Description
In criminal law, entrapment is conduct by a law enforcement agent inducing a person to commit an offense that the person would otherwise have been unlikely to commit
ABSTRACT
Title of dissertation:
FACTORS AFFECTING ENTRAPMENT BIAS: JUSTIFICATION NEEDS, FACE CONCERNS AND PERSONAL NETWORKS Anna Karavanov, Doctor of Philosophy, 2006
Dissertation directed by:
Professor Deborah A. Cai Department of Communication
This study explores the link between the entrapment bias and the concept of face (self- and other-positive) and internal and external justification processes. It examines how face-saving concerns and justification needs moderate the entrapment bias in accountability condition (i.e., presence of constituencies and reporting requirements). In addition, this research examines whether the size and influence of personal networks is associated with face-saving behaviors that, in turn, affect entrapment. The research also explores whether overall face concerns have an effect on internal and external selfjustification. Finally, the study explored messages used by individuals in a scenario potentially leading to entrapment. Respondents in the study were 236 undergraduate students majoring in communication enrolled in a large East Coast university. Study participants were assigned to one of the four conditions: (1) constituency, reporting; (2) constituency, no reporting; (3) no constituency; reporting; (4) no constituency; no reporting. The current investigation did not support the findings from previous studies that suggest that justification processes and face concerns lead to entrapment. This study
found that only internal self-justification and other-positive face concerns are related to entrapment, but instead of contributing to entrapment, these aspects prevent individuals from becoming entrapped. Personal networks were demonstrated to have positive effect on both self- and other-positive face concerns, providing empirical support for the value of using personal networks as a predictor of face goals. However, personal networks did not contribute to entrapment. Finally, the study examined messages used by individuals in a situation leading to entrapment, suggesting that when individuals try to explain their behavior, they tend to use causal accounts. Overall, this study has made a contribution to the field of communication by identifying processes and conditions (e.g., concern for other-positive face, internal selfjustification, reporting requirement, no direct observation by constituency, keeping clear record of performance success or failure) that may prevent entrapment bias from occurring. These processes and conditions could potentially improve the outcomes of negotiation with the use of effective communication strategies.
FACTORS AFFECTING ENTRAPMENT BIAS: JUSTIFICATION NEEDS, FACE CONCERNS AND PERSONAL NETWORKS
by
Anna Karavanov
Dissertation submitted to the Faculty of theGraduate School of the University of Maryland, College Park in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2006
Advisory Committee: Professor Deborah A. Cai, Chair Professor Robert Gaines Professor Cynthia Stevens Professor Monique Turner Dr. Leah Waks
©Copyright by Anna Karavanov
2006
ii
DEDICATION
This dissertation is dedicated to the memory of my father, Alexander Karavanov.
iii
ACKNOWLEDGEMENTS
As with any milestone, there are always many people who make the accomplishment possible. It is unlikely that I can name all of the people who have contributed to the accomplishment. Thank you. As this experience comes to an end, I’m amazed that I still have a husband, family and friends who are all still talking to me. Thank you for seeing me through endless frustrations, grumpiness and bouts of irritability. And a very special “thank you” to Dr. Deborah Cai, my advisor. Thank you so much for your support and guidance.
iv
TABLE OF CONTENTS
List of Tables ..................................................................................................................... vi List of Figures .................................................................................................................. viii Chapter 1 Overview........................................................................................ 1
Theoretical Rationale .............................................................................................. 3 Entrapment Bias.......................................................................................... 3 Theoretical Frameworks ................................................................. 3 Self-justification.............................................................................. 8 Gaps in Self-Justification and Entrapment Research.................... 11 Entrapment and Negotiation ..................................................................... 14 Accountability and Constituency.............................................................. 17 Accountability and Constituencies in Negotiation Context.......... 17 Accountability, Constituency, and Entrapment Bias .................... 19 Face ........................................................................................................... 20 Face-Saving and Negotiation........................................................ 21 Face-Saving and Entrapment Bias ................................................ 24 Personal Networks .................................................................................... 28 Luck .......................................................................................................... 33 Summary of Research Hypotheses ........................................................... 35 Research Question .................................................................................... 37 Chapter 2 Method ......................................................................................... 39 Experimental Design................................................................................. 39 Study Sample ............................................................................................ 40 Recruiting Procedures and Research Assistants ....................................... 41 Data Collection and Experimental Procedures ......................................... 43 Task Description ........................................................................... 44 Constituency Manipulation........................................................... 45 Reporting Manipulation ................................................................ 48 Dependent Variable ...................................................................... 49 Final Materials and Development of Measures ............................ 50 Results.......................................................................................... 80 Data Preparation........................................................................................ 80 Test of the Hypotheses.............................................................................. 87 Luck and Entrapment.................................................................... 95 Research Question .................................................................................... 97 Coding scheme and procedures .................................................... 98 Analysis....................................................................................... 108 Findings....................................................................................... 111 Discussion................................................................................. 113
Chapter 3
Chapter 4
v
Summary of the Study Design and Gaps Addressed by the Research.... 113 Research Hypotheses .............................................................................. 117 Implications............................................................................................. 130 Limitations of the Study.......................................................................... 134 Experimental Design and Manipulation ..................................... 135 Personal Networks Measures...................................................... 137 Directions for Future Research ............................................................... 138 Significance of the Study........................................................................ 141 Appendix A: IRB: Approval Letter .............................................................................. 145 Appendix B: Participant Solicitation Announcement.................................................... 146 Appendix C: ID Number Instructions............................................................................ 147 Appendix D: Informed Consent Form ........................................................................... 148 Appendix F: Personal Networks Questionnaire............................................................. 150 Appendix G: Influence Scale ......................................................................................... 160 Appendix H: Face Scale................................................................................................. 162 Appendix I: Experimental Conditions: Participant Instructions.................................... 166 Appendix J: Participant Information Packet .................................................................. 174 Appendix K: Manipulation Check ................................................................................. 186 Appendix L: Luck Questionnaire................................................................................... 188 Appendix M: Outcome Narrative .................................................................................. 192 Appendix N: Entrapment Questionnaire........................................................................ 203 Appendix O: Debriefing Form....................................................................................... 215 References....................................................................................................................... 216
vi
LIST OF TABLES Table 1 2 Page Study Design…………………………………………………………...40 Network Questionnaire: Summary of Descriptive Statistics for Total Number of People Nominated in the Networks for Items 1 through 8………………………………………………………………52 3 4 5 6 Influence Scale: Summary of Descriptive Statistics…………………...55 Face Scale: Summary of Descriptive Statistics……………………….66 Belief In Good Luck Scale: Summary of Descriptive Statistics……….67 Entrapment Bias Scale: Summary of Descriptive Statistics Condition I……………………………………………………………72 7 Entrapment Bias Scale: Summary of Descriptive Statistics Condition II…………………………………………………………....74 8 Entrapment Bias Scale: Summary of Descriptive Statistics Condition III…………………………………………………………..76 9 Entrapment Bias Scale: Summary of Descriptive Statistics Condition IV…………………………………………………………..77 10 11 12 Scale Reliability Coefficients,…………………………………………79 Correlation Matrix: Independent and Dependent Variables…………81 Independent Variables: Factor Score Descriptive Statistics for Condition I……………………………………………….83 13 Independent Variables: Factor Score Descriptive Statistics for Condition II……………………………………………….84
vii
14
Independent Variables: Factor Score Descriptive Statistics for Condition III……………………………………………...85
15
Independent Variables: Factor Score Descriptive Statistics for Condition IV…………………………………………….86
16 17 18
Codebook…………………………………………………………….104 Condition by Qualitative Category ………………………………….109 Percent of People Entrapped or Not Who Used Specific Qualitative Themes To Explain Their Actions…...…………………...110
19
Interaction of Reporting (Yes/No) and Constituency (Yes/No) on External Justification: Mean Summary…………………………...121
viii
LIST OF FIGURES Figure 1 2 3 Page The proposed value function from prospect theory……………..…….5 Research hypotheses…………………………………………….……36 Research findings……………………………………………….…….118
1
CHAPTER 1 Overview A woman decides to wait for a bus rather than walk. With the decision to wait for the bus and time invested in doing so, the woman waits for such a long time that she could have walked to the destination and back again by the time the bus finally arrives. This situation exemplifies the entrapment bias, when individuals continue to incur costs to achieve their objectives instead of changing their behavior. The entrapment bias is the tendency of people to assume that the more resources are expended, the closer they are to attaining their desired goal (Rubin, Kim, & Peretz, 1990). In research, this phenomenon is referred to as sunk cost (Arkes & Blumer, 1985), entrapment or entrapment bias (Brockner, 1977), concord fallacy (Arkes & Ayton, 1999), and escalation of commitment (Staw, 1976, 1981). All of these terms refer to the same phenomenon, and in the research literature they are used interchangeably. The phenomenon of entrapment has been observed and studied by scholars on different levels (i.e., macro, interpersonal and intrapersonal) and in different types of situations (e.g., waiting, bidding at an auction, gambling, decision making) to uncover variables and circumstances underlying the entrapment (Brockner, 1977; Schelling, 1960; Shubik, 1971; Staw, 1976). Responsibility levels, social motivations, information ambiguity, and observed rate of loss have been identified as some of the variables influencing conflict escalation (Rubin & Brockner, 1975; Staw, 1981; Teger, 1980). The entrapment bias has been studied predominantly within business decisionmaking and organizational contexts, including negotiation (Neale & Bazerman, 1985). Negotiation is used in a wide variety of settings: policy and law formulation, fiscal
2
budgets and salary discussions, house settlements and establishment of children’s curfews (Wall & Blum, 1991; Ways, 1979). Although many negotiators are able to achieve their desired outcomes, others fail to do so even when the parties involved have a zone of agreement; that is, when the maximum price the buyer is willing to offer is larger than the minimum price acceptable to the seller (Raiffa, 1982). This logic is applicable to a wide variety of negotiation contexts—not just buyer-seller interactions (Neale & Bazerman, 1985). Bazerman (1986) suggests that the entrapment bias could be one of the reasons why negotiators fail to achieve settlements. This dissertation examines the link between the entrapment bias and the concept of face—the desire to create and sustain positive identity in front of others (Goffman, 1955)—and internal and external justification processes. Because negotiation often involves the presence of other people (i.e., constituency) who influence negotiators’ performance (Gelfand & Realo, 1999), this dissertation explores how face-saving concerns and justification needs moderate the entrapment bias in the presence of constituencies. In addition, this research examines whether the size and influence of personal networks is associated with face-saving behaviors that, in turn, affect entrapment. In this chapter the theoretical rationale for the study is provided and research hypotheses are outlined. Chapter 2 provides an overview of the research methods employed in the study. Chapter 3 presents the study results. The summary of the study, chapter 4 discusses the results and their implications, and identifies limitations of the research. In addition, directions for future research and its significance are provided in chapter 4.
3
Theoretical Rationale Entrapment Bias Theoretical Frameworks Brockner (1992) defines escalation as “the tendency for decision makers to persist with failing courses of action” (p. 39). Fox and Hoffman (2002) outline five common characteristics that all escalation situations share. First, an individual is engaged in a goal-directed activity (e.g., problem-solving). Second, some type of resources (e.g., money, time, effort, or emotion) has been expended to achieve the goal. Third, expenditures have not brought the desired results. Fourth, a decision has to be made whether to continue or quit investing in the same course of action. And fifth, future prospects seem unlikely for making gains or even covering losses by continuing in the same path, yet the person continues the original course of action. A number of explanations have been provided for the escalation of commitment, such as self-justification, prospect theory, decision dilemma and persistence. Staw (1976, 1981) used Festinger’s (1957) and Aronson’s (1968) theories of cognitive dissonance to suggest that individuals become entrapped because they feel the need to provide justification for their actions. The sources of justification could be either internal or external. With internal justification, the decision maker justifies to himself or herself that the decision to pursue the course of action was rational; withdrawal would indicate that the decision was inappropriate. With external self-justification, individuals want to appear rational or do not want to expose their mistakes to others, such as bosses or stakeholders.
4
Empirical evidence supports the self-justification explanation of entrapment. For example, studies conducted by Staw (1976), Staw and Ross (1987), Brockner and Rubin (1985), and Arkes and Blumer (1985) suggest that individuals tend to make investment decisions and stick to the chosen course of action so as not to appear wasteful and to appear consistent in their decisions. For example, Arkes and Blumer (1985) conducted a study in which they manipulated the cost of theater tickets for season subscribers. Some subscribers received discounts and others did not, and the tickets they received were discretely marked accordingly. The researchers counted ticket stubs after performances and the results showed that those individuals who paid full price were more likely to attend all of the plays or at least more plays than those individuals who purchased the discounted tickets. Another theory used to explain the entrapment phenomenon is Kahneman and Tversky’s (1979) prospect theory, applied to escalation by Whyte (1986). The theory suggests that a biasing effect occurs when individuals frame situations as losses or gains relative to some neutral reference point. The value function (or utility), which is in a shape of “S,” is convex and steep in the domain of losses and concave in the domain of gains (see Figure 1). In other words, individuals are expected to be risk averse when they consider the situation (prospect) from the point of view of maximizing gains and risktaking when they view the situation from the frame of minimizing losses. According to Whyte (1986), when using a loss frame, individuals feel compelled to recover the cost that they have lost even at the risk of losing more, thus becoming entrapped.
