceal the relative contribution of the components (each side of a relationship) to the relationship between the index and a dependent variable (Edwards, 1994). Sometimes, it is likely that the relationship between the index and the dependent variable is driven by one of the components only. Third, algebraic difference indices cannot explain more variance than when the components are considered jointly (Edwards, 1994). Graphically, an algebraic difference index is a plane with two components having equal but opposite linear relationships with a dependent variable. It is simply an approximation of the actual surface relating the components to the dependent variable, and hence unable to account for more variance than its components. Lastly, the coorientation method using difference scores to represent relationships as an independent variable usually involves testing whether the coefficient shared by the two components in its regression equation is significantly different from zero.
3
Z = b0 + b1 (X-Y) + e
71
Nevertheless, it neglects to also examine whether the implied constraint is valid, i.e., if the components do share the coefficient (Edwards, 1994). Another test is necessary to show that the two components’ coefficients have opposite signs but do not differ significantly in absolute magnitude, which has not been shown in current research using the coorientation method in public relations. As noted earlier, this study used congruence of perceived organization-employee relationship characteristics as a mediator. Thus, the problems associated with using difference scores to represent relationships as dependent as well as independent variables would be certainly applicable should the coorientation method using the difference-score approach. To avoid the preceding problems, Edwards (1994, 1995) proposed an alternative multivariate procedure. Regarding analyzing the components as dependent variables, multivariate regression analysis is suggested. The two components are retained separately and are tested jointly, enabling researchers to estimate “the effects of each antecedent variable on each component measure as well as the multivariate association between the predictors and component measures as a set” (Ostroff et al., 2004, p. 346). Suppose there is dissensus or disagreement between the two relational parties; this approach can determine the sources of such discrepancy. Specifically, this approach entails first using an omnibus multivariate test (based on the overall Wilks’ ?) to examine whether the relationship between the antecedents and the two sides of organization-employee relationships was significant overall. When the test is significant, one can further examine regression coefficients with the organization’s perceptions and the employees’ perceptions as separate dependent variables.
72
With congruence of perceived organization-employee relationship characteristics as the predictor, Edwards (1994, 1995) suggested using the polynomial regression procedure and surface response tests. This procedure is based on the following three guiding principles. First, the relationship between congruence and its outcomes should be viewed in three dimensions. Perceived organization-employee relationship characteristics have two components (employees vs. organizations), the distinctiveness of which need to be retained in data analysis. Second, the relationship between congruence and a particular outcome can be shown as a three-dimensional response surface, another focus of analysis. Third, any constraints imposed on the data, such as those in the difference-score approach, need to be included in hypotheses and tested (Edwards, 1994). Using this procedure, one needs to regress each outcome variable on the components, the squared components, and the product of two components, and examine their response surfaces. These analytic procedures proposed by Edwards were used in this study. To sum up my discussion, I present the third proposition and second research question on congruence of perceived organization-employee relationship characteristics: Proposition 3: Congruence of employees’ and organizations’ (represented by managers’ perceptions) perceptions of their relationship characteristics best captures the relational dynamics. These relationship characteristics are made up of five components: trust, distrust, control mutuality, satisfaction, and commitment. Research Question 2: How do non-management employees and managers perceive the organization-employee relationships?
73
Organizational Effectiveness Organizational Effectiveness: Definition and Approaches “Organizational effectiveness is the ultimate question in any form of organizational analysis” (Hall, 1980, p. 536). The earliest approach to organizational effectiveness in the 1950s viewed organizational effectiveness as “the degree to which an organization realizes its goals” (Robbins, 1990, p. 49), which is evidently ambiguous. The goals can be either short-term or long-term or both; non-management employees’ goals or top executives’ or both; officially stated goals or actual goals. With the proliferation of organizational effectiveness studies in the next two decades, researchers agreed that organizational effectiveness may have no fixed operational definition because organizations can be evaluated as effective or ineffective on many different criteria that sometimes are independent of each other (Robbins, 1990). Such criteria included productivity, profit, absenteeism, turnover, motivation, morale, and stability, among others (Campbell, 1977). Without having to agree upon one standard definition, organization theorists adopted four major approaches to studying organizational effectiveness, each with a distinct definition (Robbins, 1990; J. Grunig & Y. Huang, 2000; L. Grunig et al., 1992). They are the goal attainment approach, the systems approach, the strategic constituencies approach, and the competing-values approach. The goal attainment approach. According to Robbins (1990), the goal attainment approach posits that an organization is effective when it accomplishes its goals (e.g., profit maximization). It assumes organizations as rational and goal-seeking entities. Other assumptions associated with this approach are:
74
First, organizations must have ultimate goals. Second, these goals must be identified and defined well enough to be understood. Third, these goals must be few enough to be manageable. Fourth, there must be general consensus or agreement on these goals. Finally, progress toward these goals must be measurable. (Robbins, p. 54) This approach is fraught with problems. Specifically, the assumed consensus on goals is hard to find in organizations because of the political nature of organizations (L. Grunig et al., 1992). The dominant coalition may desire goals different from the organization’s external publics, such as community leaders and activist groups. The officially stated goals of the organization may not be the actual goals. These different goals may contradict each other, which presents difficulty in evaluation of organizational effectiveness (Robbins, 1990). The systems approach. The systems approach suggests that organizational effectiveness is appraised by an organization’s ability to acquire resources from the environment that are critical to its survival, sustain itself internally as a social organism, and interact successfully with the external environment. Unlike the goal attainment approach that emphasizes the ends, the systems approach underscores the ends less but the means more (Robbins, 1990). The systems approach has its limitations too, mainly related to measurement and the issue of whether the means to achieve ends matter (Robbins, 1990). Measuring end goals is much easier than assessing process variables, such as “flexibility of response to environmental changes” (Robbins, p. 61). Furthermore, critics questioned whether organizations care about what means are used to achieve the end goals if the end goals
75
are accomplished (Robbins). In addition, the ultimate goal as defined in the systems approach is survival, which is extremely vague (J. Grunig & Y. Huang, 2000). The key concept environment is also vaguely defined. The strategic constituencies approach. The strategic constituencies approach states that organizations are effective when they satisfy the demands of the constituencies in their environment that are crucial for their continued existence (Robbins, 1990; J. Grunig & Y. Huang, 2000; L. Grunig et al., 1992). With an emphasis on the environment similar to the systems approach, the strategic constituencies approach however is not concerned with all elements in the organization’s environment. It is only concerned with elements of the environment that can threaten the organization’s survival. In contrast with the goal-attainment approach that assumes organizations as rational and goal-seeking entities, the strategic constituencies approach regards organizations as political entities wherein multiple interests compete for control over resources (Robbins, 1990). It also assumes that organizations have multiple constituencies in the environment who have varying levels of power and different agenda. Last, the strategic constituencies approach assumes that managers in the organization seek various goals, none of which is value free. Robbins (1990) pointed out three problems with the strategic constituencies approach: 1) difficult to separate the strategic constituencies from the larger environment; 2) hard to further separate the strategic from the “almost” strategic constituencies; 3) challenging to accurately identify the expectations of these strategic constituencies about the organization. Although the strategic constituencies approach did not suggest how to overcome these problems, public relations researchers provided solutions (Dozier et al.,
76
1995; L. Grunig et al., 1992; J. Grunig, 1997). These difficult tasks are precisely the job of public relations managers. They use environmental scanning to “identify emerging or existing publics that could either help or hurt” organizations, who are also known as strategic constituencies (Dozier et al., p. 199). The situational theory also suggests ways to segment publics into all-issue publics, apathetic publics, single-issue publics, and hotissue publics; or into latent publics, aware publics, and active publics in terms of their development stages (J. Grunig, 1997). The competing-values approach. Unlike the systems approach that emphasizes means and goal attainment approach that underscores ends, the competing-values approach focuses on both means and ends (Robbins, 1990; L. Grunig et al., 1992). This approach posits that “the criteria you value and use in assessing an organization’s effectiveness—return on investment, market share, new-product innovation, job security—depend on who you are and the interests you represent” (Robbins, p. 68). This approach assumes the existence of multiple criteria of evaluating organizational effectiveness and the unlikelihood of having the “best” criterion. It also assumes that these different criteria can be organized. Given the multiple criteria of effectiveness and conflicting interests, the management needs to use the criteria of effectiveness espoused by its strategic constituencies if the organization is to survive and prosper (Robbins, 1990). But strategic constituencies change over time. Therefore organizational theorists (e.g., Cameron & Whetten, 1981; Quinn & Cameron, 1983) connected this approach with organizations’ life-cycle stages, stating that organizations should emphasize different strategic constituencies at different stage of its life span. Robbins held that the competing-values
77
approach faces the same problems that the strategic constituencies approach encounters. As discussed previously, the public relations literature provided ways to solve these problems. In sum, the above four approaches all have strengths and problems. It is not necessarily one approach is superior to another. An integration of these approaches led to my conceptualization of organizational effectiveness as the degree to which an organization incorporates the values and demands of its strategic constituencies into its long-term (ends) and short-term (means) goals that it achieves. Employees are certainly strategic constituencies to an organization. Their values and demands should be incorporated in the goals of organizations. Just as Campbell (1977) listed, employees’ turnover and productivity are two important organizational effectiveness criteria. Therefore, I chose contextual performance and contextual performance as the variables to reflect organizational effectiveness. Contextual Performance As a criterion of organizational effectiveness, contextual performance matters markedly to an organization because high levels of activities involved in one’s contextual performance contribute to the accomplishment of organizational goals (Borman & Motowidlo, 1993, 1997; Goodman & Svyantek, 1999; J. Hogan et al., 1998; Motowidlo, 2000; Motowidlo & Van Scotter, 1994; Van Scotter, 2000). Furthermore, contextual performance-related activities tend to be similar across jobs (Borman & Motowidlo, 1997), rendering comparisons between organizations possible. Contextual performance is a distinct component of individuals’ job performance behaviors (e.g., Borman & Motowidlo, 1993, 1997; J. Hogan et al., 1998; Motowidlo &
78
Van Scotter, 1994). Job performance in essence is “the degree to which an individual helps the organization reach its goals” (Motowidlo, Borman, & Schmit, 1997, p. 72). Contextual performance has been defined as individuals’ behavior that “creates and maintains the psychological, social, and organizational environment” in which one’s work is performed (J. Hogan et al., 1998, p. 190). Borman and Motowidlo (1993) introduced the term contextual performance out of concern that research and practice in employee selection largely ignored or downplayed the non-task-related performance by employees, which are often not in one’s formal job descriptions. They emphasized, “[These activities] contribute to the organization’s technical core either directly by implementing a part of its technological process, or indirectly by providing it with needed materials or services” (p. 73). The concept of contextual performance that Borman and Motowidlo (1993) proposed has five components (Borman & Motowidlo, 1997; J. Hogan et al., 1998). They are: 1) persisting with enthusiasm and extra effort to achieve task goals; 2) volunteering to carry out actions that are not formal responsibilities of an employee; 3) helping and cooperating with others; 4) adhering to organizational rules/procedures even when inconvenient for an employee; and 5) supporting organizational objectives (Borman & Motowidlo, p. 102). Typical examples of these components of contextual performance include: extra work on one’s job, suggestions to improve organizational performance, assistance to coworkers, meeting deadlines, and complying with organizational policies. It is noteworthy that, as Motowidlo (2000) explicated, these components are not caused by contextual performance; rather, aggregation of all these behaviors or behavioral episodes by employees is contextual performance.