5
Value +
Losses
Gains
_
Figure 1. The proposed value function from prospect theory (Kahneman & Tversky, 1979). Bazerman (1984) added responsibility to the prospect theory explanations of escalation. He indicated that individuals responsible for making an original decision that led to failure have a different frame of mind than those who are not responsible for the decision. The nonresponsible decision makers are at a neutral reference point of the curve, because they have not experienced any sunk costs. In contrast, the responsible decision makers tend to be on the losing end of the curve because they have expended resources with no return and must decide whether to withdraw or continue by risking further. The failure to reinvest is perceived as certain loss, therefore, the decision makers feel pressured to re-invest to try to avoid such certain loss. Davis and Bobko (1986) conducted a study in which 50% of the participants were personally responsible for the initial funding decision and 50% were not. The researchers then manipulated the decision-making frame by providing the participants with information about a program that either was framed negatively as failing (i.e., “after 2 years of operation the program has failed to place 60.1% of all participants in either part-
6
time or full-time jobs”), or framed positively as a success (i.e., “after 2 years of operation the program has placed 39.9% of all participants in either part-time or full-time jobs”). Participants were then asked to make a decision to continue or terminate the program. The results of the study indicated that participants in the negative frame with personal responsibility for initial allocation conditions were more prone to entrapment than participants in the other three conditions. In other words, those participants both in the negative frame and personally responsible made the decision to continue with the program. Bowen (1987) proposed that the entrapment phenomenon could be explained by the framework called difficult business decision dilemma. He suggested that because many studies do not provide clear negative feedback about initial allocation of resources (e.g., the project is a failure), the escalation demonstrated by research participants could be explained by a variety of motives such as their curiosity, desire to bring a project to fruition, or desire to learn about the problem. However, Brockner (1992) argued that some research that used self-report measures has shown that the research participants did think the feedback they received about the project was indeed negative (Brockner & Rubin, 1985). Fox and Hoffman (2002) used motivation theories developed by Lewin (1935) and Atkinson and Raynor (1974) to offer another explanation of the escalation: that individuals desire to be persistent. In other words, people are goal driven and are motivated to accomplish their goals. Lewin’s theory involves psychological regions of tension, goal valence, or desirability, and psychological distance of the path to goal obtainment (how long or difficult is it to achieve the goal). The interaction between these
7
elements is used to explain the strength of motivational force driving the individual to achieve a goal. Achievement motivation theory, suggested by Atkinson and Raynor, involves goal desirability (i.e., expectancy to achieve the goal) and movement by inertia (i.e., individuals continue to pursue goals just because this is what they have been doing). An individual’s persistence to attain a goal will be determined by the combination of inertial tendencies resulting from past experiences and the strength of current goal desirability. Some empirical evidence provides support for the persistence or goal-driven explanation. For example, Garland and Conlon (1998) manipulated investment decision and project completion by telling some participants that a project designed to revamp a company’s manufacturing capabilities was 20% complete and others that a project was 80% complete. Participants in the 80% completion cell were more likely to allocate more resources to the project, leading the researchers to conclude that the participants did so just to get the project over with. Those individuals in the 20% completion preferred withdrawal. Similar studies have been carried out by Garland (1990) and Arkes and Blum (1985). However, no self-reported data studies explain that persistence is in fact behind the cognitive processes that lead to escalation. Furthermore, if persistence by itself is an explanation for escalation, it is not clear why individuals who are not responsible for the original allocation decision, but given the task of making a follow-up allocation decision (i.e., are given a set goal), refuse to allocate further resources to a given project and thus do not become entrapped. Brockner (1992) argued that none of the theories presented (e.g., self-justification, decision dilemma, or prospect theory) could explain the escalation phenomena in its
8
entirety. All of them provide explanations of entrapment in some circumstances. However, because the self-justification explanation has received most support in past research, the current study will focus on the self-justification explanation of escalation phenomenon as it relates to the desire to save face and the need to justify one’s actions. Self-justification Self-justification has been systematically explored in compliance-gaining research. According to Kelman (1961), compliance can occur when a person accepts influence from another individual or from a group in hopes to achieve a favorable reaction from others. The individual is willing to be influenced because he or she wants to receive certain awards or avoid punishment that the other individual or group controls. Compliance gaining has been extensively studied using forced compliance experiments in which participants are induced to advocate a counter-attitudinal position to motivate them to advocate views that violate their own attitudes (Eagley & Chaiken, 1993). For example, Festinger and Carlsmith (1959) conducted an experiment in which participants were asked to complete a boring task. Afterward, half of the participants were offered $1 and the other half were offered $20 to mislead a confederate that the task was enjoyable and interesting. Those participants offered higher incentives did not change their attitudes, whereas, the participants paid just $1 came to believe that the boring task they completed was indeed enjoyable. In other words, participants offered low incentives had to justify spending time completing a boring task by perceptually biasing the task as enjoyable. Other experiments on self-justification have included writing attitude-discrepant essays (Scheier & Carver, 1980) and eating a disliked food (Zimbardo, Weisenberg,
9
Firestone, & Levy, 1965). These experiments have shown that individuals faced with personal negative consequences—such as engaging in a task they do not want to perform—as a result of counter-attitudinal acts, with no external rewards to compensate for the dissatisfying nature of the experimental task, tend to change their opinions on an attitudinal issue to cognitively reduce any negative outcomes resulting from their behavior. In short, people feel compelled to justify their previous behavior and defend themselves from negative consequences through the perceptual biasing of behavioral outcomes. In addition to negative consequences, two other factors affect the biasing of outcomes within forced-compliance situations. The first factor is the commitment to behavioral consequences, which are irrevocable or not easily changed (Brehm & Cohen, 1962). The second factor is a sense of having at least a moderate degree of choice in one’s behavior. In other words, an individual should feel personally responsible for the negative consequences of his or her behavior (Cooper, 1971). A number of studies have investigated factors influencing the nonrational escalation of commitment to a previous action. For example, Staw (1976) examined how responsibility for negative consequences affects nonrational escalation of commitment. The author examines the process of escalating commitment through conducting the simulation of making a business investment decision. The participants were divided into two groups, one with high responsibility and one with low. Those participants in the group with high responsibility were asked to allocate funds to one of two corporate divisions of an organization. The participants were then told that, after 5 years of the initial allocation of funds, their investment turned out to be either successful or
10
unsuccessful. Subsequently, the participants were asked to make a second allocation to either one of the divisions. The low-responsibility group did not have to make a prior decision as to which corporate division was most deserving of the funds. This group was presented with the entire financial decision case, including successful or unsuccessful results of the first allocation. The low-responsibility participants were then asked to make the second allocation decision. The study results indicate that participants in the highresponsibility condition, who were told their first allocation was unsuccessful, committed a significantly higher amount to the original division in the second allocation than the participants in the low-responsibility group. Thus, Staw concluded that personal responsibility for negative consequences leads to increased investment of resources in a previously chosen course of action. In another study, Rubin and Brockner (1975) investigated the passage of time as a factor influencing the entrapment bias. They demonstrated how closeness to achieving a goal affects escalation of commitment to the futile investment of resources. The scholars conducted an experiment in which the passage of time could be viewed as an investment or as an expense. The experiment participants were given an initial sum of money (the “initial stake”) and an opportunity to win a bigger some of money (the “jackpot”). To win the jackpot, the participants had to solve a series of crossword puzzles. Some of the crossword puzzles were so difficult that they required the use of a dictionary. To obtain the dictionary (i.e., scarce resource), participants had to wait in line until the resource became available, which it never did. As the time passed, the amount of money in both the jackpot and initial stake decreased. Thus, the longer the time spent waiting for the dictionary, the greater the expense to the participants, and the greater the investment.
11
The authors manipulated three variables: rate of decrease of the jackpot, availability of a chart providing information about the decrease (i.e., length of waiting time), and information concerning participants’ illusory position in line for the dictionary. The study results showed that entrapment in the waiting condition was high overall. Entrapment was particularly high when the jackpot decreased slowly, when the decrease chart was not available, and when participants thought that they were first in line for the dictionary instead of third. This study illustrates that goal completion could be used as a reason to self-justify a commitment to failing action. Overall, the discussion of self-justification research illustrates that there are five conditions that could lead to an entrapment bias: negative consequences (i.e., failure of the original decision), commitment to behavior, perceived degree of freedom in commitment to this behavior, responsibility for the negative consequences, and desire for goal completion. Gaps in Self-Justification and Entrapment Research Although a number of studies have provided support for self-justification theory (see Brockner [1992] for review), only a few studies take into consideration social factors and the way these social factors could affect escalation. Brockner and Rubin (1985) identify four types of social variables that can influence escalation: group influence, behavioral modeling, presence of an audience, and competition against a social (labor vs. management) or nonsocial (waiting for a bus) entity in an attempt to achieve the goal. One factor of particular interest to the current research is the presence of an external audience, but only a few studies (Brockner, Rubin, Fine, et al., 1982; Brockner, Rubin, & Lang, 1981; Fox & Staw, 1979; Staw & Hoang, 1995) have examined the presence of an
12
audience and its effect on entrapment. Further, no studies have explored the role of cultural variables (e.g., distinguishing between dimensions of face, personal networks) in entrapment situations. Another gap in entrapment research and the self-justification explanation of the escalation phenomena, is a lack of studies that examine cognitive processes that may explain escalation. Three types of research have been conducted to provide support for the self-justification explanation of the escalation phenomena. The most popular type includes studies that “operationalize feedback from prior resource allocations and decision makers’ needs to justify those prior resource allocations; the typical finding is that escalation is greatest when both feedback is negative and justification needs are high” (Brockner, 1992, p. 49). But these studies do not examine cognitive processes— that is, what led the participants to make the decision. These studies measure commitment to a previous decision, which is often operationalized as the amount of money allocated to the previously chosen course of action. In addition to the study described earlier by Staw (1976), studies by Davis and Bobko (1986), Fox and Staw (1979) and Schoorman, Mayer, Douglas, and Hetrick (1994) also operationalized entrapment as the amount of money invested and did not examine cognitive processes leading to the decision. The second type of research providing support for the self-justification explanation of entrapment explores other manifestations of behaviors (i.e., besides escalated commitment) resulting from the self-justification motive (Brockner, 1992). For example, Conlon and Parks (1987) found that individuals, to provide self-justification, tend to look for retrospectively focused information to make a decision regarding
13
subsequent reallocation. The researchers argue that retrospective focusing results, in part, because exoneration and justification necessitate a plausible explanation of how and why a negative consequence occurred. The third type of studies uses self-report data of psychological states to establish that behavioral escalation is indeed related to individuals’ self-justification needs (Brockner & Rubin, 1985; Strube & Lott, 1984). For example, findings from the study by Brockner and Rubin suggest that individuals who manifest escalation behavior also produce self-reports that indicate a need for justification of prior allocation (e. g., “I had already invested so much, it seemed foolish not to continue,” or “Once I had invested a certain amount, I had to keep going; otherwise all of that previous investment would have been a waste” [p. 148]). Ross and Staw (1986) cite the results of an unpublished study by Bazerman, Schoorman, and Goodman (1980) who investigated the relationship between participants’ behavioral manifestation of escalation and self-reports of other measures related to self-justification. The findings indicated that escalation was “associated with the perceived importance of a decision, the extent of disappointment with initial losses, and the perceived interconnectedness of current and past decisions in the situation” (Ross & Staw, 1986, p. 276). These studies focus on psychological states, not cognitive processes associated with escalation. None of the escalation studies examined focused specifically on cognitive processes. In addition, no studies have examined the escalation of commitment phenomenon from a communication point of view, namely focusing on messages used in situations in which individuals persist with continued commitment to the failing course of action. Drummond (1994) analyzed a case study of hiring an incompetent manager. The author
14
used case study research by participant observation that employed “direct observation, questioning, diary keeping and unobtrusive methods” (Drummond, 1994, p. 46). The scholar used quotes to document the case study, however, no systematic analysis of the messages was employed. Overall, the discussion above illustrates four gaps and weaknesses in the study of the entrapment phenomenon that the current research will address. First, there are only a few studies that examine the presence of external audience and its potential to affect entrapment. Second, there are no studies that have examined cultural variables and their potential influence on the commitment to the failing course of action. Third, there are no studies that have analyzed the cognitive processes involved in an escalation of commitment. Fourth, there is no research that analyzes messages communicated in an escalation situation. This research will address these four gaps: audience effect, cultural variables (i.e., types of face concern and personal networks), cognitive processes (i.e., internal and external justification needs), and messages used to justify one’s decision to escalate. Entrapment and Negotiation Entrapment is a cognitive bias that can have a have direct effect on negotiation. According to Putnam and Jones (1982), negotiation is a process in which two or more parties that hold or believe they hold incompatible goals try to reach a mutually acceptable solution by engaging in a give and take interaction. Neale and Bazerman (1985) argue that adopting a view of negotiation as a business decision-making process could help explain failure to reach agreements. The authors propose that cognitive or judgment biases result in reduced negotiator effectiveness in reaching best outcomes and
15
reduced likelihood of attaining agreement. The authors refer to cognitive biases (including entrapment bias) as heuristics that are helpful as shortcuts in an effort to reduce the amount of information to be processed. At the same time, these heuristics can also bias outcomes in systematic ways. In the case of entrapment bias, negotiators tend to escalate their commitment to justify their earlier bids, offers, and proposals, and to avoid the financial and ego losses of coming in second. Neale and Bazerman (1985) provide a specific discussion outlining ways in which escalation and negotiation are related. The scholars suggest that escalation is likely to lead negotiators “to stand firm on their initial offers through the course of successive negotiations” (p. 48). One reason for development of such rigid negotiation positions is the presence of an external audience or constituency. The researchers maintain: The pressure from the constituency may lead to the escalation of commitment [entrapment bias], which impedes the ability of the negotiator to represent their best interests. Further, as both sides accrue losses (e.g., during a strike), both sides are likely to increase their propensity to “hold out” in order to justify their initial positions. (p. 48) Bazerman (1986) offers three additional reasons for the entrapment bias in negotiation. The first reason is that negotiators tend to look for information that is salient with their initial commitment to a position. The second reason is that the negotiators’ judgment is biased to interpret what they observe at the negotiation table in a way that justifies or supports their initial position. The third reason is that the competitive context of a negotiation situation fuels the likelihood of entrapment. The negotiators perceive the notion of unilateral surrendering of a previously stated position or even making smaller
16
demands as defeat; on the other hand, escalation of commitment or entrapment “leaves the future uncertain” (p. 57). Consistent with prospect theory, Bazerman argues that this uncertain future is perceived as more desirable by negotiators than a prospect of certain loss. Bazerman (1986) recommends that, to avoid an entrapment bias in a negotiation context, negotiators have to realize their tendency to justify past actions. They have to remember to evaluate the benefits and costs of persisting with a chosen course of action. He also suggests that awareness of the tendency to escalate can help predict the opponents’ actions and thought processes. When individuals are too psychologically invested in their positions, they tend to increase their demands or hold out on making the concessions for too long. As a result, “a negotiator should avoid pushing opponents into a corner, getting them angry or otherwise making them feel that they can’t afford to give up the struggle” (p. 57). Other approaches to reducing cognitive biases in a negotiation include providing unequivocal feedback regarding one’s performance, improving negotiator selection criteria (i.e., identifying an individual’s tendency to be vulnerable to decisional biases), and developing training protocols to help negotiators eliminate decisional biases. Because the entrapment bias in a negotiation can lead to failure to achieve desired outcomes, this issue is important to understand and empirically explore by examining how entrapment can occur during negotiation. The next sections examine variables (i.e., accountability and face) that may affect entrapment in a negotiation situation.