79
Contextual performance overlaps considerably with organizational citizenship behavior that was introduced by Organ and his colleagues (Bateman & Organ, 1983; Smith, Organ, & Near, 1983). Yet these two constructs have different roots and seek to answer different questions. Organ formally defined organizational citizenship behavior as “individual behavior that is discretionary, not directly or explicitly recognized by the formal reward system, and that in the aggregate promotes the effective functioning of the organization” (p. 4). Measures of organizational citizenship behavior were developed by asking managers what they would want their employees to do but could not require them to do. Central to the definition is “discretionary” and “unrewarded.” In contrast, contextual performance is not necessarily discretionary or nonrewarded (Organ, 1997). As indicated earlier, contextual performance taps what is not covered by task performance. Now it seems these two constructs are clearly different. Organ (1997) would not let the dust settle. He redefined organizational citizenship behavior as “contributions to the maintenance and enhancement of the social and psychological context that supports task performance (or the technical/technological/production system” (p. 91). This definition is virtually the same as contextual performance. Built upon Organ and his colleagues’ work, studies that use the term organizational citizenship behavior typically have a number of similar (if not identical) patterns of behavior included in contextual performance. Therefore, when connecting contextual performance to dimensions of organization-employee relationships, I relied on literature in areas of contextual performance as well as organizational citizenship behavior.
80
Turnover Intention Employees make up organizations. As many management researchers regard employees the most important asset to an organization (e.g., Collins, 2001; C. Pettijohn, L. Pettijohn, & Taylor, 2007; Van Scotter), their intention to leave, i.e., turnover intention is a crucial criterion of organizational effectiveness. Whether an employee intends to stay with or leave an organization shows the extent to which the organization incorporates employees’ demands and values in their goals that it achieves, namely, organizational effectiveness as defined in this study. Also, turnover intention is commonly studied as the ultimate outcome in research on human behavior in organizations (e.g., Brashear, Boles, Bellenger, & Brooks, 2002; Chandrashekaran, McNeilly, Russ, & Marinova, 2000; Z. Chen & Francesco, 2000; Lambert, N. Hogan, & Barton, 2001; Tett & Meyer; C. Pettijohn et al., 2007). Core turnover models in management literature have incorporated turnover intention (behavioral intention); and empirical research has shown turnover intention as one of the strongest predictors of actual turnover behavior (e.g., Mobley, 1977; Hom, Caranikas-Walker, Prusia, & Griffeth, 1992; Steers & Mowday, 1981). Therefore, this study used turnover intention as a variable of organizational effectiveness. Turnover intention has been conceived of as “a conscious and deliberate willfulness to leave the organization” (Tett & Meyer, 1993, p. 262). It is one’s behavioral intention. In terms of measurement, it is often associated with a particular time interval, such as intention to leave within the next six months or within the next year.
81
Model and Hypotheses: Linking the Variables Research on Relationship Maintenance Strategies as Antecedents In public relations research, few researchers have focused on relationship maintenance strategies. Only two studies (i.e., Y. Huang, 1997; Ki & Hon, 2007a) developed a measuring instrument of relationship maintenance strategies to date. A handful of studies examined relationship maintenance strategies mainly through qualitative interviews, case studies, and content analysis of company Web sites (Bortree, 2003; Hong & Kiousis, 2007; Hung, 2002, 2003; Hung & Y. Chen, 2007; Kelleher & Miller, 2006; Ki & Hon, 2003, 2006, 2007a; Wigley, 2003). In a comprehensive content analysis of journal articles on organization-public relationships from 1985 through 2004, Ki & Shin (2005) located only 38 articles from leading communication and management journals, such as Public Relations Review, Journal of Public Relations Research, Journalism & Mass Communication Quarterly, Academy of Management Review, and Journal of Communication Management. They pointed out that 18 articles suggested at least one maintenance strategy. Six of them have been studied in most of these articles: access, positivity, openness, assurances (of legitimacy), networking (i.e., participation in mutual networks), and sharing of tasks (i.e., shared tasks). Networking was the most frequently used strategy. With regard to linking relationship maintenance strategies to perceived organization-public relationship characteristics, Hung (2003) found that the maintenance strategies of positivity and shared tasks were used in achieving all four dimensions of relationship characteristics (trust, satisfaction, control mutuality, and commitment) by multinational companies in China. Similar findings were drawn by Hung and Y. Chen
82
(2007) in a meta-analysis of three studies that they conducted. Likewise, Jo and Shim’s (2005) study on organization-employee relationships indicated that employees’ perceived support from the management (analogous to positivity) who engage in empathetic, meaning-making and direction-giving communication is positively associated with the formation of trusting relationships with the management. Hung (2002) studied how multinational and Taiwanese companies built relationships with their publics in China. She found that symmetrical strategies (assurances of legitimacy and openness) were helpful in achieving integrity, a component of trust; shared tasks (as part of assurances of legitimacy) contributed to gaining competence, another component of trust. Both multinational and Taiwanese companies employed symmetrical strategies (shared tasks as part of assurances of legitimacy and networking) to show commitment to and ensure satisfaction of their publics. Multinational companies utilized symmetrical strategies such as networking and assurances of legitimacy to obtain control mutuality whereas the Taiwanese companies resorted to both symmetrical strategies (e.g., openness, assurances of legitimacy, and shared tasks as part of assurances of legitimacy) and asymmetrical ones (e.g., obligation to others, similar to compromising and accommodating). Specific to employee publics, the excellence study by L. Grunig et al. (2002) found that symmetrical communication with employees makes it more likely that employees are satisfied with their organizations, and consequently less willing to leave the organization (akin to turnover intention). In a structural equation model including symmetrical communication, job satisfaction, participative culture, organic structure, and mechanic structure, L. Grunig et al. found that symmetrical communication was
83
significantly and positively related to job satisfaction. Their measures of symmetrical communication (e.g., “Most communication between administrators and other employees in this organization can be said to be two-way communication”; “This organization encourages differences of opinion”) were similar to those of some relationship maintenance strategies (openness/disclosure and assurances of legitimacy). The indicators of job satisfaction (e.g., “I have found this organization to be a good place to work”) are similar to those of satisfaction, a component of perceived relationship characteristics (“Most people like me are happy in their interactions with this organization”). In addition, as J. Grunig (2006) stated, “the concept of relationship cultivation strategies is the heir to … two-way symmetrical model … Cultivation strategies identify specific ways in which symmetrical communication can be used to cultivate relationships” (p. 168). Taken together, it means that symmetrical maintenance strategies are positively related to perceived organization-employee relationship characteristics. Studies in the interpersonal communication literature and negotiation literature have also found associations between relationship maintenance strategies and components of perceived relationship characteristics (e.g., M. Allen, 1992; Canary & Stafford, 1994; Canary & Zelley, 2000; Guerrero, Eloy, & Wabnik, 1993; PunyanuntCarter, 2008; Putnam, 1990; Stafford, 2003; Stafford et al., 2000; Stafford & Canary, 1991). From the perspective of interpersonal communication, Canary and Stafford (1994) summarized that shared tasks, networking, and assurances of legitimacy (symmetrical relationship maintenance strategies) are positively tied to control mutuality; and that assurances of legitimacy are strongly and positively connected with commitment and
84
trust. Stafford and Canary (1991) found a positive association between individuals’ symmetrical maintenance strategies (openness, assurances of legitimacy, networking, and shared tasks) and relational satisfaction. In a similar vein, Stafford et al. (2000) found that symmetrical strategies such as assurances of legitimacy were primary predictors of control mutuality, satisfaction and commitment. Likewise, Punyanunt-Carter’s study of 198 father-daughter dyads reported that seven relationship maintenance strategies (openness, positivity, assurances of legitimacy, social networks or participation in mutual networks, conflict management, advice, and shared tasks) influenced participants’ perceptions of satisfaction. In addition, Guerrero et al. reported that openness (or disclosure), positivity, assurances (of legitimacy), participation in mutual networks, and shared tasks predict whether a couple stays or leaves a relationship (similar to relational commitment). Organizational theorists such as Putnam (1990) noted that integrative negotiation (a symmetrical strategy, as part of assurances of legitimacy in this study) can cultivate trust. Comparable patterns were found by M. Allen (1992) who examined links between communication, organizational commitment, and perceived organizational support and concluded that perceived support was a mediator between communication and organizational commitment. She defined perceived organizational support as employees’ perceptions of “the extent to which an organization values its employees’ contributions and cares about their well-being,” analogous to a few maintenance strategies explicated above, such as assurances of legitimacy (p. 359). With regard to congruence of perceived relationship characteristics, a related study by Shen and J.-N. Kim (2008) found that symmetrical communication strategies are
85
positively linked to organization-employee congruence on ethical values. Some symmetrical maintenance strategies discussed in foregoing sections are similar to the symmetrical communication strategies that Shen and J.-N. Kim measured, such as advice and disclosure. They defined congruence as compatibility between people’s and organization’s characteristics. To take one step further, one can then propose that symmetrical strategies can be positively associated with organization-employee congruence on their perceptions of relationship characteristics. On a similar note, Weigel and Ballard-Reisch (1999) who studied marriage relationships looked at the influence of relationship maintenance behaviors on joint couple-level perceptions of marital quality. Similar to this study, the authors argued that marriage relationship is more than the sum of the individual perceptions of husbands and wives. As opposed to the traditional view that examines couple relationship from the perspectives of either husbands or wives, Weigel and Ballard-Reisch posited that communication behaviors, such as communicative relationship maintenance behaviors by husbands and wives help bring about couples’ “couple marital quality.” They suggested that measures of spouses’ perceptions can be used to assess the shared couple reality. Their study of 129 married couples indicated that individuals’ use of maintenance behaviors (positivity, openness, assurances of legitimacy, networking, and shared tasks) were positively related to couples’ marital relationship quality (satisfaction, commitment, and love). In particular, wives’ use of maintenance behaviors had a stronger relationship with couples’ marital relationship quality than did the husbands’. Their conceptualization of joint couple relationship quality as shared couple reality is akin to congruence of perceived relationship characteristics in this study. Extending it to organization-employee
86
relationships, this suggests that an organization’s symmetrical (asymmetrical) relationship maintenance strategies can be positively (negatively) related to the congruence of perceived organization-employee relationship quality. To sum up the above discussion, I proposed the following hypothesis: H1: Symmetrical relationship maintenance strategies are positively correlated with congruence of perceived relationship characteristics (H1a), and asymmetrical strategies are negatively associated with congruence of perceived relationship characteristics (H1b). Congruence of Perceived Relationship Characteristics and Organizational Effectiveness Outcomes Organizational behavior and management researchers have extensively examined the effect of organizational commitment on contextual performance and turnover intention (e.g., N. Allen & Meyer, 1990; Brashear et al., 2002; S. Brown & Peterson, 1993; Chang, 1999; Z. Chen & Francesco, 2000; Gruen et al., 2000; Jaros, 1997; Morgan & S. Hunt, 1994; Mowday, Porter, & Steers, 1982; Russ & McNeilly,1995; Shore & Martin, 1989; Tett & Meyer, 1993; Williams & Hazer, 1986). Weiner (1982) stated that commitment acts as “an intervening process, mediating between certain antecedents and behavioral outcomes … [It] can be viewed as a motivational phenomenon” (p. 419). Gruen et al. found that commitment (normative and affective) partially mediates the effect of relationship-building efforts on members’ participation or involvement in their professional associations, analogous to a component of contextual performance. Brashear et al.’s research of 402 salespeople revealed a significant and negative relationship
87
between commitment and turnover intention. Russ and McNeilly established the predictive effect of satisfaction and organizational commitment on turnover intention. Shore and Martin’s (1989) survey of 72 professional staff in 15 departments of a large hospital and 71 bank tellers in 11 branches of a large bank investigated the effects of organizational commitment and job satisfaction on turnover intention and job performance (task performance, not contextual). Results indicated that commitment was more strongly and negatively associated with turnover intention than job satisfaction for the tellers, but not for the hospital staff. Satisfaction was more strongly related to supervisory ratings of performance than commitment for both samples. Psychologists Tett and Meyer (1993) used path analysis in a meta-analysis of 178 independent samples from 155 studies to examine the effects of satisfaction and organizational affective commitment on turnover intention and actual turnover behavior. They found that satisfaction and commitment each has independent effect on turnover intention. Turnover intention mediates the effect of the predictors on actual turnover behaviors. Their conceptualizations of satisfaction and organizational affective commitment coincide with the definitions in this study. They also relied on Meyer and N. Allen’s (1991) three forms of organizational commitment—affective, continuance, and normative commitment. Trust (e.g., Aryee, Budhwar, & Z. Chen, 2002; Brashear et al., 2003) was also shown to have substantial influence on turnover intention, task performance, and contextual performance. Based on an SEM analysis of 179 supervisor-subordinate dyads (179 subordinates and 28 supervisors) from an India-based company, Aryee et al. reported a significant and negative path between workers’ trust in the organization and
88
turnover intention, a significant and positive path between workers’ trust in their supervisors and contextual and task performance. Dimensions such as trust and power were also found to predict contextual performance. Campbell (1990), in a large study on the U.S. Army (N = 4,039), used general cognitive ability, achievement, dependability and adjustment to predict task performance and contextual performance. He found that dependability has a more substantial influence on contextual performance than on task performance. Dependability is a sub-dimension of trust. Likewise, Motowidlo and Van Scotter (1994) found that dependability and dominance have significantly higher correlations with contextual performance than with task performance. A meta-analysis by Organ and K. Ryan (1995) of attitudinal correlates of organizational citizenship behavior revealed that perceived fairness (a sub-dimension of trust), satisfaction, perceived leader supportiveness, and organizational commitment (dimensions of organization-employee relationships) predicted behaviors such as altruism and compliance, which were identified as components of contextual performances (e.g., Borman & Motowidlo, 1993, 1997; Goodman & Svyantek, 1999; Motowidlo et al. 1997). Likewise, scholars including Bateman, Organ (1983) and Smith et al. (1983) and Organ (1997) who were credited for coining the term organizational citizenship behaviors that overlaps with contextual performance, proposed the term because of their belief that employees’ job satisfaction could lead to such behaviors. In addition to the aforementioned studies that showed the predictive effect of dimensions of perceived relationship quality on contextual performance and turnover intention, congruence was found to be a predictor of contextual performance and turnover
89
intentions as well (e.g., Arthur, Bell, Doverspike, & Villado, 2006; Goodman & Svyantek, 1999; Kristof-Brown, Zimmerman, & Johnson, 2005; Saks & Ashforth, 2002; Vancouver & Schmitt, 1991; Van Vianen, 2000). Goodman and Svyantek reported that congruence of perceived organizational culture and the discrepancy between employees’ ideal and perceived actual organizational culture are positively associated with their contextual performance. Kristof-Brown et al.’s comprehensive meta-analysis of literature on personorganization congruence and other types of person-environment congruence indicated positive associations between congruence and contextual performance but negative relationships between congruence and turnover intention. Day and Bedeian’s (1995) study of 206 nursing service employees indicated that organization-employee congruence on conscientiousness positively influenced employees’ job performance, but had a negative impact on employees’ turnover. Saks and Ashforth’s longitudinal study on job search, congruence perceptions, and employment quality followed 113 students prior to their organizational entry and four months after entry. They found a negative association between person-job congruence (example item: “To what extent is the organization a good match for you?”) and intention to quit a job. Finally, the Attraction-SelectionAttrition framework (Schneider, 1983, 1987) posits that individuals whose characteristics are similar to those of an organization are more likely to be selected in the hiring process and choose to stay on the job. Extending this body of literature to congruence of organization-employee relationship characteristics, I hypothesized that it is positively related to employees’ level of contextual performance but negatively tied to their likelihood of turnover intention. Specifically, higher congruence of perceptions of their relationship characteristics by the
90
employees and the organization (as often represented by the management) will be linked to employees’ higher contextual performance (such as helping others, and voluntarily working extra hours) and lower turnover intention. Therefore, I propose the following hypotheses: H2: The level of congruence of perceived relationship characteristics is negatively associated with the likelihood of turnover intention. H3: The level of congruence of perceived relationship characteristics is positively associated with the level of contextual performance. These hypotheses are shown in a structural model (see Figure 2). Summary In summary, after arguing for the need of theory-building on internal relationship management, I extended the dialectical approach to relationship management and integrated existing relationship models in public relations to refine the general theory of relationship management proposed by Ledingham (2003). I critiqued literature from different disciplines on conceptualizations of organization-public relationships, relationship maintenance strategies, congruence of perceived organization-employee relationship characteristics, and behavioral outcomes of organizational effectiveness, and proposed a new three-stage model of organization-employee relationships. I also discussed issues related to measurement of organization-public relationships and suggested a new analytic framework. In order to test relationship management inside organizations, I have developed the following propositions, research questions, and hypotheses in this chapter:
91
Proposition 1: The dynamic and constant tension and change in organizationpublic relationships around negotiated interests and goals lead to benefits of varying degrees for the organization and its publics and better understanding of each other. Proposition 2: Symmetrical relationship maintenance strategies are composed by disclosure or openness, assurances of legitimacy, networking, and advice. Asymmetrical relationship maintenance strategies include distributive negotiation, avoiding, and compromising. Research Question 1: How do organizational members perceive relationship maintenance strategies used by their organizations? Proposition 3: Congruence of employees’ and organizations’ (represented by managers’ perceptions) perceptions of their relationship characteristics best captures the relational dynamics. These relationship characteristics are made up of five components: trust, distrust, control mutuality, satisfaction, and commitment. Research Question 2: How do non-management employees and managers perceive the organization-employee relationships? H1: Symmetrical relationship maintenance strategies are positively correlated with congruence of perceived relationship characteristics (H1a), and asymmetrical strategies are negatively associated with congruence of perceived relationship characteristics (H1b). H2: The level of congruence of perceived relationship characteristics is negatively associated with the likelihood of turnover intention. H3: The level of congruence of perceived relationship characteristics is positively associated with the level of contextual performance.
92
Symmetrical Relationship Maintenance Strategies
?
H2 H1a
?
Congruence of Perceived Relationship Characteristics
Turnover Intention ?
H3
H1b
Asymmetrical Relationship Maintenance Strategies
Contextual Performance
Figure 1. The structural model of internal relationship management theory. In this figure, Maintenance Strategies vary from high to low likelihood of using certain strategies (symmetrical or asymmetrical4). Congruence of Perceived Relationship Characteristics varies from high to low congruence, empirically represented by employees’ perceptions of relationship characteristics (person variable), organizations’ perceptions of relationship characteristics (organization variable), squares of employees’ perceptions of relationship characteristics (squared person variable), squares of organizations’ perceptions of relationship characteristics (squared organization variable), and product of person variable and organization variable. Turnover Intention varies from high to low likelihood of intention to leave an organization. Contextual Performance5 is participants’ evaluation of their own citizenship behaviors, ranging from high to low level of performance.
Symmetrical maintenance strategies are defined as strategies that are geared towards mutual or joint gains of both relational parties whereas the asymmetrical ones are focused on maximization of self-gains. 5 Contextual performance is defined as individuals’ behavior that “creates and maintains the psychological, social, and organizational environment” in which one’s work is performed (Hogan, Rybicki, Motowidlo, & Borman, 1998, p. 190). It complements task-related performance.
4
93
Examining these propositions, research questions, and hypotheses helped develop a theory of internal relationship management, moving the general theory of relationship management further. It not only provided a newer conceptualization of organizationpublic relationships and a fuller understanding of relationship management inside organizations, but also demonstrated the contributions public relations function makes to organizational effectiveness — the ultimate goal for which organizations build relationships with their publics.
94
Chapter III Methodology The survey method was used in this dissertation. This chapter discusses the choice of this method, sampling strategy, the data collection procedure, and data analysis techniques. Survey Fink (2006) defined surveys as “information collection methods used to describe, compare, or explain individual and societal knowledge, feelings, values, preferences and behavior” (p. 1). It is appropriate for both exploratory and explanatory purposes. Surveys can help “determine the incidence, distribution, and interrelations among …variables, and…the vital facts of people, and their beliefs, opinions, attitudes, motivations and behavior” (Kerlinger & Lee, 2000, p. 600). Respondents’ beliefs, opinions, attitudes, motivations, and behaviors are central to this study, such as perceptions of relationship maintenance strategies and relationship characteristics, and self-reported contextual performance behavior. Why Survey? Randomized experimental research designs are preferred if one wants to maximize internal validity and infer causality from the association between variables (Hoyle, Harris, & Judd, 2002). However, not all variables are manipulable (Hoyle et al.). As a result, survey method6 is chosen to assess people’s attitudes, beliefs, and behaviors, and draw causal inferences about the sources of such attitudes, beliefs, and behaviors (Weisberg, Krosnick, & Bowen, 1996). Statistical methods for multivariate data analysis,
A survey can be a true randomized experiment. Here the term survey method does not involve manipulation of independent variables and random assignment of participants into experimental conditions.