17
Accountability and Constituency Normative social influence describes the force that compels people to conform to the positive expectations of another. Normative influence implies “real or imagined group pressure, and a person may conform to avoid sanctions, to gain approval or simply to avoid the embarrassment of being different” (Nail, 1986, p. 202). The desire for social approval is related to social normative influence that is associated with compliance (Asch, 1951) and consequently with the self-justification process. One condition that demonstrates normative influence is accountability, or “the extent to which representatives are required to justify their actions, and are going to be evaluated and rewarded by their constituents” (Gelfand & Realo, 1999, p. 721). A constituency can exert normative influence, which can lead to entrapment bias. Accountability and Constituencies in Negotiation Context According to Gelfand and Realo (1999), within a context of negotiation, accountability is activated when negotiators have to provide justification for their actions after the negotiation. Lewicki, Saunders, and Minton (1985) maintain that constituencies control individuals by administering rewards and punishments. Wall and Blum (1991) argue that constituencies tend to exert as strong an influence on negotiators as opponents do. These researchers explain, “The reasons are quite clear; constituents are powerful and will use their power, bringing pressures to bear that keep the negotiator committed to the constituents’—group’s or organization’s position” (p. 282). Lewicki, Saunders, and Minton (1985) posit that accountability occurs under two conditions: (a) when the constituency can observe and judge the individual’s performance, and (b) when the
18
individual is aware that the constituency depends on him or her to achieve positive outcomes. Research suggests that accountability to a constituency significantly influences negotiators’ behaviors; negotiators accountable to a constituency tend to behave more competitively and bargain tougher than those in low- or no-accountability conditions (Bartunek, Benton, & Keys, 1975; Carnevale, Pruitt, & Britton, 1979; Gruder, 1971; Gruder & Rosen, 1971; Organ, 1971). For example, findings from Neale (1984) indicate that when constituents evaluate negotiators’ bargaining and determine payoffs, the negotiators are more likely to reach impasses and less likely to concede. Lewicki, Saunders, and Minton (1985) maintain that the individual’s desire for consistency is often amplified by a desire to save face and maintain an illusion that he or she is in control in front of an audience. The authors suggest that such behavior is a result of unwillingness on the part of the individuals to admit an error or failure, particularly when the other party might interpret doing so as a weakness. They state, “The mere presence of an audience, particularly one that can easily observe the negotiator, motivates a negotiator to seek a favorable evaluation from that audience and to avoid an unfavorable evaluation” (p. 295). Neale and Bazerman (1985) and Rubin, Kim, and Peretz (1990) argue that the pressure from the audience may lead to the escalation of conflict (entrapment) and impede the negotiator’s ability to represent the best interests of this audience. Furthermore, they state that the pressure generated by constituencies toward adherence to a certain position may not be in the best interest of either the constituency or the
19
negotiator and may impede the negotiators’ ability to represent the constituency’s best interests. Accountability, Constituency, and Entrapment Bias A number of studies suggest that accountability contributes significantly to entrapment (Fox & Staw, 1979; Staw & Hoang, 1995). Staw (1981) attributes this effect to external justification. For example, Staw and Hoang (1995) studied National Basketball Association (NBA) draft choices and found that coaches gave greater playing time to players who were selected earlier in the draft and who were paid more money independent of the individual’s performance. In this case, owners, fans, and media could be considered as the external audiences to whom coaches are responsible. In another study, Fox and Staw (1979) conducted a simulation in which business students played the role of business executives who had to make funding allocation decisions under high or low conditions of job security. The scholars manipulated the popularity of a recently implemented policy at the time that the initial allocation decision was made. Fox and Staw hypothesized that those participants whose job was threatened or who implemented an unpopular policy would be motivated to protect themselves against failure. Judgment by the board of directors regarding allocation simulated the accountability condition. The participants in the job insecurity condition were informed that they were temporarily assigned the role of a vice president; those participants in the job security condition were told that their job was permanent. Half of the respondents were told that the board of directors was dissatisfied with their initial funding decision (resistance condition) and the remaining half were informed that the board was pleased with the initial resource allocation (no resistance condition). The participants were asked
20
to reallocate their funds. The findings indicated that when a course of action led to negative results (resistance condition), the participants in the low job security condition were more likely to escalate their commitment to the losing course of action by making greater subsequent resource allocations. In other words, negative input from the constituency (the board of directors) in combination with being dependent on this constituency for job security resulted in greater resource allocation. These studies did not examine the possibility that one of the underlying reasons for entrapment in the presence of constituency is an individual’s need to save face. Based on the discussion presented in the sections above, this research puts forth the following hypotheses: Hypothesis 1 (H1): Entrapment will be more likely when a constituency is present. Hypothesis 2a (H2a): Justification would be greater when respondents are required to report their behavior. Hypothesis 2b (H2b): When individuals have to report their behavior, they will have greater need for internal and external self-justifications leading to entrapment. Hypothesis 3 (H3): Presence of constituency is likely to lead to greater internal and external justification. Face Goffman (1955, 1959) introduced the concept of face or the desire to create and sustain positive identities in the eyes of significant others. This desire motivates individuals to appear strong and capable, and to avoid situations in which they could be publicly embarrassed in front of an audience. Goffman argued that people will try to prevent loss of face even if they have to incur costs. Facework is communication aimed at
21
enhancing or repairing face that has been damaged. Brown (1968, 1970) adopted Goffman’s definition of face and facework and conceptualizes facework as face-saving and face-restoration. Face-saving is defined as an attempt to prevent another from causing an individual to appear foolish or incapable to significant others and facerestoration is conceived as an individual’s attempt to seek revenge from another after the other has already damaged face. Brown and Levinson (1987) discuss two types of face: negative face and positive face. Negative face is conceived as an individual’s desire to maintain his or her autonomy (i.e., the need not to be imposed upon), whereas positive face is conceptualized as the need to seek inclusion or approval from significant others. In addition, Brown and Levinson (1987) charge that, because of the interdependent nature of social relations, individuals can threaten or support other people’s face and protect their own face. TingToomey (1988) adds another dimension to the concept of face: self and other. Self-face implies concern for one’s own image and self-interest, whereas other-face implies concern for another’s image and other-interest. Ting-Toomey (1988) proposes four types of face maintenance. First, self-negative face is associated with one’s need to protect one’s autonomy from other’s infringements. Second, other-negative face implies the need to demonstrate respect for other person’s need for autonomy. Third, self-positive face is concerned with the need to defend one’s need for inclusion. Fourth, other-positive face is defined as the need to support the other person’s need for association. Face-Saving and Negotiation The concept of face has been addressed by scholars examining conflict in the negotiation context. Wilson and Putnam (1990) state that face goals exert strong
22
influence on the negotiation process. These scholars state that, during negotiation, negotiators have a certain image of themselves that they would like to preserve. Scholars (e.g., Thompson, Nadler, & Kim, 1999; White, Tynan, Galinsky, & Thompson, 2004; Wilson, 1992) have examined the role of face threats in negotiation. According to White et al. (2004), positive face threats include situations in which one of the parties is embarrassed, disrespected or criticized. Negative face threats, on the other hand, are associated with situations in which one party desires to avoid imposition. White et al. (2004) argue that both positive and negative aspects of face could be threatened during a negotiation. However, they charge that research conducted by Cupach and Carson (2002) and Cupach and Messman (1999) suggests that positive face is more important for relationships, and therefore is more likely to be associated with integrative outcomes. Some direct face threats include nonnegotiable offers (Tjosvold, 1977), criticizing an opponent’s position (Brown & Levinson, 1987), pressure (Thompson, Nadler & Kim, 1999), and resisting making a concession (Tjosvold & Huston, 1978). In addition to direct threats to face by an opponent, certain situational factors increase face threat in a negotiation (White et al., 2004). One of these factors is constituency. The presence of constituency is likely to heighten face concerns. The discussion in the previous section illustrates that the presence of an audience can exert significant influence on an individual in a variety of contexts. Brown and Garland (1971) suggest two reasons why presence of an audience increases face-saving behavior. First, audiences can provide evaluative feedback directly to an individual. Second, an audience could also communicate its evaluation to others. Brown (1968) maintains that negotiators seek to communicate a positive image of themselves not only to the counterparts
23
involved in a bargaining process but also any other audience interested in the outcome. According to Brown (1968), “The latter may include the group the bargainer represents or in which he holds membership; they may be ‘real’ or imagined, physically present at or absent from the setting in which the bargaining occurs” (p. 109). Similarly, Wilson and Putnam (1990) argue that face-maintenance goals become more salient when negotiators receive feedback from and feel highly accountable to constituents. Stevens (1963) links face saving to concession making. He states that bargainers face a dilemma when they have to make concessions to reach an agreement. He maintains that the act of making the concession itself can be perceived by others as a sign of weakness, which can cause face loss and possibly increase attempts of exploitation by the other party. Therefore, the facesaving and economic motives may require mutually incompatible responses by individuals. The study conducted by Brown (1968) showed that bargainers are willing to sacrifice their own economic gain just to inflict worse losses on their counterparts, especially when they have been made to appear foolish in front of a salient audience. Brown conducted an experiment in which the face saving was induced by informing the participants that they would be observed by an audience while participating in a bargaining task. The task was based on a two-person trucking game in which each player runs a trucking company and has to move his or her truck over a road system to a final destination. The faster one reaches the final destination, the more money the player will earn. The game is set up in such a way that for each player to win the most money, they have to cooperate. In addition, each player has control over a tollgate through which the
24
counterpart’s truck has to pass. Players are required to either charge the opponent a specified toll or grant him or her free passage. During the first ten trials, a confederate controlled the tollgate and systematically charged the participant high tolls resulting in significant monetary losses for the participant. Then, the participants received feedback from the audience members, who were supposedly, but not actually, observing the interaction from behind a one-way mirror. The participants received feedback that either said that they looked foolish and weak (e.g., “Bolt was out to beat Acme and he really made Acme look like a sucker”) or that they looked good (e.g., “Bolt made Acme pay a lot of high tolls but Acme looked good because he tried hard and played fair). The control group received no feedback. In the second round, the participants had control over the tollgate. The participant had two choices: He or she could either retaliate against the confederate or increase his or her own winnings. The results indicated that those participants who received negative feedback from the audience were much more likely to retaliate against their counterparts than participants who were provided with positive feedback. Further, participants in the negative feedback condition were willing to lose money to restore face. Those in the positive feedback condition proceeded to maximize their profits. The results from the post-experimental questionnaire showed that the negative feedback participants were more concerned with looking strong than were the other participants. Face-Saving and Entrapment Bias Only two studies have explored the relationship between face-saving behavior and the entrapment bias. Brockner, Rubin, and Lang (1981) conducted two experiments to explore how the presence of an audience affects entrapment. In the first experiment, the
25
participants were told that they are about to participate in a study on decision making. The participants were asked to make an investment decision. Half of the participants were told about the virtues of investing conservatively (cautious condition) and the other half was instructed about advantages of investing a larger amount (risky condition). To investigate the role of face saving, experimenters assigned half of the participants to the large-audience condition (experimenter plus two confederates who were supposedly working for psychology professors, and who were interested in observing the procedures because they were considering using them in their own experiments) and the second half to the small audience condition (just the experimenter). The researchers also hypothesized that, because social anxiety is related to self-presentation, the participants with low social anxiety would be less influenced by the experimenter’s instructions than would those with high social anxiety. The results showed that investments were less than half in the cautious condition than in the risky condition. The face-saving analysis showed that (1) the instructions had a greater effect on subjects with high rather than low social anxiety, and (2) individuals with high social anxiety who participated in front of a large audience were more influenced by the instructions than were individuals with low social anxiety who participated in front of a small audience. (p. 68) In the second experiment, the procedures were the same as the ones in the first experiment, except Brockner, Rubin, and Lang (1981) orthogonally varied the importance of costs and rewards. In the high-cost-importance condition, the participants were given a chart with information regarding their costs at various points in the
26