6
95
such as structural equation modeling (SEM), are helpful to test adequacy of one’s proposed models and to draw causal inferences. In the case of this study, manipulating independent variables such as the organizations’ relationship maintenance strategies and randomly assigning employees to different levels of the independent variables was not logistically realistic and possible. Thus the survey method was a better choice than randomized experiments. On the other hand, surveys have a number of weaknesses. First, survey research often gathers self-reported information. Social desirability biases can be a problem, which compromises validity (Weisberg et al., 1996). This occurs for example when sensitive topics are touched upon. Second, the typical format of close-ended questions in surveys makes it hard to contextualize respondents’ answers (Krosnick, 1999). Third, when data are collected face to face or over the telephone by researchers, interviewer effects create biased data. Specific problems include “reading errors, speech variations, improper probes, and unprogrammed feedback to the respondent” (Weisberg et al., p. 118). Lastly, surveys with low response rates have been regarded as non-representative, i.e., lacking external validity. However, research found that surveys with lower response rates can be more accurate than those with higher response rates (see Visser, Krosnick, Marquette, & Curtin, 1996). Strengths and Weaknesses of Online Surveys Surveys can be conducted in person, over the phone, via mail or email, or teleconference (Fink, 2006). Compared with mailed questionnaires, telephone surveys, and face-to-face surveys, the biggest advantages of electronic surveys are convenience and efficiency. Online surveys save more time and money than telephone surveys and
96
face-to-face surveys in terms of survey distribution, data collection, respondent reminding, and data entry, among others (Fink; Schmidt, 1997). Participants can choose to fill out questionnaires at their convenience and keep personal records (Neuman, 1994). This method allows researchers to obtain data instantly and be downloaded to a spreadsheet (Fink, 2006). In addition, it has the “ability to make complex skip pattern questions invisible to the respondent” (Fink, p. 9). Lastly, the dynamic features of online surveys can increase participant motivation, because participants can receive feedback right after filling out information (Schmidt). However, online surveys have a few disadvantages compared with other survey modes, including mailed questionnaires, telephone surveys, and face-to-face surveys. First and foremost, its advantage of convenience can become a disadvantage. The researcher can have little control over the conditions under which an electronic survey is completed, which may result in invalid and unreliable data. For example, questionnaires may not look the same on different monitors. Technology may be unstable and create difficulties for respondents to complete surveys. Respondents may have different levels of computer expertise (Fink, 2006). In a related vein, a common problem with online surveys is incomplete answers (Schmidt, 1997). Respondents may overlook certain questions or simply do not want to provide answers, which is hard to determine unless researchers force explanations in their questionnaire design. When incomplete answers only occur in certain demographics, analysis without taking into account this will lead to biased findings. Third, online surveys are only limited to respondents who know how to use a browser and have ready access to the Internet. Research has shown that typical Web users
97
have higher socio-economic and education status than the general population (Schmidt, 1997). Measures should be taken by researchers to control the impact of such biases on research findings. For example, researchers may screen participants to meet certain demographics criteria or use demographics variables as control variables (Schmidt). Nevertheless, considering the fact that face-to-face or telephone surveys might make participants hesitant to answer certain questions that they view as sensitive, I decided to choose electronic surveys for the ensured confidentiality and anonymity that can make participants feel more at ease. The questions on organization-employee relationships, turnover intentions, and contextual performance can be sensitive to many respondents. Also, the missing-data problem is present in mailed questionnaires. Online surveys technology enables researchers to remind respondents when they forget to answer questions. Combined with the efficiency and convenience of the method, I chose online surveys for data collection. Sampling The Appropriateness of Non-Probability Sampling Probability sampling is often recognized as superior to non-probability sampling; however non-probability sampling is commonly used because it is not as expensive and/or statistical generalizability is not necessary (Hoyle et al., 2002). For example, when one obtains a purposive sample of foods from a well-known cuisine, the purpose is to understand the variety of elements available in the population, but not to estimate population value—statistical generalizability being unnecessary. In addition, a probability sample is not necessary for making theoretical generalizations. As Shapiro (2003) argued, maximizing the ability to generalize means
98
more than probability sampling. A study should be examined for its contributions to theoretical development and its choice of theoretically driven variables. Provided that theoretical generalization was the focus of this study, the non-probability sampling method was deemed acceptable. Sample Size Estimation and Procedures Given that I used 60 items for the four constructs, the hypothesized model (see Figure 2) has degrees of freedom of 1,830, using the counting rule—([p*(p+1)/2] – total number of parameters to be estimated]). If I were to use composites to represent the indicators, then symmetrical relationship maintenance strategies had four composites, asymmetrical ones had three, congruence of perceived relationship characteristics had five, turnover intention had one, and contextual performance had five. Consequently, the degrees of freedom would become [18* (18+1)/2] -40 parameters to be estimated = 131. Based on Hancock’s (2006a) procedure of a priori power analysis for testing data-model fit as a whole, with a df =131, ?1 = .02, to have a level of power ? = .80 for the ? = .05level test associated with the root mean square error of approximation, the estimated sample size would be approximately 145. In other words, given that the true level of datamodel fit in the population is ?1 = .02, to have a .80 probability of rejecting ?0 ? .05 in favor of acceptable data-model fit using an ? = .05-level test, this study would need a minimum of 145 participants (see Hancock, 2006a, for a detailed discussion of arriving at this estimated sample size). However, Hancock further added a general rule: When ?1 = .02 and models have df ? 60, to achieve power of ? = .80 sample sizes of n = 300 appear to suffice for testing overall data-model fit … For models with
99
df ? 30 when ?1 = .02, sample sizes of around n = 500 appear to be adequate. (p. 103) Therefore, I decided on an estimated minimum sample size of 500. As elaborated in Chapter II, I used managers’ perceptions of relationship characteristics to represent the organization side of an organization-employee relationship, i.e., the organization variable. To ensure enough variance on the organization variable, I needed multiple managers within a single organization and multiple organizations. In terms of the actual number of organizations and number of participants (non-management employees plus managers), past research has suggested 30 organizations with 30 individuals in each organization or 150 organizations with 5 individuals in each organization (Hofmann, 1997) or an average of 10 employees in 113 bank branches (Ostroff, Shin, & Kinicki, 2005). Taking into consideration the minimum number of participants for the overall model, I successfully recruited on average 26.17 participants within a single organization (with a 1: 3 ratio between the managerial and non-management employees) and 30 organizations. I tried to recruit participants from different departments (within each organization) and have organizations of different sizes from different industries to ensure variance on the person variable (employees’ perceptions of relationship characteristics) as well as the organization variable (organizations’ perceptions of relationship characteristics). Pilot Studies Pilot Study 1 Sample and procedures. The objective of this pilot study was to test if distrust was a new dimension of perceived organization-employee relationship characteristics
100
distinct from trust. One hundred and forty-nine students from a large university in Northern China were recruited to complete paper-and-pencil questionnaires with 7-point Likert-type items on trust, distrust, and symmetrical communication strategies. Participants were between 18 and 22 years of age, and were predominantly female (81.9 %). Eighty were freshman, 61 were sophomore, and eight did not report their year in school. Five items on distrust were adapted and revised from Cho’s (2006) items. Six items on trust from Hon and J. Grunig (1999) and eight items on symmetrical communication strategies from L. Grunig et al. (2002) were used. Based on the meaning of these items, I wrote a Chinese version (not literal translation). The Chinese version of items was sent to two Chinese public relations professors and an English studies professor in China. Revisions were made based on their comments. The items were back translated into English to ensure their comparability. Results. The reliability of the trust, distrust, and symmetrical communication items were obtained (Cronbach ? = .863 for trust, .827 for distrust, and .600 for symmetrical communication). I conducted exploratory factor analysis using principal component analysis with Varimax rotation. Four factors with eigenvalues greater than 1 emerged, with trust items loading highly on one factor, distrust on another, and symmetrical communication on two factors (eigenvalue = 10.45, 65.337% variance explained). I then did exploratory factor analysis using principal component analysis with Varimax rotation on trust and distrust items only. Trust and distrust items loaded on separate factors (with the exception of one item of distrust “I am skeptical about whether the university will keep the students’ interests in mind when it makes decisions”)
101
(eigenvalue = 6.80, 61.786% variance explained). This finding showed preliminarily that distrust and trust may be different factors. I then conducted confirmatory factor analysis using the EQS 6.1 Program (Bentler, 2005) for trust and distrust as a two-factor oblique model (because conceptually these two factors are linked but distinct). To evaluate data-model fit, I used the oft-cited Hu and Bentler (1999) joint-criteria approach. According to the joint-criteria, a model is considered tenable when it achieves Non-Normed Fit Index (NNFI) ? .96, Comparative Fit Index (CFI) ? .96, and Standardized Root Mean Squared Residual (SRMR) ? .09 or Root Mean Squared Error of Approximation (RMSEA) ? .06 and SRMR ? .09. The model chi-square = 14.81 (df = 15, p = .47 > .05). NNFI = 1.00 > .96, CFI = 1.00 > .96, SRMR= .03 < .09. According to Hu and Bentler’s (1999) joint criteria of datamodel fit, the two-factor oblique model has excellent fit with the data. I tentatively concluded that distrust is a new dimension of perceived organization-employee relationship characteristics, linked to but distinct from trust. The English version of the distrust items developed in this pilot study was used in the dissertation study. Focusing on a different sample and in a different country, the dissertation study helped crossvalidate these items by examining its reliability and validity. I discuss measurement validity and reliability in Chapter IV. Pilot Study 2 Sample and procedures. The objective of this study was to have a final revision of the questionnaire items in terms of wording and clarity of instructions. I recruited 10 working professionals in both managerial and non-managerial positions from 10 different organizations to complete a draft questionnaire with 93 questions in March 2009. Two of
102
them filled out the questionnaire at home and emailed me revision suggestions and reported questionnaire completion time. A focus group was conducted with the remaining eight participants. Participants filled out the questionnaire and recorded the completion time. A two-hour discussion followed of each item’s meaning. Based on these results, I revised and shortened the questionnaire to 66 questions, including six questions on demographics (see Appendix A). Measures This section describes the final questionnaire (see Appendix A). Each question was in a five-point Likert format, with “totally disagree” being 1, “neutral being” 3, and “totally agree” being 5. I present items on the antecedent variables first, then on the mediating variable, and the behavioral outcomes of organizational effectiveness (see Figure 2 for the proposed model). As different reliability tests were conducted, the reliability values of all measures are reported in Chapter IV. Antecedents: Relationship Maintenance Strategies As is shown in Figure 2, symmetrical and asymmetrical relationship maintenance strategies are antecedents in the model. I discuss the measures for symmetrical and asymmetrical strategies herein. Symmetrical maintenance strategies. Twelve questions assessed four symmetrical maintenance strategies—disclosure, assurance of legitimacy, networking, and advice. Ki and Hon’s (2007a) items on disclosure, assurances of legitimacy, and networking were modified and used. Sample items included “The organization’s annual report is a valuable source of information for employees about what it has done (disclosure);” “The organization makes a genuine effort to provide personal responses to employees’
103
concerns (assurances of legitimacy);” “The organization builds coalitions with groups that help employees’ career growth (networking).” Advice items were adapted from Stafford (2000). An example was “The organization tells employees like me what to do about problems on the job” (advice). Asymmetrical maintenance strategies. Ten questions measured three asymmetrical maintenance strategies—distributive negotiation, avoiding, and compromising. Distributive negotiation items were revised and adapted from Beersma and De Dreu (2002), De Dreu, Evers, Beersma, Kluwer, and Nauta (2001) and Y. Huang (1997a). An example was “When there is a conflict between employees and the management, the management applies pressure so that employees may make concessions.” Avoiding items were revised and adapted from Hess (2003), De Dreu et al. and Y. Huang. A sample item was “When there is a conflict between employees and the management, the management gives as little attention as possible to it.” Items on compromising were adapted from De Dreu et al. A sample item was “When there is a conflict between employees and the management, the management tries to realize a middle-of-the-road solution with employees.” Mediator: Congruence of Perceived Organization-Employee Relationship Characteristics Congruence was measured by using both organizations’ and employees’ responses on the five dimensions of perceived organization-employee relationship characteristics: trust, distrust, control mutuality, commitment, and satisfaction. I present example measures of each dimension in this section.
104
Trust. I revised six items from Hon and J. Grunig (1999). An example was “I am willing to let the organization make decisions for employees like me.” Distrust. Four items developed from Pilot Study 1 were used. An example was “I feel that this organization will exploit employees’ vulnerability given the chance.” Control mutuality. Four items developed by Hon and J. Grunig (1999) were used. An example was “This organization believes the opinions of employees like me are legitimate.” Commitment. Seven items created by Allen and Meyer (1990) and one item by Hon and J. Grunig (1999) were revised and used. Example items were “I do not feel like ‘part of the family’ at my organization (reverse)” and “I feel that I have too few options to consider leaving this organization.” Satisfaction. Four items by Hon and J. Grunig (1999) were used. An example was “Most employees like me are happy in their interactions with this organization.” Dependent Variables Two variables were used to assess organizational effectiveness: turnover intention and contextual performance. I present sample items on each variable below. Turnover intention. Two items developed by O’Reilly, Chatman, and Caldwell (1991) and two items by Chen and Francesco (2000) were modified and used. A sample item was “I may leave this organization and work for another one in the next year.” Contextual performance. I pooled and modified eight items from Borman and Motowidlo (1993), Goodman and Svyantek (1999), and Smith et al. (1983). An example was “I often help other employees with their work when they have been absent.”
105
Data Collection Procedures Data collection was conducted in two phases from March 18 to June 3, 2009. First, I searched the Hoover’s Company Information Database for listed organizations with more than 200 employees and contact information of their human resources, public relations, and corporate communications personnel in Virginia, Maryland, West Virginia and the District of Columbia. I then randomly selected 150 organizations out of the total of 946 organizations, and called these organizations from March 18 to mid-April. Seven organizations agreed to participate. Only 2 organizations were included in the data analyses because there were fewer than 10 participants from each of the other 5 participating organizations. Second, I reached out to friends, acquaintances, and undergraduate students to recruit organizations between March and early June. Each of the 21 undergraduate students was paid 100 dollars for every organization they successfully recruited. They recruited organizations where they did or still were working at and where their parents and relatives were chief executives. I used a template email for each organization that I myself, my friends, and undergraduate students contacted (see Appendix B). Once the organization agreed to participate, they received a link in email. They were directed to Surveymonkey.com to read the informed consent form. Once they agreed, they would click on the “Yes, I would like to participate” button to start the questionnaire. But if they withdrew, they clicked on “No, thanks, I do not want to participate” and exited the Web site. I created a separate questionnaire (same questions, different links) for each organization to keep track of the number of responses per organization. Reminder emails
106
with the link were sent out each week until at least 25 participants were recruited from each organization. Data collected were confidential and anonymous, except that the organization names were known to the researcher. Their Web Internet Protocol (IP) addresses were blocked through Surveymonkey’s service to ensure anonymity. Finally, 48 organizations participated in the study, but 18 organizations were excluded from data analyses because there were fewer than 10 participants from each of these organizations. During the recruitment process, organizations were promised that they would have a summary report of the data if they wanted to, and participants were informed of a raffle drawing with the prize being four American Express gift cards ($50 worth each). To ensure anonymity, I included the following statement and the raffle link at the very end of the questionnaire. Thank you for your participation! To express our gratitude, we give away 4 gift cards (50 dollars value each). If you would like us to enter your name in the raffle for the gift cards, please click on (or Copy and Paste) the following link to enter a separate 1-question survey where you can put down your name and email. This way we can protect the anonymity and confidentiality of information you entered on this survey. A total of 263 participants entered their names and emails in the raffle. Even though some of these participants were from organizations that were excluded in the data analyses, I still included them in the raffle and four participants won the gift cards.
107
Data Analysis Data Aggregation As this study aimed to use managers’ perceptions to represent the organization side of employee-organization relationships, it was necessary to justify the aggregation of individual-level data in the first place. To do so, one should establish within-group agreement, or “greater than chance similarity” (Bliese, 2000, p. 350). I calculated rwg, ICC (1) and ICC (2). The index rwg compares an observed group variance to an expected random variance. ICC (1) assesses either non-independence, i.e., the extent to which variance in a dependent variable is because of group membership; or reliability of a measure—whether aggregation is justified (Kenny & Judd, 1986). When ICC (1) is calculated on independent variables, it is considered a measure of reliability. When it is calculated on dependent variables, it measures non-independence (Bliese). ICC (2) shows the reliability of group means (Bartko, 1976). Statistical Procedures for Data Analyses First, I recoded three four reverse coded items. As a few participants skipped questions, I then conducted missing value analysis (MVA) using the SPSS 16.0 Program. Next, I used the following statistical procedures to test the three propositions, two research questions, and three hypotheses. As the propositions and research questions pertain to construct validity and reliability, I performed exploratory factor analysis (EFA), Cronbach’s alpha reliability test and then confirmatory factor analysis (CFA) after removing items with low loadings for a given factor or significant cross-loadings; then conducted the Cronbach’s alpha reliability test. CFA compared different models to further examine these structures suggested by EFA.
108
To test the three hypotheses, I used multivariate regression to test H1 and polynomial regression procedures followed with surface response tests to examine H2 and H3. Distinct from a series of univariate regression analyses, multivariate regression examines the relationships among the independent variables and among the dependent variables. It allows for multivariate tests of the variance explained by all the equations jointly and provides estimates of the coefficients across equations (see Edwards, 1995 for more details of this procedure). In this study, I used the person variable and organization variable as the dependent variables jointly and the symmetrical and asymmetrical relationship maintenance strategies as the predictors. I examined the overall Wilks’ ?, a multivariate test of the two equations (i.e., for employees’ perceptions and for organizations’ perceptions) jointly. The coefficients’ direction and significance were examined next. Then, I checked whether each predictor had an equal but opposite effect on the person variable and the organization variable, a constraint that is implied but not tested in the difference-score approach. With respect to polynomial regression, I regressed each outcome variable on the mediator, represented by employees’ perceptions of relationship quality (person variable), managers’ perceptions of relationship quality (organization variable), squared person variable, squared organization variable, and the product of person variable and organization variable to specifically examine the links between the mediator and the two dependent variables (see Edwards, 1994; Edwards & Parry, 1993 for discussions of this procedure). The measures of perceived organization-employee relationships were centered on the midpoint of their scales to “reduce multicollinearity and allow for more meaningful interpretation of congruence relationships” (Ostroff et al., 2005, p. 603). Next,
109
the estimated response surfaces were examined to fully understand the relationships between the mediator and the outcome variables. Lastly, I used the four-step approach by Baron and Kenny (1986) to examine the mediation effect in the overall model (see Figure 2).
110
Chapter IV Results This study first used MVA to examine patterns of missing values, and conducted aggregation analysis to justify using managerial respondents’ data to represent the organization side of organization-public relationships. Next, descriptive statistics were obtained to understand the sampled participants and organizations. Also, reliability tests, exploratory factor analysis, and confirmatory factor analysis were performed to assess measurement validity and reliability, the research questions, and the last two propositions. Based on the discussion in Chapter II on the advantages of polynomial regression, surface response tests, and multivariate regression in examining both sides of organization-public relationships, I used these analytic procedures to test the three hypotheses. Lastly, I used the Baron and Kenny (1986) four-step approach to test the mediation effect, yielding insights into the first general proposition. In this chapter, I first discuss the preliminary analysis (MVA and aggregation analysis), then findings related to the research questions, last two propositions, as well as measurement reliability and validity, and lastly results on the hypotheses and overall model. Description of Participants Forty-eight organizations participated in this study, 18 of which were excluded from data analyses as there were fewer than 10 participants in each of these organizations. In total, 1293 participants agreed to participate, and 785 questionnaires with under 5% or fewer than 3 unanswered questions were deemed usable (see Hair, Black, Babin, Anderson, & Tatham, 2006 for discussion on handling missing values). The response rate was 61%, above the desirable benchmark of 50% suggested by Babbie (1992). Table 2
111
shows descriptions of the 30 participating organizations. Table 3 summarizes demographic information of participants. Table 2 Descriptions of Participating Organizations Org No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Industry/Type NGO, justice Health care Government NGO, environment Recreational services Market research Market research Printing Financial consulting Food Recreational services Food Government Recreational services Energy Media Manufacturing Participants no. 26 26* 25 21 28 20 34 28* 34 26 25 27 27 80* 28 23 27 Managerial 5 7 7 4 7 2 10 8 9 4 9 6 7 12 8 9 5 Non-Managerial 21 18 18 17 21 18 24 18 25 22 16 21 20 67 20 14 22
112
Table 2 (continued) Org No. 18 19 20 21 22 23 24 25 26 27 28 29 30 Industry/Type Education Aerospace Food Computer software Construction Contract research Publishing NGO, health Education Telecommunications Food Government Health care Participants no. 30 21 29 29 27 27* 28 16* 15 18 15 13 12* Managerial 6 6 5 6 5 6 5 8 6 9 4 3 7 Non-Managerial 24 15 24 23 22 20 23 7 9 9 11 10 4
Note. Total participants with asterisks have 1 or 2 missing values. Participants did not indicate their position in their respective organizations. Total missing = 7.
113
Table 3 Demographic Information of Participants (N = 785*) Variable Managers Frequency Male Female Sex Total Caucasian American African American Latin American Native American Pacific Islander Asian American Other Ethnicity Total Bachelor Master’s Doctorate Other Education Total Total* 79 114 193 (2 missing) 144 13 12 0 0 13 11 193 (2 missing) 90 57 11 32 190 (5 missing) 195 Percent (%) 40.5 58.5 99.0 73.8 6.7 6.2 0 0 6.7 5.6 99.0 46.2 29.2 5.6 16.4 97.4 24.8 Non-Managers Frequency 226 354 580 (3 missing) 387 65 29 3 4 56 32 576 (7 missing) 315 77 7 169 568 (15 missing) 583 Percent (%) 38.8 60.7 99.5 66.4 11.1 5.0 .5 .7 9.6 5.5 98.8 54.0 13.2 1.2 29.0 97.4 74.3
Note. 7 participants failed to report their position levels. Above demographic was based on 778 participants’ data, although the total N was 785.
114
As is shown in Table 2, 30 participating organizations were in a variety of industries and of different types, such as health care, recreational services, printing, financial consulting, energy, construction, education, and aerospace. Also, there were multiple managerial and non-managerial participants in each organization. Managerial employees numbered from 3 to 12, and non-managerial employees varied from 4 to 67. The ratio of managers to non-managers was 1:3. The average size of the organizations was 26.17. Such purposeful arrangements helped ensure enough variance on the organization variable (organizations’ perceptions of relationship characteristics) and person variable (employees’ perceptions of relationship characteristics) of congruence of perceived relationship characteristics. Table 2 and Table 3 indicated that the majority of respondents was female, Caucasian American, and holds a bachelor’s degree, regardless of their position levels. On average, respondents were 32 years old and had 4 years of tenure in their organizations. Most participants were younger than 45 (84.6%). A comparison of the descriptive statistics of managerial and non-managerial employees revealed that management employees on average were older and had a longer tenure at their organizations (Mage = 39, Mtenure = 7) than the non-management employees (Mage = 29, Mtenure = 3). Managers were more likely to be Caucasian American and have higher education levels than non-managers. Missing Values Analysis To minimize the impact of missing values, I first excluded cases that had more than 3 to 5 % unanswered questions. Still there were cases with missing data. To determine the extent and patterns of missing data, a rule of thumb is that missing data
115
under 10% for an individual observation can generally be ignored, unless the missing data occur in a nonrandom pattern (Hair et al., 2006). Also, the number of cases with no missing data ought to be sufficient for statistical tests to be performed if no substitution of missing values is planned (Hair et al.). Accordingly I used the SPSS 16.0 program to conduct an MVA on the final data set (N = 785) and found that the percent of missing values ranged from 0 to 3.1% (education being the variable with maximum missing data, i.e., 3.1%). Furthermore, I performed the Little’s Missing Completely at Random (MCAR) test and found that missing values occurred in a completely random fashion (?2 = 6482.34, df = 6308, p = .061 > 0.5). Even though the missing data occurred completely at random, they may cause serious problems. To begin with, most statistical procedures by default exclude cases with missing data from analysis, which leads to sample size reduction. Second, missing data can lead to misleading results and inferences (SPSS Inc., n.d.). The expectationmaximization (EM) method is the best way to deal with these problems. It generates estimates that are closest to the parameter values (see SPSS Inc., 2007 for demonstration example comparing the listwise, all-values estimate, and EM method). This method also introduces the least biases while providing the best representation of original distribution of values, compared with the other imputation methods, such as mean substitution, regression imputation, and all-available data (Hair et al., 2006). Therefore, I proceeded with the EM method of imputation.