experiment, and they were asked to plot their progress toward the jackpot. In the lowcost-importance condition, the participants were not given a chart, and they did not have to track their progress. The results showed that the reward importance variable did not have any effect. The effect of perceived costs was mediated by participants’ concern about how their investment would make them look in front of others. Participants in the high-cost-importance condition quit earlier and stated that they became less entrapped to make a desirable self-presentation (e.g., “I thought that it would look good to quit”). The authors concluded, “Individuals will become more or less entrapped to the extent that doing so will portray them in a more favorable light” (p. 78). The second study exploring face and entrapment was conducted by Brockner, Rubin, Fine, et al. (1982). These researchers varied decision makers’ face-saving concerns and the point in time at which these concerns arose. The scholars conducted two experiments. The first experiment dealt with perceived importance of costs and rewards associated with continued investment and did not involve face-saving concerns. In the second experiment, the face-saving manipulation was present and was operationalized through the presence of an evaluative audience. The participants were told that the purpose of the experiment was to simulate gasline-waiting procedures. The participants were asked to make a decision about whether to stop or continue waiting in a gas line. Half of the participants were told that experts in decision making would be observing and evaluating their behavior (evaluative condition) and the other half were told that they would be watched by a non-evaluative audience (non-evaluative condition). In the evaluative condition, the participants were provided with a form that the observers would supposedly use to evaluate them. The form included
27
such items as how much participants “appeared to have the situation under control,” “were being made to look foolish,” and “were using their money wisely” (p. 258). Brockner, Rubin, Fine, et al. also manipulated the timing of observations so that half of the participants were observed by the audience from the start of the experiment (early condition) and the other half were observed after a significant portion of resources was already invested (late condition). The results showed that the presence of an evaluative audience had no effect on participants’ resource allocation early in the process, but it did have a significant effect on investment behavior later in the process. More specifically: When the audience was described as “experts in decision making,” subjects high in public self-consciousness (or social anxiety) became less entrapped than those low on these dimensions. When the audience consisted of individuals who “wished to simply observe the experimental procedure,” however, high public self-consciousness (or social anxiety) individuals were significantly more entrapped than lows. Moreover, these interaction effects occurred when the audience was introduced late, but not early, into the entrapment situation. (pp. 247–248) Given the small number of studies related to, but not directly testing, face-saving behavior in the entrapment situation, further investigation is needed of how the desire to save face, particularly in the presence of constituency, can lead to entrapment. The current research examines the effect of face on entrapment. Because scholars have indicated (White et al., 2004) that concerns for positive face are more important to negotiators than the desire to avoid an imposition (concern for negative face), this study
28
will primarily focus on concerns for self- and other-positive face. This research puts forth the following hypotheses: H4: Individuals with greater self- and other-positive face concerns will be more likely to become entrapped. H5: Concern for saving face will be greater when the individuals have to report their behavior. H6: Overall other- and self-positive face concerns that an individual has will lead to greater need to justify one’s actions in a scenario leading to entrapment. H7: Concerns for saving face will be greater when constituency is present. Personal Networks During the last decades, network analysis has emerged as a way to examine social structures (Emirbayer & Goodwin, 1994). According to Valente (1995), networks are patterns of support, advice, friendship, and communication that are shared by members of a social system. Personal networks include an individual, people who are in contact with this individual, social relationships between the individual and other people, and social relationships between the people in contact with the individual. Emirbayer and Goodwin (1994) argue that personal network analysis is “one of the most promising currents in sociological research” (p. 1412). In fact, the authors argue that network analysis “offers a more powerful way of describing social interactions than do other structural perspectives that focus solely on the categorical attributes of individuals and collective actors” (p. 1413). According to Wellman (1983), network analysis explains social behavior as the result of individuals’ involvement in structured social relations as opposed to common attributes and norms they possess. Network
29
analysis looks at patterns of relations (Burt, 1986) as opposed to such individual attributes as political affiliation, gender, social status, or ethnicity. The study of personal networks benefits not only sociological research; it can also be applied to the field of communication. Over the past decades, individualismcollectivism and independent and interdependent self-construal have dominated crosscultural research as explanations for cultural differences in types of relationships. Both sets of constructs have been widely used to explain the relationship between individuals and their relevant others (Hofstede, 1980; Markus & Kitayama, 1991). The concepts of collectivism and inter-dependent self-construal emphasize the importance of and dependence on relevant others. On the other hand, the concepts of individualism and independent self-construal emphasize self-reliance and focus on oneself. The concept of face has been closely related to individualism-collectivism. TingToomey (1988) posits that members of individualistic cultures and people with independent self-construals are more concerned with self-face maintenance than members of collectivistic cultures or people with interdependent self-construals. On the other hand, individuals in collectivistic cultures are more concerned with mutual-face and other-face maintenance. Further, Ting-Toomey argues that members of individualistic cultures tend to use autonomy-preserving strategies, and members of collectivistic cultures tend to use approval-seeking strategies when managing conflict. However, Fiske (2002) criticized individualism-collectivism research, and Levine et al. (2003) offered criticism of the independent and interdependent self-construals pointing out that both sets of constructs have significant operationalization and measurement drawbacks. Massett (1999) suggested that personal networks provide more
30
accurate representation of the relevant others than the two sets of constructs. Massett outlined three advantages of using personal networks as a framework for examining individuals’ differences. First, personal networks allow for more accurate representation of one’s social interactions with relevant others within a society. Second, examination of the individual’s personal networks, when he or she can identify his or her own set of significant others, reduces the problems associated with arbitrary definitions of groups. Third, examination of personal networks can provide insight into quality and frequency of communication among individuals in a given society or culture. Networks not only allow for more careful description of relevant others but also allow for the measure of range and strength of the relationship with relevant others (Granovetter, 1973; Marsden & Campbell, 1984). According to Emirbayer and Goodwin (1994), network range is the number of an individual’s ties to other people and strength of ties is “the relative frequency, duration, emotional intensity, reciprocal exchange and so on which characterize a given tie or set of ties” (pp. 40-41). Granovetter (1982) provided an overview of studies that have examined tie strength. Among other applications, tie strength was a good indicator of social mobility, in which different degrees of tie strength were positively related to the outcomes of job search efforts. The most common and best indicator of tie strength is the closeness of a relationship (Marsden & Campbell, 1984). Massett (1999) conducted a study comparing the effects of culture and otherorientation on personal communication networks and behavioral intentions in the United States and Mexico. The study focused on individuals’ health networks and diabetesrelated behaviors. Massett found that Americans were significantly more independent and
31
that Mexican respondents were significantly more interdependent. In addition, the comparison of personal networks indicated that, although U.S. participants nominated more people in their networks, Mexican respondents nominated more family members, communicated with members of networks more often, reported more high-context communication with network members and indicated greater degree of closeness with network members than did the U.S. respondents. These findings suggest support for using personal networks to investigate the relationships between individuals and their significant others. More specifically, the results of the network comparison parallel closely the characteristics traditionally associated with interdependent (Mexico) and independent (U.S.) self-construals; that is, individuals with interdependent self - construal are more likely to have strong, close tie networks and individuals with independent selfconstrual are more likely to have weak, loose-tie networks. Given a plausible relationship between self-construals and networks (Massett, 1998), and taking into consideration that self-construals have an effect on face concerns (Oetzel & Ting-Toomey, 2003; Ting-Toomey, 1988), it is possible that personal networks affect face concerns. In addition, as the previous section established a connection between face concerns and entrapment, an indirect relationship may exist between personal networks and entrapment bias with face concerns as a moderating variable. Further, because a concept of a network range is at the core of network analysis (Granovetter, 1973), the size of an individual’s network along with the strength of ties within the network is expected to affect face concerns and entrapment bias. This research argues that the range and strength of one’s personal networks will predict face-saving behavior, which in turn will affect the entrapment bias. The greater
32
the size of the network and the stronger the ties in the network, the more the individual should be concerned with protecting self-positive face because of his or her desire to look good to the members of the network in order to maintain his or her inclusion in that network. In addition, individuals with stronger ties and larger personal networks are expected to be more concerned with protecting other-positive face, because they feel compelled to protect the other party’s interests. Further, individuals with larger, strong-tie personal networks are expected to be more prone to entrapment in the presence of a constituency, because members of the constituency are likely to be members of the negotiator’s professional network. Therefore, the negotiator may feel that his or her reputation, as well as reputation of the constituency, are at stake and persist with a failing course of action. Along the same lines of reasoning, these individuals are also more likely to feel more compelled to justify their behavior than individuals with smaller, loose tie networks. Based on the discussion presented above, this research will test the following hypotheses: H8: Individuals with more expansive and strong ties in their personal networks will be more concerned with their self-positive face. H9a: When participants have expansive personal networks, entrapment will be positively associated with protecting the constituency’s positive face. H9b: In the presence of constituency, individuals with larger, strong-tie personal networks will be more prone to entrapment than the individuals with smaller, weak-tie networks.
33
H10: In the presence of constituency, individuals with larger personal networks and strong ties would report greater justification for making their decisions. Luck Although unrelated to the theoretical rationale of this research, the concept of luck is discussed here, because luck was used in the method to foil the nature of the study; that is, the participants were told that the study is about luck. It was necessary not to disclose the nature of the research so as not to bias the outcomes. To identify measures of luck suitable to be employed in this research, a literature review of studies examining luck was conducted. Summarized below is a brief overview of luck research. Luck measures used in this dissertation are discussed in chapter 2. Traditionally, luck is discussed within the framework of conditions related to expectations of control and success (Darke & Freedman, 1997a). According to Darke and Freedman (1997a), “luck is a random, uncontrollable factor which should have little effect on future expectations” (p. 487). Luck is typically considered to be an external, unstable factor that can explain achievement outcomes or social events (Darke & Freedman, 1997b). Social learning theory of personality developed by Rotter (1955) posits that perceptions of control decrease if events are attributed to luck or other people (external locus of control) and increase when events seem to be a result of a person’s own actions (internal locus of control). Overall, individuals deceive themselves as having less control if they believe that luck is involved. An attributional model explaining the origin of perceived control identifies four causal factors to which failure or success is usually attributed: luck, task difficulty, effort, and ability (Weiner et al., 1972). In addition to the locus of control aspect specified by
34
the social learning theory, attributional theory introduces stability—a degree of consistency across time and situations. According to the attributional model, luck is considered to be an external and unstable factor. That is, the model predicts that any success attributable to luck should be regarded as uncontrollable; in addition, future successes cannot be predicted on the basis of luck. Both the social learning theory and the attributional model assume that individuals have rational views about causes of luck; they see luck as unstable and external. However, studies conducted by Darke and Freedman (1997a) suggest that not all people have rational views about luck, some hold irrational beliefs that luck is a stable factor that influences events in their favor. Individuals who hold the latter view of luck tend to have positive expectations for the outcome of future events. This result supports findings from another study conducted by Darke and Freedman (1997b) that demonstrates that those who believe in luck are more confident and positive about future success. Further, a study conducted by Wohl and Enzle (2002a) shows that individuals who believe they are lucky also believe that they can use their luck intentionally to influence the outcome of chance events. For example, research participants acted “as though luck could be transmitted from themselves to a wheel of fortune and thereby positively affect their perceived chance of winning” (p. 1388). Another study conducted by Wohl and Enzle (2002b) indicates that individuals’ differences in self-perceived luck influence their future behaviors. More specifically, Wohl and Enzle (2002b) conducted an experiment in which participants were assigned to either near win or near loss conditions. The experimental task consisted of playing a computerized slot-machine style wheel of fortune game. In the near-loss condition, the wheel appeared to almost stop at
35
the Bankrupt section but inched forward towards a small payoff section. In the near-win condition, the wheel appeared ready to stop at the Jackpot section, but came to a complete stop at the small payoff section. The near escape of big loss led participants to deem themselves lucky. As a result, they were more prone to gamble more in comparison to participants who experienced a near big win. The discussion above suggests that luck is a plausible disguise to be used in this research. Specifics of the experimental procedures are detailed in the chapter 2: Methods. Summary of Research Hypotheses Figure 2 provides visual representation of study hypotheses and relationships among study variables. Circles represent constructs that incorporate more than one variable. For example, personal networks are comprised of the size of the network and influence of network members have on the person or tie strength.
36
Personal Networks (size and influence)
Self-positive face
Constituency Other-positive face Entrapment Reporting Internal Justification
Overall selfand otherpositive and negative face concerns
External Justification
Figure 2. Research Hypotheses. Overall, this research attempts to provide support for the following hypotheses: Hypothesis 1 (H1): Entrapment will be more likely when a constituency is present. Hypothesis 2a (H2a): Justification would be greater when respondents are required to report their behavior. Hypothesis 2b (H2b): When individuals have to report their behavior, they will have greater need for internal and external self-justifications leading to entrapment. Hypothesis 3 (H3): Presence of constituency is likely to lead to greater internal and external justification.