116
Data Aggregation As discussed in Chapter II, managers’ data were to be aggregated to represent the organization side of organization-employee relationships. To justify aggregation, I calculated rwg, an estimator of within-group agreement. It ranges from 0 to 1, with a larger value indicating higher agreement. An rwg of 0.70 or above is considered acceptable (Klein & Kozlowski, 2000). For multi-item scales, James, Demaree, and Wolf (1984) proposed the following equation:
rwg ( J ) =
2 J [1- ( sxj / ? EU )] 2 2 J [1- ( sxj / ? EU )] + ( sxj / ? EU )] 2 2
2
where rwg ( J ) is the index of within-group agreement based on J parallel items on a
2 is the variable; sxj is the mean of the observed variance on the J parallel items; ? EU
2
variance on xj that would be expected if all cases on xj were due exclusively to random
2 measurement error. For scales using the 5-point Likert format such as in this study, ? EU
is equal to 2.0 (James, Demaree, & Wolf, 1993). The SPSS 16.0 program was used to calculate the rwg values. Because the aggregation involved only variables of congruence of perceived relationship characteristics, I present the rwg values for its five dimensions (trust, distrust, control mutuality, commitment, and satisfaction) in Table 4. The average rwg values were all above .80, and median rwg values were all above .85, sufficiently large to justify aggregation.
117
Table 4 Within-Group Agreement (rwg) Variable Trust Distrust Control Mutuality Commitment Satisfaction No. of items 6 4 4 8 4 Minimum rwg .60 .67 .41 .00* .50 Maximum rwg Average rwg 1.00 1.00 1.00 .98 1.00 .90 .86 .87 .85 .92 Median rwg .92 .85 .90 .89 .91
Note. A negative rwg value was recoded to 0 (James et al., 1984). In addition, I calculated ICC (1) and ICC (2). As previously stated, ICC (1) assesses non-independence when being calculated on dependent variables and measures reliability when being calculated on independent variables (Bliese, 2000). ICC (2) estimates reliability of group mean (Bartko, 1976). Both indices are recommended as criteria of aggregation (Bliese). Bartko suggested a formula for calculating ICC (1) from a one-way random-effects analysis of variance (ANOVA) model:
ICC (1) = MSB ? MSW MSB+ ( K ? 1) * MSW
Where MSB is the between-group mean square, MSW is the within-group mean square, and k is the group size. If group sizes differ, average group size for k is used. In this study, the average group size was 26.17. The range of ICC (1) calculated from the ANOVA model is -1 to +1. Negative values mean that the between-group variance is smaller than the within-group variance. According to Bliese, a large non-zero ICC (1) estimate suggests that group membership largely affects individual responses. If non-
118
independence is present, it should be accounted for to avoid biases caused in standard errors in further ordinary least squares analyses, such as ANOVA and regression. Similarly, from a one-way random-effects ANOVA model, ICC (2) can be calculated with the following formula (Bliese, 2000): ICC (2) = MSB ? MSW MSB
I obtained mean squares by conducting one-way ANOVA analyses with organization membership as the predictor and manager scores on the five dimensions of perceived relationship characteristics as the dependent variables. ICC (1) and ICC (2) values were calculated from the mean squares. The ICC (1) and ICC (2) estimates are presented in Table 5-I. Because I calculated ICC (1) on dependent variables, it assessed non-independence. It can be interpreted as the percent of variance in individuals accounted for by organization membership. The ICC (1) values in this study were low, ranging from .04 to .11. This indicated the influences of organization membership on managers’ responses, which is of small magnitude in this case. In contrast, the ICC (2) values were much greater. They showed that the group/organization means were relatively reliable, mostly above or near .70. In summary, based on the rwg, ICC (1), and ICC (2) values, I concluded that managerial employees’ data could be aggregated to represent their organizations’ perceptions of relationship characteristics.
119
Table 5-I ICC (1) and ICC (2) Results for Managers’ Data (n = 195) Variable Trust Distrust Control Mutuality Commitment Satisfaction No. of items 6 4 4 8 4 MSB 1.72 1.68 .44 1.50 1.41 MSW .42 .54 .22 .41 .36 ICC (1) .11 .07 .04 .09 .10 ICC (2) .76 .68 .49 .72 .74
In addition to the analysis on managers’ data, I calculated the ICC (1) values based on employee data to assess the extent to which employees’ data were nonindependent, which could result in biased standard errors in ordinary least squares analysis (G. Hancock, personal communication, July 16, 2009). The calculations were based on one-way ANOVA analyses with organization membership as the predictor and employee scores on all the dependent variables as the outcomes. All the ICC (1) estimates were low, ranging from .09 to .23. This indicated that the impact of organization membership on non-managerial employees’ responses was small, between .09 and 23. It was sufficient to conclude that the likelihood of having biased standard errors was small. The results are summarized in Table 5-II.
120
Table 5-II ICC (1) Results for Employees’ Data (n = 583) Variable Trust Distrust Control Mutuality Commitment Satisfaction Turnover Intention Contextual Performance No. of items 6 4 4 8 4 4 8 MSB 3.99 3.63 2.48 1.05 3.16 3.66 1.38 MSW .46 .56 .38 .30 .49 .81 .34 ICC (1) .23 .17 .18 .09 .17 .12 .11
Means, Standard Deviation, and Correlations I performed descriptive data analysis on both the managers’ and non-managers’ data. The managers’ data (n = 195) were aggregated in analyses involving the latent construct congruence of perceived relationship characteristics as the mediating variable. Descriptive statistics on the independent variables (symmetrical and asymmetrical relationship maintenance strategies) and the dependent variables (turnover intention and contextual performance) were based on non-managerial employees’ data (n = 583). The descriptive statistics yielded information on the two research questions and provided preliminary checks on content validity of the measures. I discuss these preliminary findings below. Table 6 and Table 7 summarize respectively the means, standard deviations, and correlations of independent and dependent variables. Table 8 and Table 9 provide the
121
means, standard deviations, and correlations of the dimensions of congruence of perceived relationship characteristics respectively based on the managerial respondents’ data (n = 195) as well as the organizational-level data (n = 30). Research Question 1: How do organizational members perceive relationship maintenance strategies used by their organizations? Descriptive statistics provided information on the first research question. With regard to independent variables (see Table 6), non-managerial employees perceived their organizations’ relationship maintenance strategies to be fairly symmetrical. The organizations reportedly had a relatively high level of disclosure, e.g., actively communicating with employees and encouraging employees to share thoughts and feelings about the organization (M = 3.50); showed considerable amount of assurances of legitimacy, e.g., taking employees’ concerns seriously (M = 3.62); made efforts to network with employees, e.g., involved in coalitions that benefit employees (M = 3.42); and provided advice to employees (M = 3.49). In addition, these employees appeared to believe that their organizations engaged less in asymmetrical relationship maintenance strategies, such as distributive negotiation (M = 2.70) and avoiding (M = 2.54). Yet they did feel that compromising was often used by organizations in conflict situations (M = 3.23). Taken together, non-managerial employees believed that their organizations employed more symmetrical relationship maintenance strategies than asymmetrical ones. Table 6 also demonstrated that all the independent variables were significantly correlated at the .01 level. The symmetrical relationship maintenance strategies were significantly and positively correlated with each other. On the other hand, two of the asymmetrical relationship maintenance strategies, i.e., distributive negotiation and
122
avoiding, were positively and significantly correlated with each other (r = .46, p < .01). Nevertheless, the third asymmetrical relationship maintenance strategy compromising correlated negatively and significantly with distributive negotiation (r = -.46, p < .01) and avoiding (r = -.32, p < .01). Combined with its higher mean value than distributive negotiation and avoiding, it seemed that non-managerial employees did not perceive compromising as much to be asymmetrical. Overall, the correlation between symmetrical and asymmetrical relationship maintenance strategies was -.20 (p < .01), preliminarily indicating that symmetrical and asymmetrical relationship maintenance strategies might be inter-related but distinct concepts. Table 6 Means, Standard Deviations, and Correlations of Independent Variables (n = 583) Variable 1. Disclosure 2. Assurances of 3.62 Legitimacy 3. Networking 4. Advice 5. Distributive Negotiation 6. Avoiding 7. Compromising **p < .01. 3.42 3.49 2.70 2.54 3.23 .85 .78 .87 .76 .71 .61** .51** .60** .53** 1.00 .53** 1.00 1.00 .46** -.46** 1.00 -.32** 1.00 .87 .75** 1.00 M 3.50 SD .77 1 1.00 2 3 4 5 6 7
-.35** -.40** -.26** -.16** -.38** -.41** -.28** -.26** .43** .52** .38** .33**
123
Research Question 2: How do non-management employees and managers perceive the organization-employee relationships? Congruence of perceived relationship characteristics from the non-managerial employees’ perspective. Descriptive statistics offered insights into the second research question. Table 7 showed that these non-managerial employees reported that they had good relationships with their organizations, judging from the relatively high means of trust (M = 3.67), control mutuality (M = 3.58), commitment (M = 3.11), and satisfaction (M = 3.77), as well as low mean level of distrust (M = 2.14). In addition, these five dimensions of the congruence of perceived relationship characteristics were all significantly correlated with each other at the .01 level. Coupled with the fact that distrust had significantly negative correlations with the other four dimensions, these significant correlations yielded preliminary support for the third proposition that relationship characteristics consist of these five dimensions. Table 7 Means, Standard Deviations, and Correlations of Dependent Variables (n = 583) Variable 1. Trust 2. Distrust 3. Control Mutuality 4. Commitment 5. Satisfaction 6. Turnover Intention 7. Contextual Performance M 3.67 2.14 3.58 3.11 3.77 2.98 3.79 SD .80 .83 .68 .57 .79 .97 .56 1 1.00 -.67 ** .72** .41** .80** -.51** .18** 1.00 -.66** -.28** -.66** .43** -.09* 1.00 .37** .74** 1.00 .45** 1.00 1.00 -.26** 1.00 2 3 4 5 6 7
-.47** -.48** -.55** .19** .27** .18**
124
*p < .05. **p < .01. Congruence of perceived relationship characteristics from the organizational perspective. I also examined the means, standard deviations, and correlations of congruence of perceived relationship characteristics from both the individual level (managerial employees’ responses) and the organizational level perspective (aggregated managers’ data). I averaged managerial employees’ responses (n = 195) in each organization to form the organizational-level data (n = 30). According to Table 8, managers reported high means of trust, control mutuality, commitment, and satisfaction, and low means of distrust (M = 2.08). Except distrust, the other four dimensions were all positively highly and significantly correlated with each other, at the .01 level. Distrust was significantly but negatively correlated with the other four dimensions. Similar patterns emerged from the organizations’ data in Table 9. The statistics suggested that organizations perceived good relationships with their employees. Table 8 Means, Standard Deviations, and Correlations of Congruence of Perceived Relationship Characteristics (n = 195) Variable 1. Trust 2. Distrust 3. Control Mutuality 4. Commitment 5. Satisfaction **p < .01. M 3.77 2.08 3.66 3.28 3.92 SD .78 .84 .76 .51 .72 1 1.00 -.79** .78** .43** .84** 1.00 -.78** -.35** -.78** 1.00 .34** .80** 1.00 .40** 1.00 2 3 4 5
125
Table 9 Organizational-Level Means, Standard Deviations, and Correlations of Congruence of Perceived Relationship Characteristics (n = 30) Variable 1. Trust 2. Distrust 3. Control Mutuality 4. Commitment 5. Satisfaction **p < .01. Turnover Intention and Contextual Performance from the Non-Managerial Employees’ Perspective Table 7 also presented descriptive information on turnover intention and contextual performance. Non-managerial employees reported moderate intention to leave their current organizations (M = 2.98) and considered themselves good performers on the job (M = 3.79), such as helping others’ work and contributing to organizations’ accomplishments. Also, these two dependent variables showed significant and negative correlation with each other (r = -.26, p < .01). Lastly, turnover intention had significantly negative correlations with four dimensions of congruence of perceived relationship characteristics (trust, control mutuality, commitment, and satisfaction), and significantly positive correlations with the dimension distrust. Similar patterns were found in contextual performance’s correlations with these five dimensions. M 3.78 2.07 3.66 3.30 3.92 SD .53 .52 .49 .27 .48 1 1.00 -.89** .92** .66** .96** 1.00 -.91** -.58** -.89** 1.00 .66** .95** 1.00 .68** 1.00 2 3 4 5
126
Reliability Tests and Factor Analyses After examining the data and descriptive statistics, I assessed reliability and validity of the measures with Cronbach’s alpha test, exploratory factor analysis (EFA), particularly principal component analysis (PCA), and confirmatory factor analysis (CFA). PCA was the extraction method used in EFA. Kaiser’s rule (eigenvalue > 1) was chosen for selection of number of components extracted. Varimax rotation was used to handle the issue of cross-loadings. EFA evaluates the dimensionality of measures and provides a preliminary set of measures. CFA assesses the degree to which data fit expected structure, yielding information on construct validity and construct reliability. I used the maximum likelihood estimation method in CFA. Reliability Coefficients for Initial Measures Table 10 summarizes alpha coefficients for the initial measures. Nunnally (1978) suggested .70 as an acceptable reliability coefficient threshold. Most of the initial measures had alpha values above .70. Commitment scale’s alpha coefficient was borderline (? = .68) for non-managerial data, and poor (? = .53) for managerial data. A possible explanation for this is that managers may be more homogenous than the nonmanagers on commitment. With everything else being equal, a more homogenous sample generally results in a lower reliability estimate than a more heterogeneous sample (Crocker & Algina, 1986, p. 144). The advice scale also has relatively low reliability values (? = .61 for managerial data, .62 for non-managerial data). This was probably because there were only two items for the variable.
127
Table 10 Cronbach’s Alpha of Initial Measures Variable Disclosure Assurances of Legitimacy Networking Advice Distributive Negotiation Avoiding Compromising Trust Distrust Control Mutuality Commitment Satisfaction Turnover Intention Contextual Performance Managers .75 .84 .87 .61 .90 .83 .83 .91 .87 .83 .53 .92 .81 .80 Non-Managers .78 .85 .89 .62 .88 .81 .81 .92 .85 .79 .68 .93 .83 .80 N of Items 4 3 3 2 3 4 3 6 4 4 8 4 4 8
EFA and Reliability Results For EFA, I retained items that loaded higher than .50 on a single factor (Hair et al.). Some items loaded highly (? .50) on more than one factor. To deal with the issue of cross-loadings, I used Varimax rotation so as to arrive at a simpler structure (Hair et al.). If an item still loaded on multiple factors after the rotation, it was deleted. The analyses
128
of the five dimensions of congruence of perceived relationship characteristics were based on all data (N = 785). The analyses on other variables were based on non-managerial employees’ responses (n = 583). Table 11, 12, and 13 present the factor loadings and alpha values of the revised scales. I discuss the results of different variables below. Relationship maintenance strategies. Proposition 2: Symmetrical relationship maintenance strategies are composed by disclosure or openness, assurances of legitimacy, networking, and advice. Asymmetrical relationship maintenance strategies include distributive negotiation, avoiding, and compromising. Research Question 1: How do organizational members perceive relationship maintenance strategies used by their organizations? The second proposition and first research question explored the kinds of relationship maintenance strategies used by organizations. I proposed four main symmetrical relationship maintenance strategies and three asymmetrical ones. I used 12 items to measure the four different kinds of symmetrical relationship maintenance strategies. Four items measured disclosure and three items assessed each of assurances of legitimacy and networking. Advice had two items. The EFA results showed that that disclosure, assurances of legitimacy, networking, and advice emerged as one factor—symmetrical relationship maintenance strategies. One item on networking (“the coalitions that the organization forms with other groups benefit the organization’s employees”) was deleted due to significant cross-loading. I then examined the reliability of symmetrical relationship maintenance strategies with all the remaining 11 items. It had excellent reliability (? = .90). The loadings of each item are presented in Table 11.
129
With regard to asymmetrical relationship maintenance strategies, 10 items were used to measure it. Three items were respectively used to assess distributive negotiation and compromising. Four items measured avoiding. The initial extraction in EFA resulted in multiple measures having cross-loadings. After the Varimax rotation, all the items loaded highly on a single factor. As a result, no item was deleted. The reliability coefficients of the three kinds of strategies were the same as the initial measures (see Table 11). All had highly reliable measures (> .80). Distributive negotiation, avoiding, and compromising stood out as three separate factors, instead of one factor of asymmetrical relationship maintenance strategies. It is possible that the latent construct asymmetrical relationship maintenance strategies is a second-order factor, and distributive negotiation, avoiding, and compromising are three first-order factors that are caused by asymmetrical relationship maintenance strategies. The EFA results also suggested that symmetrical and asymmetrical relationship maintenance strategies were related but conceptually distinct latent factors. These structures from EFA were further tested in CFA, and are discussed later. The factor loadings of each item obtained in EFA are shown in Table 11.
130
Table 11 Items, Factor Loadings, and Reliability for Relationship Maintenance Strategies (n = 583) Factor Construct Item Loading The organization’s annual report is a valuable source of information for employees about what it has done. (Disclosure 1) The organization shares enough information with employees about the organization’s governance. (Disclosure 2) The organization’s employee meetings are a valuable way for employees to communicate their opinions to the organization. (Disclosure 3) The organization encourages employees to disclose thoughts and feelings about the organization. (Disclosure 4) The organization makes a genuine effort to provide personal .80 responses to employees’ concerns. (Assurances 1) Symmetrical Relationship Maintenance Strategies The organization’s policy development process allows employees adequate opportunity to raise an issue and propose a solution. (Assurances 2) When employees raise concerns, the organization takes these concerns seriously. (Assurances 3) .76 .80 .76 .76 .71 .57 Alpha
131
Table 11 (continued) Construct Item The organization builds coalitions with groups that help employees’ career growth. (Networking 1) The activities that the organization is involved in with other groups are helpful to its employees. Symmetrical (Networking 3) Relationship The organization tells employees like me what Maintenance to do about problems on the job. (Advice 1) Strategies The organization shares opinions on things going on in my career. (Advice 2) Eigenvalues % of Variance Explained 50.96 .68 5.61 .90 .51 .71 .74 Factor Loading Alpha
132
Table 11 (continued) Construct Item Factor Loading When there is a conflict between employees and the management, the management will not give up, but keep on persuading until the employees concede. the management applies pressure so that employees may make concessions. Distributive Negotiation the management strongly adheres to its point of view until the employees agree. When there is a conflict between employees and the management, the management gives as little attention as possible to it. the management avoids addressing it. the management tries to avoid a confrontation. the management tries to avoid expressing differences of Avoiding opinion with employees involved. .77 .81 .70 .73 .84 .82 .88 .83 .90 Alpha
133
Table 11 (continued) Construct Item Factor Loading When there is a conflict between employees and the management, the management tries to realize a middle-of-the-road solution with employees. the management emphasizes a compromise solution. the management insists that both employees and the Compromising management give in a little. Eigenvalues % of Variance 73.47 Explained .85 .76 7.35 .81 .86 Alpha
Congruence of perceived relationship characteristics. Proposition 3: Congruence of employees’ and organizations’ (represented by managers’ perceptions) perceptions of their relationship characteristics best captures the relational dynamics. These relationship characteristics are made up of five components: trust, distrust, control mutuality, satisfaction, and commitment. Research Question 2: How do non-management employees and managers perceive the organization-employee relationships? The third proposition and second research question proposed that perceived relationship characteristics have five dimensions. I used six items for trust, four items for
134
distrust, four items for control mutuality, eight items for commitment, and four items for satisfaction. Because both employees and organizations were involved here, the EFA was based on all data (N = 785). A Varimax rotation was performed to eliminate multiple cross-loadings. No item was deleted. Factor loadings are provided in Table 12. The EFA results provided preliminary support for the proposition that trust, distrust, control mutuality and satisfaction are distinct latent factors. Commitment items emerged as three separate factors, i.e., the three sub-dimensions of affective, continuance, and normative commitment. Similar to the case of asymmetrical relationship maintenance strategies, it was likely that commitment was a higher-order factor that caused these three first-order factors. These structures from EFA were further tested in CFA and will be discussed later. I examined the reliability coefficients for these seven factors. The coefficients for trust (? = .92), distrust (? = .85), control mutuality (? = .79), and satisfaction (? = .93) were high. I also obtained separate reliability coefficients for the three sub-dimensions of commitment. Affective commitment had a reliability of .82, continuance commitment .61, and normative commitment .57. Except affective commitment, the other two subdimensions’ reliability were minimally acceptable. It was probably because each subdimension only has two or three items. Generally, the shorter a scale becomes, the smaller alpha is (Crocker & Algina, 1986). Take together, the measures of trust, distrust, control mutuality, commitment, and satisfaction were reliable. The reliability coefficients are also included in Table 12.
135
Table 12 Items, Factor Loadings, and Reliability for Relationship Characteristics (n = 785) Construct Item Factor Loading I believe that the organization treats employees fairly and justly. Whenever this organization makes an important decision, I know it will be concerned about its employees. The organization can be relied on to keep its promises to the employees. I am willing to let the organization make decisions for employees like me. I feel very confident about the organization’s capabilities. The organization has the ability to accomplish what it says Trust it will do. Eigenvalues % Variance 70.62 Explained .80 4.24 .92 .80 .87 .87 .86 .84 Alpha
136
Table 12 (continued) Construct Item Factor Loading I am skeptical about whether the organization will keep the employees’ interests in mind when it makes decisions. I feel that this organization will exploit employees’ vulnerability given the chance. I feel that this organization will engage in damaging and harmful behavior to employees to pursue its own interest. I feel that the way this organization is run is irresponsible Distrust and unreliable. Eigenvalues % of Variance Explained 71.25 .86 2.85 .85 .87 .88 .76 Alpha
137
Table 12 (continued) Construct Item Factor Loading This organization and employees like me are attentive to what each other say. This organization believes the opinions of employees like me are legitimate. In dealing with employees like me, this organization has a tendency to throw its weight around. (R) Control Mutuality Eigenvalues % of Variance Explained 63.59 This organization really listens to what employees like me have to say. .86 2.544 .79 .67 .87 .77 Alpha
138
Table 12 (continued) Construct Item Factor Loading I do not feel like “part of the family” at my organization. (R) This organization has a great deal of personal meaning for me. Affective Commitment There is a long-lasting bond between this organization and employees like me. I feel that I have too few options to consider leaving this organization. Continuance Commitment Too much in my career would be disrupted if I decided to leave the organization now. I think that people these days move from organization to organization too often. Jumping from organization to organization does not seem at all unethical to me. (R) One of the major reasons I continue to work for this Normative organization is that I believe that loyalty is important and Commitment therefore feel a sense of moral obligation to remain. Eigenvalues % of Variance Explained 67.84 .58 5.43 .57 .83 .70 .82 .61 .82 .91 .82 .90 .72 Alpha
139
Table 12 (continued) Construct Item Factor Loading I am happy with this organization. Both the organization and employees like me benefit from the relationship. Most employees like me are happy in their interactions with this organization. Generally speaking, I am pleased with the Satisfaction relationship this organization has established with me. Eigenvalues % of Variance Explained Note. (R) means that the items were reverse-coded. Turnover intention and contextual performance. As I discussed in Chapter II, organizations are supposed to incorporate the values and demands of their strategic constituencies into their long-term and short-term goals in order to be effective. Turnover intention and contextual performance are indicators of the extent to which organizations do so. I used four items to measure turnover intention, and eight items to evaluate contextual performance. Statistical analyses were based on non-managerial employees’ data (n = 583). EFA showed the appropriateness of these measures and dimensionality of these two dependent variables. 82.13 .92 3.29 .93 .88 .90 .93 Alpha
140
A Varimax rotation was performed on contextual performance items to eliminate significant cross-loadings. The last two items of contextual performance were deleted because they failed to load highly on a single factor. The items were: 1) “I tend to follow proper procedures and avoid unauthorized shortcuts;” 2) “I have put in extra hours to get work done on time.” All three scales were very reliable (? > .75). Table 13 presents the factor loadings as well as reliability coefficients of the revised scales. The EFA extraction indicated that turnover intention emerged as one factor, while contextual performance items yielded two factors. The first three items of contextual performance grouped as one factor, whereas the next three items loaded highly on another. The first three items assessed how much employees helped co-workers and volunteered to do extra work. The other three items dealt more with the contributions employees made toward their organizations. Organ (1997) respectively labeled them as altruism and conscientiousness. Altruism behaviors are geared towards individuals, whereas conscientiousness actions are focused on organizations. They both are components of contextual performance. Therefore, I named these two factors as altruism and conscientiousness. Similar to the three factors of asymmetrical relationship maintenance strategies and three factors of commitment, these two factors were likely governed by one underlying higher-order latent factor contextual performance. CFA results on this are to be discussed in the following section.