37
Hypothesis 4 (H4): Individuals with greater self- and other-positive face concerns will be more likely to become entrapped. Hypothesis 5 (H5): Concern for saving face will be greater when the individuals have to report their behavior. Hypothesis 6 (H6): Overall other- and self-positive face concerns that an individual has will lead to greater need to justify one’s actions in a scenario leading to entrapment. Hypothesis 7 (H7): Concerns for saving face will be greater when constituency is present. Hypothesis 8 (H8): Individuals with more expansive and strong ties in their personal networks will be more concerned with their self-positive face. Hypothesis H9a (H9a): When participants have expansive personal networks, entrapment will be positively associated with protecting the constituency’s positive face. Hypothesis 9b (H9b): In the presence of constituency, individuals with larger, strong-tie personal networks will be more prone to entrapment than the individuals with smaller, weak-tie networks. Hypothesis 10 (H10): In the presence of constituency, individuals with larger personal networks and strong ties would report greater justification for making their decisions. Research Question Finally, because currently there are no known studies that have examined the entrapment phenomenon from a communication point of view—that is, no studies have examined messages used by individuals in situations potentially leading to entrapment—
38
this study poses the following research question: In a situation leading to entrapment, are individuals likely to use social accounts to explain their behavior? Bies (1987) defines social account as “a verbal strategy employed by a person to minimize the apparent severity of the predicament or convince the audience that the wrongful act is not a fair representation of what the actor is ‘really like’ as a person” (p. 294). Sitkin and Bies (1993) state that social accounts are used to influence an individual’s perception of responsibility for an action and motivation for an action, and to paint the unfavorability of an action. Sitkin and Bies (1993) identify three broad categories of accounts: (a) mitigating responsibility, (b) legitimizing the action by appealing to some higher-order values or norms, and (c) reframing outcomes. Mitigating, or causal, accounts suggest that the situation forced the individual to take the action. Sitkin and Bies state, “by suggesting that the offending party had no other alternatives to the chosen action, a social account claiming mitigating circumstances should reduce the amount of blame attributed to the party” (p. 350). When individuals use the second type, exonerating accounts, they try to explain their actions by placing them within a broad normative framework that will give legitimacy to their motives. The third type, the reframing account, attempts to put the actions in the best possible light by suggesting to the offended party the appropriate context for interpretation. To this end, this research will examine the messages produced by the research participants to determine whether they are likely to be social accounts and explore which of the social accounts are used most often.
39
CHAPTER 2 Method This chapter describes the process of developing research questionnaires and data collection procedures. The first section presents describes the experimental design. The second section provides a description of the study sample. The third section describes in detail the recruitment procedures. The fourth section discusses the experimental procedures and manipulations. Finally, the fifth section provides a detailed discussion of the development of study measures. Experimental Design One of the purposes of this research is to examine whether the presence of constituency leads to entrapment. The presence of constituency is expected to pose face threats that lead to entrapment. The research also hypothesizes that accountability, in this case reporting in person to the constituency, will pose greater face threat than nonaccountability. In addition, face concerns are expected to be moderated by the size and strength of personal networks. There are three conditions essential to simulate accountability (Gelfand & Realo, 1999). First, constituents need to have control over rewards. Second, individuals must justify their performance. Third, individuals must be evaluated by the constituency. For the purposes of this study, accountability is manipulated using two requirements in the accountability condition: (a) presence of a constituency who has control over rewards and evaluates performance, which meets the requirements of the first and third conditions; and (b) reporting, so that individuals have to explain their behavior to the constituency in person, meeting the requirement of the second condition. To this end, two experimental
40
conditions are employed: presence of constituency and reporting on the outcomes of the experiment. Overall, the respondents were assigned to one of the four conditions (see Table 1): 1. Working in groups and reporting in-person to the group (COND I: constituency; reporting). 2. Working in groups, with no reporting (COND II: constituency; no reporting). 3. Working individually and reporting in-person to the researcher (COND III: no constituency; reporting). 4. Working individually (COND IV: no constituency; no reporting). The research design is 2X2 (constituency [yes/no] by reporting [yes/no]), with face and justification as mediating variables, personal networks as a moderating variable, and entrapment as a dependent variable. Table 1. Study Design (N = 236) Constituency Yes Yes Reporting No n = 58 n = 62 n = 57 No n = 59
Study Sample The study sample consisted of 236 undergraduate students majoring in communication enrolled in a large East Coast university (N = 236). The overwhelming majority of the participants (97.4%) were full-time undergraduate students. The remaining 2.6% were part-time undergraduate students. Of the participating students
41
2.5% were freshmen, 33.9% were sophomores, 39.4% were juniors, and 23.7% were seniors. Ethnic composition of the sample was as follows: White, non-Hispanic or Caucasian (64.4%), African American or Black (14%), Hispanic (6.4%), Asian or AsianAmerican (6.8%). In addition, 6.8% of the respondents reported their ethnicity as “Other.” Participants were between 17 and 33 years old (M = 20.3 years, SD = 1.86, median = 20). Female participants constituted 80.9% of the sample and male participants constituted the remaining 19.1%. Such an imbalance in female to male student ratio is not unusual, as this is a growing trend in communication classes, where the majority of majors are female. A slight majority of the participants (53%) reported being employed. The most often-cited categories of employment were sales, administrative/clerical (e.g., secretary, administrative assistant, account clerk), and service industry (e.g., waiter, nanny, chef) positions. Recruiting Procedures and Research Assistants Participants volunteered to take part in the experiment. They received extra credit in their communication courses for their participation. In addition to extra credit, each participant received ten raffle tickets for four raffle drawings; each drawing was worth $50. Each ticket was worth $1. Participants were not allowed to purchase tickets outside of the study. When the study was completed, the researcher administered one lottery and awarded $50 to the winner. The study was announced by each class instructor (see Appendix B). The announcement was made several times during the course of the semester. The prospective participants were told that if they would like to receive extra credit for the class, one option available to them was to participate in an experiment investigating an individual’s
42
luck and decision-making processes. An alternative written assignment was offered to those students who wanted extra credit but did not want to participate in research. The instructor passed around the sign up sheet. Those individuals who chose to participate in the study received an informed consent form (see Appendix D) and a number of questionnaires that they were asked to fill out prior to arriving at the experimental laboratory. The questionnaires included demographic information (see Appendix E), a personal network instrument (see Appendix F), a network influence instrument (see Appendix G), and a face scale instrument (see Appendix H). Two female undergraduate students were recruited to help the researcher and act as the “confederate constituency” in conditions in which constituency was present. Several hours were spent training the undergraduate students on how to act and “exert pressure” on a participant. The confederates were encouraged to put full responsibility for deciding to continue or withdraw from the task on the research participant. Sample statements used by the confederates included the following: “It is all up to you, but it sure would be nice to win some money,” “Win us some money,” “I hope you win, I can use some extra cash” and “If you think it is time to withdraw, it is your call, but extra cash for Christmas presents would be nice.” In addition, on several occasions, the confederates also acted as research assistants, administering instruments to the participants in the noconstituency conditions. During their training the undergraduate assistants also received instructions on how to use the instruments, and they practiced administering them. In addition, the researcher sat in during the first four experiments conducted by the assistants. The researcher was available for questions when the research assistants
43
administered the study. The undergraduate students helping the researcher received three independent-study credits for their involvement in this research project. Data Collection and Experimental Procedures Data were collected between June and December 2004. Prior to starting the experimental task, the participants received an information packet that included details of the task (see Appendix J) and an informed consent form (see Appendix D). In addition, they were asked to fill out the following questionnaires: manipulation check and luck (see Appendices K and L, respectively). After participating in the experiment, the participants were asked to provide a narrative explaining the experiment outcomes and complete the entrapment bias questionnaire and the face-work scale instrument (see Appendices M, N, and O). At the beginning of each experiment, the researcher explained the procedures (see Appendix I) and distributed the information packet (see Appendix J). The participants were informed that the experiment was designed to study luck. Prior to participation in the study, the respondents were told that the purpose of the experiment was to study luck “but some features of the research will not be revealed until the research is completed. This will ensure that study results are not biased” (see Appendix I). During the pre-experiment instructions, the researcher also mentioned that results of many past studies investigating luck indicate that how well one performs in this particular experimental task is “a strong indicator of how lucky one is in other areas of life, such as finding a good job, finding a partner for life, etc.” After participants completed the pre-task instruments, the researcher mentioned, “Even though some people feel unlucky, their performance on this task does not depend on it. And, furthermore,
44
even if you feel that you are unlucky, there is still about one in ten chances that you will be successful in accomplishing the task.” This statement was designed to preclude participants from quitting too early because they consider themselves unlucky. In social research, not fully disclosing the nature of the study is not uncommon and sometimes is necessary to prohibit biases and confounding variables (Schrag, 2003). According to Schrag (2003), this type of deception is justified when validity of research is at stake, when risk to participants is minimal, and when the researcher has a debriefing plan. In this research, all three criteria were met. First, the researcher could not disclose that the investigation concerns entrapment, because then the participants would have been aware of the potential for escalation of commitment and would be unlikely to become entrapped. Second, the risk to the participants was indeed minimal, and they were not likely to object once they were told about how they were deceived. Third, at the end of the experiment the researcher debriefed the participants and explained to them the true purpose of the experiment, defined entrapment, and identified the experimental conditions and manipulated factors. The debriefing also stressed that the experiment did not “have anything to do with luck and your [participant’s] performance is not indicative of how lucky you are as a person.” The debriefing form can be found in Appendix O. Task Description The participants were told that the task would consist of drawing ping-pong balls out of a box. The experimenter explained that the box contained 100 ping-pong balls, of which 90 were white and 10 were red, and that the participant’s luck would be measured by the number of red balls he or she drew from the box. In fact, there were no red balls in the box. The participants were told that the goal of the task is to draw five out of the ten
45
red balls that are mixed in with white balls. The participants were informed that they could make up to 20 attempts to draw five red balls. It was also emphasized that the experiment was structured so that only drawing five red balls counted as a successful completion of the task; in other words, drawing one, two, three, or four red balls constituted failure to complete the task. All of the participants received extra credit for participation regardless of their performance and were reminded that their performance would not affect receiving extra credit. All of the participants started the experiment with ten raffle tickets. The participants were told that if they successfully completed the task (i.e., drew 5 red balls) they would receive an additional 20 raffle tickets, bringing the total number of raffle tickets to 30. However, because there were no red balls to draw, it was impossible for anyone to receive 30 tickets. Constituency Manipulation Half of the participants were told that they would be working on their own (no constituency). In the no constituency condition, the respondents were told that after they received instructions, they would be escorted into the experiment room where they would perform the drawing task. They received a pay-off sheet reflecting potential gains and losses (see Appendix J). More specifically, the pay-off sheet detailed that for every white ball drawn, the participant would lose one raffle ticket (of the ten tickets he or she started with) and for every red ball drawn he or she would win two raffle tickets. Each participant was informed that if he or she were to get “in the negative,” he or she would have to pay for the number of negative points accumulated, such that one point was worth one dollar. Because the participants started with ten raffle tickets each and could make up to 20 draws, the maximum penalty possible was $10. However, no money was actually
46
collected in the end. The participants were also informed that regardless of how many points he or she lost, if all 5 red balls were drawn by the end of the task, the person would receive 20 raffle tickets with no penalties. For example, if the respondent had negative points at the time of drawing all 5 balls, she or he would not have to pay and would receive all 20 raffle tickets for successfully completing the task. The other half of the participants were told that they would be working in groups of three participants (constituency condition). These groups actually consisted of two confederates (constituency) and one participant. By the time the participant arrived at the lab, the confederates were already waiting. The researcher pretended that she did not know them and took down their names and classes they were enrolled in along with the participant’s information for the purposes of providing this information to his or her instructor to notify the instructor that the student should receive extra credit for participating in the study. In the constituency condition, the participants were told that, although they were working as a group, only one person would be performing the task, and that person would be decided by number drawing. The participants in the constituency condition were told that the experimenter was interested in investigating “how an individual’s luck affects groups; particularly, because there have been studies that have shown that not only individuals could be lucky, but also groups.” To vote on the person to perform the task, the participants were asked to draw a number out of a hat; the person who got number “3” had to perform the drawing. To ensure that the participant was the one who would be performing the task, all of the folded pieces of paper had the number “3.” The confederates had an agreement that one
47
would always say that she had “1” and the other one would always say that she had “2.” Because of these arrangements the participant was always the one to perform the task. Each participant in the constituency condition was instructed in front of the confederates that, for every round in which the participant failed to draw a red ball, the constituency would lose one raffle ticket per person and the participant would also lose one raffle ticket. On the other hand, for every round in which the participant did draw a red ball, the constituency would win two tickets each, and the participant would win two tickets. If the participant got “in the negative,” losing more raffle tickets than the number held by each person, the number of points that the constituency would lose would be the same as the number of points lost by the participant. For example, if the participant drew eleven white balls—that is, one negative point—he or she would lose $1, and each member of the constituency would also lose $1. The maximum penalty for each member of the constituency and the participant was $10. Furthermore, to meet the requirement that the constituency control rewards, the participants were instructed that, at the conclusion of the task, the constituency would determine an amount (from $0 to $5) to be awarded to the participant for his or her performance (this award is in addition to the extra credit that all participants would receive regardless of their performance). The experimenter told the participants that she would pay him or her the amount indicated by the constituency. Modeling after the procedures used by Organ (1971), the participant was told that the constituency would also receive compensation, but the amount or payment structure was “left unspecified so as not to bias their decisions on how much to award.”
48
Reporting Manipulation In each accountability condition, half of the participants were asked to report on their performance in the task (reporting condition). The other half of the participants were told that, after they finished the experimental task, they would complete some questionnaires and would be free to go (no reporting condition). Thus, prior to the experiment, half of the participants were informed that they would report in person either to the constituency (one-fourth of the participants) or to the researcher (one-fourth of the participants) at the end of the task. All participants were asked to write a paragraph justifying his or her strategy and statements he or she would use to explain the success or failure in performing the task (outcome narrative). Respondents in the reporting condition were told that they would have to complete the outcome narrative before facing the constituency or the researcher. In fact, the participants did not have to report to the constituency or researcher. As soon as the respondents completed the post-experimental questionnaires they were debriefed by the researcher and were free to go. The information packet for the reporting/constituency condition read: “As a group representative you will perform the drawing task on behalf of your group. At the conclusion of the task, you will write a short essay (one or two paragraphs) summarizing the outcome of the task and explaining your strategy and the results. You will use this essay to report to the group members after you are finished with the drawing. After the explanation is presented to the group members, they will carefully weigh the information and decide how much money to award you for your performance.” The information packet for the reporting/no constituency condition read: “You are to perform the experimental task on your own. At the conclusion of the task, you will write a short essay
49
(one or two paragraphs) summarizing the outcome of the task and explaining your strategy and the results. You will use this essay to report to the researcher after you are finished with the drawing.” In the no reporting/constituency condition, each participant was told that the experimenter would inform the group members of the outcomes of the task and the group members would make a decision regarding the monetary reward to be received by the participant. In the no reporting/no constituency condition, the respondent was told that after completion of the task and the post-experimental questionnaires the participant would receive a debriefing and would be free to go. The statements produced by the participants were collected by the experimenter and analyzed for the types of messages most often used by the respondents in an attempt to explain their behavior. Dependent Variable The dependent variable in this study was the entrapment bias. For the purposes of this study, the entrapment bias was conceived of as a participant’s commitment to continuing the task. It was operationalized by the number of drawing rounds (the greater number of rounds, the greater the entrapment bias) the participant chose to conduct. Analysis of descriptive statistics of the total number of draws shows M = 9.78, SD = 5.17 and median = 9.00. Number of draws ranged from 0 (2.5% of the respondents) to 20 (12.7% of the sample). About 75% of the participants withdrew by the 10th round. Approximately 8% of the sample lost all 10 tickets, and about 31% stopped at round nine and thus, were left with just one raffle ticket. The remaining 25% continued the task
50
incurring the monetary penalty, with approximately 18% (of the total sample) incurring a penalty over $5. Final Materials and Development of Measures For anonymity purposes, each participant selected a unique four-digit number that he or she wrote on all study instruments (see Appendix C). When the respondents handed back their questionnaires, the researchers made sure that all surveys had the matching identification numbers. The respondents filled out the materials in the following order: (1) prior to arriving to the lab, the respondents signed an informed consent form and completed the demographics questionnaire, personal network questionnaire, network influence instrument, and face scale; (2) after receiving the experimental instructions, the respondents signed another informed consent form and completed pre-task questionnaires including the manipulation check and luck scale; (3) after finishing the experimental task, the participants completed a series of post-task questionnaires, which included an outcome narrative, an entrapment scale, and face scale (identical to the one they completed prior to participating in the research study); and (4) after the respondents turned in their post-task questionnaires they read a debriefing form and received explanations about the study from the researcher. Institutional Review Board (IRB) approval was received to conduct this study. The IRB approval letter can be seen in Appendix A. The discussion of measures used in the study (presented below) follows the order in which each instrument was administered. Table 3 summarizes reliability coefficients (Cronbach’s ) for the scales used in this study. Informed consent form. The informed consent form assured the respondents of the confidentiality of their performance, informed them that there were no long-term effects
51
associated with this research, and explained that they were free to terminate the session at any time without penalty (see Appendix D). In other words, if a participant chose to withdraw from the task, he or she would still receive extra credit for participation. The respondents were asked to sign the form twice. The first time, the participants received the form along with other questionnaires when they signed up for the study. They were asked to bring the signed form and completed questionnaires to the lab on the day of their participation in the study. The second time, the participants were given the form and asked to sign it after they received experimental task instructions. Demographic questionnaire. The demographics questionnaire contained questions regarding participant gender, age, employment status, student status and standing, and race (Appendix E). Personal networks questionnaire. Prior to arriving at the lab, the participants were asked to complete a personal networks questionnaire (see Appendix F). The questionnaire used items developed by McAllister and Fischer (1978). The questionnaire included a total of 11 items. Items 1 through 8 covered various aspects of an individual’s life. Examples of questions included in the instrument are “Who would care for your home if you were to go out of town?”; “With whom do you talk about school or work decisions?”; “With whom do you engage in social activities?”; and “From whom would or could you borrow a large sum of money?” Descriptive statistics for these questions are summarized in Table 2. Item 9 asked respondents to provide the number of all adult members in their household. As the respondents identified members of various networks, they were asked to indicate either the name of the network member or the relationship of the person to the participant (e.g., my father). The questionnaire allowed the respondents
52
to nominate up to 15 network members for each of the questions. However, in case there was anyone else important to them that they would have liked to nominate, item 10 offered them an opportunity to list those people. The final question asked the respondents to count the total number of people they named on all of the lists and asked them not to count a person multiple times. However, only 79% of the sample provided answers to this question, reporting 15 people as an average number of network members (M = 14.64, SD = 6.69 and median = 14). The number of members in the network ranged from 1 to 38. Because about 20% of the respondents failed to indicate the total number of people nominated in their networks, a new variable NETWORK was created in which a total number of people nominated in the network was calculated such that every person nominated, even if they were nominated more than once, were counted (Massett, 1999). The range of the network size for this new variable was 78, with minimum of 9 and maximum of 87 people and with the mean of 39.94 (M = 39.94, SD = 15.18, median = 39). In addition, for analysis purposes the NETWORK variable was dichotomized, with the network size between 1 and 39.9 considered as small and network size between 40 to100 was considered large. This variable was called NETSIZE.
53
Table 2 Network Questionnaire: Summary of Descriptive Statistics for Total Number of People Nominated in the Networks for Items 1 through 8 Question (Total number of people listed) Who would care for your home if you were to go out of town? With whom do you talk about school or work decisions? Who, if anyone, has helped with household tasks in the last three months? With whom do you engage in social activities (e.g., going to a movie, having dinner)? Whom do you talk with about your interests or hobbies? With whom do you talk about personal worries? Whose advice do you consider in making important decisions? From whom would or could you borrow a large sum of money?
Mean (M) 4.46
Standard Deviation (SD) 2.90
Median 4.00
4.75
2.41
5.00
2.85
1.84
3.00
6.70
3.63
6.00
5.39
3.12
5.00
4.36 3.87
2.43 1.98
4.00 4.00
2.60
2.00
1.36
Because closeness is an important predictor of strong ties, three items were added to the questionnaire (Marsden & Campbell, 1984). Two of these questions, “Please indicate how well do you know each of the people you mentioned?” and “How close do you feel to the person?”, were used by Massett (1999). Responses to these questions were not used in the final analysis; instead, the influence scale (see section below) was used to measure strength of the relationships within the network. The third question, used by
54
Marsden and Campbell (1984), read, “How would you characterize this person?” In other words, the respondents were asked to indicate whether each person named was a relative, an acquaintance, a good friend, or a very close friend. Overall, the respondents nominated mostly their close friends (M = 18.01, SD = 10.09, median = 17) and relatives (M = 16.75, SD = 7.85, median = 16) as members of their networks. The friends category (M = 2.76, SD = 3.58, median = 1) received just a few nominations, and almost no one nominated acquaintances (M = .71, SD = 1.57, median = 0) as part of their network. Influence scale. In addition to the personal network questionnaire, an influence scale was also administered. The scale was developed to measure how close the participants were to the members of their network by asking them how much influence people nominated in their network have on the respondent’s life. This 12-item instrument used a 7-point Likert scale in which 1 indicated “no influence” and 7 indicated “strong influence.” The respondents were asked to indicate how much influence people nominated in the networks would have on their job choice, education choice, social life, personal life decisions, social habits, outward appearance, interests and hobbies, neighborhood choice, ways to resolve conflict, political position, and decision to undergo a serious medical procedure. Reliability analysis indicated that the scale had strong reliability (Cronbach’s items. = .82). Table 3 summarizes descriptive statistics for the scale
55
Table 3 Influence Scale: Summary of Descriptive Statistics Standard Deviation (SD) 1.36
Question Overall how much influence do people you listed in Q1 through 10 have on a type of job you might choose (e.g., type of company, position)? Overall how much influence do people you listed in Q1 through 10 have on your decision to pursue education beyond college? Overall how much influence do people you listed in Q1 through 10 have on your social life (e.g., choice of friends)? Overall how much influence do people you listed in Q1 through 10 have on personal development (e.g., books you read, music you listen to)? Overall how much influence do people you listed in Q1 through 10 have on your personal life decisions (e.g., whom to date, sex conduct)? Overall how much influence do people you listed in Q1 through 10 have on your social habits (smoking, alcohol consumption)? Overall how much influence do people you listed in Q1 through 10 have on your outward appearance (e.g., clothing)
Mean (M) 4.83
Median 5.00
5.08
1.57
5.00
5.23
1.41
6.00
4.61
1.52
5.00
4.78
1.58
5.00
4.85
1.63
5.00
4.27
1.54
5.00
56
Question Overall how much influence do people you listed in Q1 through 10 have on your interests and hobbies? Overall how much influence do people you listed in Q1 through 10 have on your neighborhood choice? Overall how much influence do people you listed in Q1 through 10 have on ways you choose to resolve conflict? Overall how much influence do people you listed in Q1 through 10 have on political position you might take (e.g., support a candidate, party, voting behavior)? Overall how much influence do people you listed in Q1 through 10 have on your decision to undergo a serious medical procedure (e.g., surgery)?
Mean (M) 4.27
Standard Deviation (SD) 1.48
Median 5.00
4.26
1.75
5.00
4.49
1.58
5.00
3.69
1.84
4.00
4.93
1.78
5.00
According to Levine (2005), in communication research confirmatory factor analysis (CFA) is preferred to exploratory factor analysis (EFA) because in comparison to exploratory factor analysis, “CFA provides stronger evidence for dimensionality than EFA because EFA can under-factor correlated constructs and because model fit is typically tested with CFA” (p. 337). Therefore, throughout this research CFA was used to confirm scales employed by the study. CFAs were performed using principal components extraction and varimax rotation. To this end, confirmatory factor analysis (CFA) was used to confirm a composite measure of influence that people nominated in the networks have on the participants’ lives (INFLUENC). This initial factor analysis
57
yielded three factors. However, loadings for the two of the factors were weak. To create a factor score, the items were forced into one factor. The resulting factor score was used in analyses discussed in the chapter 3, Results. The overall combined network size (NETWORK) was then multiplied by the overall influence measure (INFLUENC) to create an overall measure of the size and strength of the nominated network (NTWRKINF). Face scale. The participants were asked to complete a face scale developed specifically for this experiment (see Appendix H). The scale was administered twice, once prior to arriving to the lab and the second time after the experiment. The scale was adapted from the instrument developed by Wilson and Kunkel (2000), who examined face threats in relationship issues using a series of scenarios with follow-up questions. For the purposes of this research, instead of scenarios, one-sentence items were created. The instrument was developed using four dimensions: self-positive face, other-positive face, self-negative face, and other-negative face. The 24-item instrument used a 7-point Likert scale in which 1 indicated “strongly disagree” and 7 indicated “strongly agree.” Among items examining self-positive face were questions such as “I worry about how other people judge me when it comes to my physical appearance” and “In general, it is important to me that people do not think that I am nosy.” The reliability analysis of items measuring this dimension yielded strong reliability (Cronbach’s = .73). In
addition for the purposes of analysis, confirmatory factor analysis was used to create a composite measure of self-positive face (SELFPOS). The initial factor analysis yielded one factor with all of the loadings above .50. The resulting factor score was used in analyses using self-positive face discussed in the Results chapter.
58
Examples of items looking at other-positive face included “Overall, when I talk to people, it is important to me that what I say does not make them look inadequate” and “I don’t like to get into arguments with people because it might make them look uncooperative.” The reliability analysis of items measuring this dimension yielded strong reliability (Cronbach’s = .74). Review of the reliability of specific scale items
suggested that one item, “When I talk to people, I want to make them feel comfortable discussing issues with me,” should be deleted. The new reliability analysis yielded stronger reliability (Cronbach’s = .75). Confirmatory factor analysis was used to create
a composite measure of other-positive face (OTHERPOS). The initial factor analysis yielded one factor with all of the loadings above .50. The resulting factor score was used in analyses discussed in the Results chapter. Items “In general, it is important to me that others feel that I’m an independent person” and “When my close friend does me a favor, I worry about the fact that I will be obliged to return the favor in the future” illustrate examples of questions measuring selfnegative face. Review of the reliability of specific scale items measuring this dimension suggested that one item, “Overall, I dislike when people give me an advice when I did not ask for it,” should be deleted, resulting in an improved reliability coefficient (Cronbach’s = .71). Confirmatory factor analysis was used to create a composite measure of selfnegative face (SELFNEG). The initial factor analysis yielded two factors. However, loadings for one of the two factors were weak. To create a factor score, another confirmatory factor analysis was performed in which the items were forced into one factor. All of the new loadings were above .50. The resulting factor score was used in analyses discussed in the Results chapter.