141
Table 13 Items, Factor Loadings, and Reliability for Turnover Intention and Contextual Performance (n = 583) Construct Item Factor Loading I would prefer another more ideal job than the one .85 I now work in. I have seriously thought about changing .78 Alpha
organizations since beginning to work here. I may leave this organization and work for another one in the next year. Turnover Intention Eigenvalues % of Variance Explained 66.31 I plan to stay in this organization to develop my career for at least three years. (R) .73 .83 .88
2.65
142
Table 13 (continued) Construct Item Factor Loading I often help other employees with their work when they have been absent. Altruism (Contextual Performance) I help other employees when their work load increases. I often volunteer to do things that are not formally required by the job. I have made suggestions to improve my organization. I have informed the management of potentially Conscientiousness (Contextual Performance) Eigenvalues % of Variance Explained Note. (R) means that the item was reverse-coded. Summary. In conclusion, exploratory statistical tests of the variables resulted in four factors of relationship maintenance strategies, three of which were likely first-order factors of asymmetrical relationship maintenance strategies. Seven factors were extracted for perceived relationship characteristics, three of which were likely first-order factors of commitment. The other four factors were trust, distrust, control mutuality, and 57.92 unproductive policies and practices. .86 I am willing to speak up when policy does not contribute to goal achievement of the organization. .85 4.63 .82 .76 .76 .78 .80 Alpha
.84
143
satisfaction. Turnover intention items emerged as one factor, whereas contextual performance grouped as two factors (altruism and conscientiousness). One item of networking and two items of contextual performance were deleted. Most items had high internal consistency. Continuance commitment and normative commitment had acceptable internal consistency. However, EFA is exploratory in nature. To confirm these structures, particularly to test whether some of the extracted factors were first-order factors (i.e., commitment factors, asymmetrical factors, and contextual performance factors), I conducted CFA. CFA Results I used the EQS 6.1 Program for CFA (Bentler, 2005). Following the two-step process, I first specified the initial measurement models. Then, using the Lagrange Multiplier test in EQS, I checked for model specification errors, such as error covariances and cross-loadings where a measured variable serves as an indicator of more than one latent factor. Respecifications arrived at the final measurement model (Hancock & Mueller, 2007). To evaluate data-model fit, I used the oft-cited Hu and Bentler (1999) joint-criteria approach. According to the joint-criteria, a model is considered tenable when it achieves NNFI ? .96, CFI ? .96, and SRMR ? .09 or RMSEA ? .06 and SRMR ? .09. Based on the EFA results, I compared different models for each latent construct. Regarding comparison among hierarchically nested models, I performed chi-square difference tests. Concerning comparison among models that were not nested within each other, I compared model AIC values. The models with smaller AIC values were considered a better representation of the data (Hancock & Mueller).
144
Relationship Maintenance Strategies Symmetrical relationship maintenance strategies. For symmetrical relationship maintenance strategies, I compared a one-factor model, a four-factor oblique model, and a second-order factor model. I did not include a four-factor orthogonal model because it is locally under-identified with advice and networking having only two measures respectively. The one-factor model was the most parsimonious model. Along with the second-order factor, it was hierarchically nested within the four-factor oblique model. Therefore I used nested model chi-square difference tests for nested model comparison. Table 14 summarizes the model comparison results and goodness-of-fit indices. The null model’s information as a baseline for computing goodness-of-fit indices is also presented. The statistics in Table 14 showed that the one-factor model had significantly datamodel fit than the null model: ??2 = 2729.122 with ?df = 11, p < .01. Next, I compared the one-factor model with the four-factor oblique model. The latter demonstrated significantly better data-model fit: ??2 = 213.509 with ?df = 6, p < .01. Also, its fit indices met the Hu and Bentler (1999) criteria: NNFI = .958 (approaching .96), CFI > .96, and SRMR = .031 (< .09). In addition, its RMSEA was very close to .06. Its model AIC was much smaller than that of the one-factor model. Thus I concluded that the four-factor oblique model was a more viable representation of the relations underlying the data.
145
Table 14 Symmetrical Relationship Maintenance Strategies CFA Results (n = 583) Model Null Model One-Factor Model 338.737 (Model A) Four-Factor A vs. B Oblique Model (Model B) Second-Order B vs. C Factor Model (Model C) 150.589 40 < .01 .950 .963 .037 .069 70.589 2 25.361 < .01 125.228 38 < .01 .958 .971 .031 .063 49.228 6 213.509 < .01 44 < .01 .878 .902 .053 .107 250.737
?2
3067.859
df 55
p
NNFI
CFI
SRMR
RMSEA
AIC 2957.859
?df
??2
p
146
Then, I compared the four-factor oblique model with the second-order factor model: ??2 = 25.361 with ?df = 2, p < .01. The nested model chi-square difference test showed that the four-factor oblique model was significantly better. Furthermore, as stated above, its fit indices met the Hu and Bentler criteria. Its model AIC was also much smaller than that of the second-order factor model. Based on the comparisons, I concluded that the four-factor oblique model was the most viable structure for the data with respect to symmetrical relationship maintenance strategies. The CFA results provided evidence to Proposition 2 that symmetrical relationship maintenance strategies were composed by disclosure or openness, assurances of legitimacy, networking, and advice. I provide the loadings for the measures on the four factors, correlations among the factors, and error variances in Figure 3. All the loadings, correlations, and path coefficients were significant at least at the .05 level. It was not surprising that these four factors were all significantly correlated, because they were all measuring the underlying concept of symmetrical relationship maintenance strategies.
147
V1 .533 Disclosure .687
.
.
846
E1
V2
.727
E2
758 V3
.653
E3
.
762 V4 .648 E4
.
926 V5
.
575
.
751 Assurance
.818 .809
.
E5
.
720 785
V6
.
587
E6
V7
.
.
620
E7
731 .835 Networking .805 .745 .558 Advice .805 V11
. .
.
700
V8
.
550
E8
V9
594
E9
V10
830
E10
.
594 E11
Figure 3. Final four-factor oblique model of symmetrical relationship maintenance strategies.
148
Asymmetrical relationship maintenance strategies. For asymmetrical relationship maintenance strategies, I compared a one-factor model, a second-order factor model, a three-factor orthogonal model, and a three-factor oblique model. The first three models were all hierarchically nested within the three-factor oblique model. I used both the nested model chi-square difference test and model AIC to compare these models. Table 15 summarizes the results. Table 15 Asymmetrical Relationship Maintenance Strategies CFA Results (n = 583) Model Null Model One-Factor Model Three-Factor Orthogonal Model Three-Factor Oblique Model Second-Order Factor Model 204.953 32 < .01 .919 .942 .061 .096 140.953 204.954 32 < .01 .919 .942 .061 .096 140.954 590.411 35 < .01 .762 .815 .254 .165 520.411
?2
3039.310 1147.60
df 45
p
NNFI
CFI
SRMR
RMSEA
AIC 2949.310
35 < .01
.522
.628
.131
.234
1077.603
The statistics in the above table showed that the one-factor model had significantly better data-model fit than the null model: ??2 = 1891.71 with ?df = 10, p < .01. Next, I compared the one-factor model with the three-factor oblique model. Because the two models were not nested within each other, it was incorrect to do the model chi-square difference test. I compared their model AIC. The three-factor
149
orthogonal model had a much smaller AIC value (520.411) than that of the one-factor model (1077.603), which suggested that the three-factor orthogonal model was better than the one-factor model. Then, I compared the three-factor orthogonal model with the three-factor oblique model. As the three-factor orthogonal model was nested within the three-factor oblique model, I conducted the chi-square difference test: ??2 = 385.457 with
?df = 3, p < .01. The test showed that the three-factor oblique model had significantly
better data-model fit than the three-factor orthogonal model. Also, the fit indices of the three-factor oblique model were close to the Hu and Bentler (1999) criteria: NNFI = .919 (approaching .96), CFI = .942 (borderline), and SRMR = .061 (< .09). Lastly, I compared the three-factor oblique model with the second-order factor. It was worth noting that these two models were equivalent models (see Hershberger, 2006 for more a detailed discussion on equivalent models). Given that equivalent models had the identical fit indices and chi-square values among others, one criterion for selecting the best model among equivalent models was the information complexity criterion (ICOMP). ICOMP = model misfit + complexity. The model with lower correlations among the parameter estimates would have smaller discrepancy between the implied and observed covariance matrix (i.e., small misfit) and smaller complexity, hence a better model for the data (Hershberger). I obtained correlations among the parameter estimates for both the three-factor oblique model and the second-order factor model. The second-order factor model had lower correlations among its parameter estimates than the three-factor oblique model, indicating that it was a better model for the data. Furthermore, the second-order factor model’s fit indices had acceptable fit based on the Hu and Bentler (1999) joint criteria: NNFI = .919 (approaching .96), CFI = .942 (borderline), and SRMR = .061
150
(< .09). As a result, I concluded that the second-order factor model was a more viable structure for the data. To sum up, the CFA results supported Proposition 2 that asymmetrical relationship maintenance strategies consisted of distributive negotiation strategies, avoiding, and compromising. I present the loadings for the measures on the three factors and error variances in Figure 4. All the loadings and path coefficients were significant at the .05 level. The three first-order factors were all highly significantly correlated with the underlying factor of asymmetrical relationship maintenance strategies, ranging from .663 to .819.
151
Asymmetrical Relationship Maintenance Strategies
.819 D1 .574 D2 .677 Distributive Negotiation .789 .873 .893
.736
-.663 D3 .748
Avoiding
Compromising
.859 V2 .512 E2
.875 .566
.475
.698
.952
.669
V1 .614 E1
V3 .487 E3 .451
V4
V5 .484
V6
V7
V8 .716 E8
V9 .305 E9
V10 .743 E10
.824 .880 E6 E7
E4
E5
Figure 4. Final second-order factor model of asymmetrical relationship maintenance strategies. Turnover Intention and Contextual Performance Turnover intention. As the EFA results yielded a one-factor structure for turnover intention, I tested the one-factor model in CFA. The results are summarized in Table 16. The one-factor model appeared to fit the data significantly better than the null model: ??2 = 880.653 with ?df = 4, p < .01. However, it did not have a very good data-model fit according to the Hu and Bentler (1999) criteria: NNFI = .679 (< .96 criterion), CFI = .893 (< .96 criterion), and SRMR