59
Finally, examples of items examining other-negative face included “When I ask my close friend for a favor, I worry that he/she might find it hard to say ‘no’” and “When I ask my close friend to do something for me, I worry that he/she might feel pushed into agreeing with what I want.” Original reliability analysis of items measuring this dimension revealed that item “In general, I do not feel comfortable sharing my problems with other people because I’m afraid I’m going to overburden them” should be deleted, resulting in an improved reliability (Cronbach’s = .96). Confirmatory factor analysis
was used to confirm a composite measure of other-negative face (OTHENEG). The initial factor analysis yielded one factor solution with strong loadings, all above .70. The resulting factor score was used in analyses discussed in the Results chapter. Overall, for the purposes of analysis, four new variables were created: SELFPOS, OTHERPOS, SELFNEG, and OTHERNEG. Table 4 summarizes descriptive statistics for the scale items. Table 4 Face Scale: Summary of Descriptive Statistics Standard Deviation (SD) 1.66
Question When I ask my close friend for a favor, I worry that she/he might feel obliged to comply with my request When I ask my close friend for a favor, I am concerned that he/she might feel like they have to say "yes" to my request When I ask my close friend for a favor, I worry that she/he might find it hard to say "no"
Mean (M) 3.84
Median 4.00
3.69
1.67
3.00
3.68
1.68
3.00
60
Question When I ask my close friend to do something for me, I'm concerned that she/he might feel like she/she has to go along with my request When I ask my close friend to do something for me, I worry that she/he might fee pushed into agreeing with what I want In general, I do not feel comfortable sharing my problems with other people because I'm afraid I'm going to overburden them In general, it is important to me that others feel that I'm an independent person When I ask somebody for a favor, I worry about the fact that I will feel indebted to the person In general it is important to me to be self-sufficient because I do not like owing anything to other people In general it is important to me that people in my life realize that I'm capable of making decisions myself When my close friend does me a favor, I worry about the fact that I will be obliged to return the favor in the future Overall, I dislike when people give me an advice when I did not ask for it
Mean (M) 3.54
Standard Deviation (SD) 1.61
Median 3.00
3.30
1.60
3.00
3.00
1.79
3.00
5.24
1.36
5.00
3.49
1.65
3.00
5.13
1.53
5.00
5.73
1.17
6.00
2.98
1.59
2.98
3.85
1.64
4.00
61
Question Overall, when I talk to people, it is important to me that what I say does not make them look inadequate In general, when I give advice to a person, it is important to me that she/he does not feel like I'm implying that she/he has difficulty handling the situation Overall, when I give an advice to my friend, it is important to me that my friend does not feel like I'm implying that she/he does not understand the consequences of his/her actions In general, when my friend inadvertently breaks promise, I don't want him/her to feel like I think that she/he is person who never honors his/her commitments I do not like to get into arguments with people because it might make them look uncooperative When I talk to people, I want to make them feel comfortable discussing issues with me I worry about how other people judge me when it comes to my physical appearance I want people to find me physically attractive
Mean (M) 4.71
Standard Deviation (SD) 1.41
Median 5.00
4.83
1.41
5.00
4.66
1.50
5.00
4.17
1.47
4.00
3.18
1.37
3.00
6.04
1.08
6.00
4.76
1.59
5.00
5.85
1.16
6.00
62
Question In general, when a group of colleagues is having a conversation, I like to be included in the conversation In general, it is important to me that people do not think that I am nosy In general, it is important to me to make a positive impression on people In general, it is important to me that people think that I am smart and capable
Mean (M) 5.66
Standard Deviation (SD) 1.20
Median 6.00
4.86
1.50
5.00
6.29
.80
6.00
6.29
.89
6.00
Manipulation check. Manipulations were checked through questions in the information packet given to the participants prior to performing the experimental task. After reading the materials, the participants were asked questions to evaluate their understanding of the instructions (see Appendix K). The questionnaire included such items as “Will you be evaluated?” (1 = yes, 2 = no); “How closely do you believe your actions will be examined?” (1 = not at all closely to 7 = very closely); “How much will you be required to justify your outcomes and strategy?” (1 = not at all required to 7 = very much required); “To receive 20 extra raffle tickets at the end of the task, how many red balls do you have to draw?” (“please circle the appropriate answer: 1, 2, 3, 4, 5”); “What is the maximum number of drawing rounds you can conduct?” (“please circle the
appropriate answer: 3, 7, 10, 15, 17, 19, 20, 22, or as many as I want”); “If you receive
two negative points, what will be the monetary penalty you will incur?” (“please circle the
appropriate answer: $1, $2, $3, $4, $5, $6, $7, $8, $9, $10”), and only for the constituency
63
condition “If you receive three negative points, what will be the monetary penalty incurred by each group member?” (“please circle the appropriate answer: 1, 3, 4, 6, 9, 10”). Two of the experimental manipulations, reporting Yes/No and constituency Yes/No, were checked using an independent-samples t-test. The t-test was performed on items designed to show meaningful differences based on the assigned conditions. Question one (“Will you be evaluated?”) contrasted the difference between the conditions in which the participants had to report in person to either constituency or the researcher (M = 1.07, SD = .26) with the conditions with no reporting (M = 1.38, SD = .49). Results indicated a significant difference between these two conditions t(231) = 6.17, p < .01. In addition, significant differences were also observed in reporting (M = 4.96, SD = 1.43) versus no reporting (M = 4.28, SD = 1.75) conditions on question 2, “How closely do you believe your actions will be examined?” (t[233] = 3.25, p < .01). Furthermore, the results showed that those participants in the no reporting conditions (M = 3.36, SD = 1.78) indicated that they believed that they would not have to justify their outcomes and strategy (question 3) in comparison to those in reporting conditions (M = 4.90, SD = 1.48; t[232] = 7.14, p < .01). For the constituency (M = 1.05, SD = .22) versus no constituency conditions (M = 1.39, SD = .49), only one item (“Will you be evaluated?”) was hypothesized to show significant differences between the conditions. The null hypothesis was rejected (t[231] = 6.70, p < .01), such that those participants in the constituency conditions knew that they would be evaluated. The majority (94.40%) of the individuals assigned to the constituency/reporting condition accurately reported that they would be evaluated by the group members. All (100%) individuals assigned to the no constituency/reporting condition correctly
64
indicated that they would be evaluated by the researcher. In addition, 98.70% of the participants correctly indicated that they would need to draw five red balls to receive 20 extra raffle tickets. Approximately 94% of the respondents reported that they could conduct a maximum of 20 rounds of drawing, as they were instructed. Furthermore, 83% of the participants accurately pointed out that if they were to lose 12 raffle tickets, they would incur a $2 monetary penalty. The results also suggested that the majority of the participants (89.30%) understood that if, on the first four tries, they were to draw only white balls, they would be left with 6 raffle tickets. The data indicated that 81.60% of the individuals assigned to the constituency conditions understood that if they were to lose 13 raffle tickets, each group member would incur a $3 monetary penalty. Luck scale and scenarios. Because the participants were told that the experiment was about luck, to make the situation more believable the participants were asked to fill out a number of luck scales and scenarios borrowed from studies investigating luck (Darke & Freedman, 1997a; Wohl & Enzle, 2002a). The items from these studies were placed into one instrument called “Luck Questionnaire” (see Appendix L). Although luck was not part of the experimental manipulation, reliability of the scales was assessed. The first scale included in the Luck Questionnaire was called Belief in Good Luck (BIGL), a 12-item-scale developed by Darke and Freedman (1997a). This scale was designed to measure the belief that luck is a stable and personal trait. The questionnaire consisted of 6-point Likert scale items ranging from 1 (strongly agree) to 6 (strongly disagree). For example, items 1, 4, and 10 respectively asked, “Luck plays an important part in everyone’s life,” “I believe in luck” and “Even the things in life I can’t control tend to go my way because I’m lucky.” Darke and Freedman reported strong reliability
65
(Cronbach’s
= .85). The reliability analysis using data from the current research study = .78). The review of the scale items
also yielded good reliability (Cronbach’s
suggested that item 7, “It’ s a mistake to base any decisions on how lucky you feel,” and item 12, “Luck is nothing more than random chance,” should be deleted. The reliability analysis was conducted without items 7 and 12 resulting in higher reliability (Cronbach’s = .88). To make the case that this was a study about luck more convincing, three short scenarios about luck were also included in the experimental materials. The scenarios were borrowed from Darke and Freedman (1997a). Originally, the scenarios were used to establish the external validity of the BIGL scale. For example, scenario number 1 read, “If you were walking down a street that was full of people and someone dropped a $20 bill in the middle of the crowd, do you feel that you would: (1) most certainly find it; (2) probably find it; (3) have a slightly better than even chance of finding it; (4) have no feeling one way or the other; (5) have a slightly better than even chance of not finding it; (6) probably not find it; (7) most certainly not find it.” Reliability analysis of these scenarios, however, showed that they were not reliable (Cronbach’s = .26).
In addition to the BIGL scale and luck scenarios, a questionnaire developed by Wohl and Enzle (2002a) was adopted and slightly modified for this study. This questionnaire consisted of four 7-point scale items, which included such items as “When it comes to games of chance (gamble), usually my chances of winning are” (1 = very bad to 7 = very good); “How often do you play games of chance (gamble)?” (1 = never to 7 = more than once a week); “To what extent do you feel that luck is a quality of the person or a quality of the situation?” (1 = quality of situation to 7 = quality of the person), and
66
“If you were to say that you were ‘lucky’ in terms of some event, to what extent do you mean that you a lucky type of person or that a lucky thing happened to you?” (1 = A lucky thing happened to me and 7 = I am a lucky type of person). These items were poorly interrelated (Cronbach’s = .50).
Low reliabilities for the scenarios and Wohl and Enzle’s (2002a) questionnaires had no implications for the results of the study, because luck was not the focus of this research and there were no research hypotheses associated with luck. However, because the BIGL scale did have strong reliability, for exploratory research purposes, a luck factor score was computed on that scale (LUCKSCAL) using confirmatory factor analysis. First, the analysis yielded two components. However, only two items in the second factor had a loading above .50, so another principal component analysis was performed forcing items to load as one factor. All of the new loadings were above .50, and the factor score from this analysis was used for analyses performed for the current research. Table 5 summarizes descriptive statistics for the BIGL scale items. Table 5 Belief in Good Luck Scale: Summary of Descriptive Statistics Standard Deviation (SD) 1.55 1.55 1.40 1.54 1.41 1.29
Question Luck plays an important part in everyone's life Some people are consistently lucky, and others are unlucky I consider myself to be a lucky person I believe in luck I often feel like it's my lucky day I consistently have good luck
Mean (M) 4.30 4.30 3.94 4.34 3.45 3.37
Median 4.00 4.00 4.00 5.00 3.00 3.00
67
Question It's a mistake to base any decision on how lucky you feel Luck works in my favor I don't mind leaving things to chance because I'm a lucky person Even the things in life I can't control tend to go my way because I'm lucky There is such a thing as luck that favors some people, but not others Luck is nothing more than a random chance
Mean (M) 4.67
Standard Deviation (SD) 1.54
Median 5.00
3.58 2.77
1.25 1.23
4.00 3.00
3.11
1.27
3.00
3.78
1.60
4.00
4.76
1.62
5.00
Outcome narrative. This instrument asked participants to summarize the results of the task and explain the st rategy that he or she chose to pursue while performing the drawing (see Appendix M). The respondents were first asked, “What was the color of the ball you drew?” Then, they were asked, “What did you choose to do next?” to which they could respond that they chose to withdraw from the task or continue drawing. If the respondents chose to continue to draw, they were asked in an open-ended question to explain why they made this decision. Participants were asked the same set of questions for each round of drawing. At the end of the questionnaire the respondents were asked to identify some specific statements that they might used to explain what happened during the drawing task. These data were collected to explore whether individuals are likely to use social accounts to explain their behavior. Coding procedures and analysis of the open-ended questions will be discussed in the Results chapter.
68
Entrapment bias scale. The scale measuring cognitive processes contributing to conflict escalation was constructed specifically for this study. It was administered after the participants completed the experimental task (See Appendix N). The instrument used a 7-point Likert scale in which 1 indicated “strongly disagree” and 7 indicated “strongly agree.” The questionnaire was administered to participants in all conditions. The instrument consisted of 20 items for the constituency conditions and no constituency/reporting condition and 18 items for the control condition. For the control condition, two items were omitted because they were not applicable to the respondents. The scale was developed using four factors: internal self-justification, external self-justification, and other- and self-positive face goals. The internal self-justification measured the need to appear to be a rational decision maker (Aronson, 1968). This portion of the scale consisted of six items (i.e., 1, 2, 3, 4, 5, and 16) including “I feel that my strategy choice was rational,” “I made my decision to continue drawing unemotionally,” “I weighted potential risks and gains carefully prior to making each round of drawing,” “I calculated my odds prior to each drawing,” “I felt that my investment in the process will be justified by the eventual pay off,” and “It was important for me to do well on this particular task.” Review of the reliability coefficients (Cronbach’s = .60) for this dimension revealed that one item, “I made my decision to
continue (or stop) drawing unemotionally,” should be omitted, which improved the reliability for this factor (Cronbach’s = .67). Confirmatory factor analysis was used to
create a composite measure of internal self-justification (INTRNJS). The initial factor analysis yielded one factor with all factor loadings above .50.
69
The external self-justification measured the desire to appear rational to others and to demonstrate to them that persisting in a failing course of action was a correct decision in the long term (Staw, 1976). The external self-justification items (i.e., 6, 7, 8, 9, 10) included five questions, among them were such items as “In general, it was important to me that the group members see me as a rational person” and “I wanted the other group members to understand the logic used in choosing my course of action.” Items administered to the participants in the no constituency condition were adapted for relevancy. For example, “I wanted my group members to feel that my choices were wellcalculated” was adapted to read “If I were to perform this task in a group in which I was a group representative, I would want the group members to feel that my choices were wellcalculated.” To improve reliability coefficients (Cronbach’s = .79) on this dimension,
two items, “Under the circumstances, I believe that the course I pursued had the best potential for earning the most extra raffle tickets for my group members” and “I felt it was important to perform well inn this task for my group members,” were omitted resulting in stronger reliability score (Cronbach’s = .83). Confirmatory factor analysis
was used to create a composite measure of external self-justification (EXTRNJS). The initial factor analysis yielded one factor with strong factor loadings, all of them above .50. In addition, the discussion in the preceding sections regarding face argues that participants who have expansive personal networks are likely to be concerned with protecting group image (other-positive face). Five items (i.e., 11, 12, 13, 14, 15) were developed to measure other-positive face goals. Sample items included “I felt it was important to perform well in this task for my group members” and “I was concerned for
70
my group’s needs.” Again, items were adapted to be applicable to all conditions. For example, the item “I wanted to help my group” was administered to the participants in constituency present conditions; for the no constituency conditions the item was adapted to read, “When working in a team, I want to help my group.” Reliability analysis resulted in strong reliability (Cronbach’s = .87), but also suggested that item 15, “I was worried
what the group members will think about me after the task was completed,” should be removed, which resulted in even greater reliability (Cronbach’s = .91). Confirmatory
factor analysis was used to create a composite measure of other-positive face goals (OTHRPENT). The initial factor analysis yielded one factor with strong factor loadings, all of them above .50. Items measuring the self-positive face dimension (i.e., 17, 18, 19, 20, 21, 22) included “I was worried what [the researcher/group members] will think about me after the task was completed,” “It was important for me to do well on this task,” “I wanted to perform well to make a positive impression on [the researcher/group members],” “I don’t want my group members to be mad at me for loosing some of their raffle tickets and/or money,” “I think this was an unfair task” and “I didn’t want [the researcher/group members] to think that I’m unable to calculate the odds.” However, because the items were largely inapplicable to the control group—that is, by definition, this condition (no reporting/no constituency) did not include any threats to self-positive face—this condition was excluded from the analysis for the control group. Reliability analysis on items for conditions I, II, and III revealed that reliability of the scale for this dimension could be improved (Cronbach’s = .67) if items 19, 21 and 22 were omitted (Cronbach’s
= .80). Confirmatory factor analysis was used to create a composite measure of self-
71
positive face goals (SELFPENT). The initial factor analysis yielded one factor with strong factor loadings, all of them above .50. Overall, to measure how much internal and external self-justification and otherand self-positive face goals contributed to the entrapment bias and to distinguish which of the factors contributed to the entrapment, four new variables were created: INTRNJS (measure for internal justification); EXTRNJS (measure for external justification); OTHRPENT (measure for other positive face), and SELFPENT (measure for own positive face). Again, the self-positive face factor score (SELFPNET) did not include COND IV. Tables 6 through 9 summarize descriptive statistics for the entrapment bias scale items by each condition. Table 6 Entrapment Bias Scale: Summary of Descriptive Statistics for Condition I Standard Deviation (SD) 1.60 1.89
Question
I feel that my strategy was rational I made my decision to continue (or stop) drawing unemotionally I weighed potential risks and gains carefully prior to making each round of drawing I calculated my odds prior to each drawing I felt that my investments in the process will be justified by the eventual pay off Under the circumstances, I believe that the course I pursued had the best potential for earning most extra raffle tickets for my group members
Mean (M) 5.38 5.05
Median 6.00 6.00
3.87
1.96
4.00
2.91 3.96
1.87 1.90
3.00 4.00
4.29
1.69
4.00
72
Question
I wanted the other group members to understand the logic used in choosing my course of action I wanted my group members to feel that my choices were well-calculated In general, it was important to me that the group members see me as a rational person I felt it was important to perform well in this task for my group members I was concerned for my group’s needs I wanted to help my group I did not want to let my group down When performing the task it was important to me to consider the consequences my behavior might have for my group I was worried what the group members will think about me after the task was completed It was important for me to do well on this task I wanted to perform well to make a positive impression on my group members I don’t want my group members to be mad at me for loosing some of their raffle tickets and/or money I think this was an unfair task I didn’t want my group members to think that I’m unable to calculate the odds
Mean (M) 4.91
Standard Deviation (SD) 1.58
Median 5.00
4.80
1.48
5.00
4.65
1.76
5.00
4.94
1.56
5.00
4.75 5.40 5.24 5.04
1.52 1.24 1.42 1.52
5.00 5.00 5.00 5.00
3.73
1.60
4.00
4.44 4.27
1.60 1.72
4.00 4.00
5.35
1.51
5.00
3.93 4.04
1.84 1.68
4.00 4.00
73
Question
I don’t think I understood the instructions well I believe that any person in my position would have behaved the same way I did
Mean (M) 3.00 4.89
Standard Deviation (SD) 1.89 2.00
Median 3.00 5.00
Table 7 Entrapment Bias Scale: Summary of Descriptive Statistics for Condition II Standard Deviation (SD) 1.49 1.45
Question
I feel that my strategy was rational I made my decision to continue (or stop) drawing unemotionally I weighed potential risks and gains carefully prior to making each round of drawing I calculated my odds prior to each drawing I felt that my investments in the process will be justified by the eventual pay off Under the circumstances, I believe that the course I pursued had the best potential for earning most extra raffle tickets for my group members I wanted the other group members to understand the logic used in choosing my course of action I wanted my group members to feel that my choices were well-calculated In general, it was important to me that the group members see me as a rational person
Mean (M) 5.55 5.40
Median 6.00 6.00
4.43
2.04
5.00
3.29 4.48
1.95 1.72
3.00 4.00
4.28
1.79
4.00
5.02
1.63
5.00
4.83
1.71
5.00
5.14
1.33
5.00
74
Question
I felt it was important to perform well in this task for my group members I was concerned for my group’s needs I wanted to help my group I did not want to let my group down When performing the task it was important to me to consider the consequences my behavior might have for my group I was worried what the group members will think about me after the task was completed It was important for me to do well on this task I wanted to perform well to make a positive impression on my group members I don’t want my group members to be mad at me for loosing some of their raffle tickets and/or money I think this was an unfair task I didn’t want my group members to think that I’m unable to calculate the odds I don’t think I understood the instructions well I believe that any person in my position would have behaved the same way I did
Mean (M) 5.09
Standard Deviation (SD) 1.48
Median 5.00
4.86 5.45 5.55 5.21
1.56 1.38 1.37 1.39
5.00 6.00 6.00 5.00
4.29
1.68
4.00
4.57 4.33
1.54 1.46
5.00 4.00
5.07
1.58
5.00
4.14 4.03
1.89 1.73
4.00 4.00
2.60 5.03
1.85 1.57
2.00 5.00
75
Table 8 Entrapment Bias Scale: Summary of Descriptive Statistics for Condition III Standard Question Mean (M) Deviation (SD) Median
I feel that my strategy was rational I made my decision to continue (or stop) drawing unemotionally I weighed potential risks and gains carefully prior to making each round of drawing I calculated my odds prior to each drawing I felt that my investments in the process will be justified by the eventual pay off Under the circumstances, I believe that the course I pursued had the best potential for earning most extra raffle ticket I wanted the researcher to understand the logic used in choosing my course of action I wanted the researcher to feel that my choices were wellcalculated In general, it was important to me that the researcher sees me as a rational person In general, when performing in a group/team environment, it is important to me to perform well for my group members When working in a team, I’m concerned for my group’s needs When working in a team, I want to help my group When working in a group, I do not want to let my team down
5.13 4.95
1.63 1.89
5.00 5.00
4.36
1.93
5.00
3.32 4.13
1.98 1.61
3.00 4.00
3.63
1.90
4.00
3.89
1.74
4.00
3.91
1.71
4.00
3.75
1.70
4.00
6.04
0.97
6.00
5.95
.90
6.00
6.11 6.25
1.06 1.06
6.00 6.50
76
Question
When working in a group, it is important to me to consider the consequences my behavior might have for my group When working in a group, I worry what the group members will think about me after the project is completed It was important for me to do well on this particular task I wanted to perform well to make a positive impression on the researcher I don’t want the researcher to think less of me for losing raffle tickets and/or money I think this was an unfair task I didn’t want the researcher to think that I’m unable to calculate the odds I don’t think I understood the instructions well I believe that any person in my position would have behaved the same way I did
Mean (M) 6.20
Standard Deviation (SD) .92
Median 6.00
5.25
1.47
6.00
4.11 3.57
1.65 1.72
4.00 4.00
3.34
1.67
3.00
3.41 3.34
1.91 1.62
3.00 3.00
2.14 4.20
1.45 1.35
2.00 4.00
Table 9 Entrapment Bias Scale: Summary of Descriptive Statistics for Condition IV Standard Deviation (SD) 1.66 1.61
Question
I feel that my strategy was rational I made my decision to continue (or stop) drawing unemotionally
Mean (M) 5.60 5.57
Median 6.00 6.00
77
Question
I weighed potential risks and gains carefully prior to making each round of drawing I calculated my odds prior to each drawing I felt that my investments in the process will be justified by the eventual pay off Under the circumstances, I believe that the course I pursued had the best potential for earning most extra raffle ticket In general, when I do something it is important to me that others understand the logic I used in choosing my course of action. If I were to perform this task in a group in which I was a group representative, I would want the group members to feel that my choices were well-calculated. In general, when working in groups, it is important to me that group members see me as a rational person In general, when performing in a group/team environment, it is important to me to perform well for my group members When working in a team, I’m concerned for my group’s needs When working in a team, I want to help my group When working in a group, I do not want to let my team down
Mean (M) 4.15
Standard Deviation (SD) 2.29
Median 5.00
2.93 3.82
1.91 1.70
2.00 4.00
4.00
1.77
4.00
4.47
1.73
4.00
5.63
1.21
6.00
5.88
1.04
6.00
6.13
0.89
6.00
5.95
1.05
6.00
6.27 6.38
0.86 0.86
6.00 7.00
78
Question
When working in a group, it is important to me to consider the consequences my behavior might have for my group When working in a group, I worry what the group members will think about me after the project is completed It was important for me to do well on this particular task When working in a group, it is important to me to perform well to make a positive impression on my team members I think this was an unfair task I don’t think I understood the instructions well I believe that any person in my position would have behaved the same way I did
Mean (M) 6.08
Standard Deviation (SD) 0.94
Median 6.00
4.43
1.57
5.00
3.90 5.72
1.89 1.20
4.00 6.00
3.48 2.47 4.10
1.89 1.64 1.56
4.00 2.00 4.00
79
Table 10 Scale Reliability Coefficients Scale Influence scale Face scale Self-positive face Other-positive face Self-negative face Other-negative face Belief in good luck Entrapment bias scale Internal self-justification External self-justification Other-positive face Self-positive face (only conditions I, II and III) 0.80 0.67 0.83 0.91 0.73 0.75 0.71 0.91 0.88 Cronbach’s 0.82
80
CHAPTER 3 Results This chapter presents the results of the study. This research explores a number of factors affecting entrapment bias. The discussion below presents the findings for each of the hypotheses and research question. Data Preparation A number of transformations were performed in attempts to correct skewness in the data while maintaining variance homogeneity. However, no transformation was found that would eliminate skewness. Overall, most variables had no or less than three missing values. Missing values were a result of respondents leaving the items blank, apparently because they either refused to or forgot to answer the questions. Table 11 presents a correlation matrix for the independent and dependent variables.
81
Table 11 Correlation Matrix: Independent and Dependent Variables
Overall face Overall face Overall face Overall face concern: concern: concern: Other concern: Self Other positive Self positive negative face negative face face scale face scale
Total number of draws Constituency (present or not) Reporting (yes or no) Overall face concern: Other negative face Overall face concern: Self negative face Overall face concern: Other positive face scale Overall face concern: Self positive face scale Network size multiplied by strength of network influence Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other Positive Face Entrapment scale: Self positive face (Conditions 1,2 and 3) 0.02 0.13* -0.01 0.04 -0.10 0.07 -0.03 0.07 -0.23** -0.23** -0.20** -0.05
Constituency (present or not)
Reporting (yes or no)
0.01 0.01 0.02 0.10 0.02 0.01 0.08 0.06 0.09 -0.41** 0.34** 0.06 0.01 -0.12 0.04 0.05 -0.03 -0.04 -0.30** -0.05 -0.16* 0.47** 0.26** 0.15* 0.07 0.07 0.05 0.10 0.07 0.12 0.33** 0.30** 0.06 0.08 0.08 0.10 0.04 0.20** 0.34** 0.03 0.13* 0.07 0.20** 0.13* 0.30** 0.13 0.02 0.05 0.16* 0.25** 0.24** .
82
Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other Positive Face Entrapment scale: Self positive face (Conditions 1,2 and 3)
Network size multiplied by strength of network influence 0.03 0.10 0.13 0.13 0.15*
Luck
Entrapment scale: Internal justification
Entrapment scale: External justification
Entrapment scale: Other Positive Face
0.17* 0.06 0.09 0.15* 0.50** 0.31** 0.41** 0.42** 0.60** 0.33**
* Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed).
83
Table 12 Independent Variables: Factor Score Descriptive Statistics for Condition I Variable Mean Std. Deviation
Total Number of Draws Overall face concern: Other negative face Overall face concern: Self negative face scale Overall face concern: Other positive face scale Overall face concern: Self positive face scale Network size multiplied by strength of network influence Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other positive face Entrapment scale: self positive face
10.52 0.15 -0.01 -0.01 -0.07 5.80 -0.03 -0.09 .019 -0.49
4.72 1.04 1.00 1.00 1.04 35.89 0.86 0.97 1.00 1.06
0.26
0.95
84
Table 13 Independent Variables: Factor Score Descriptive Statistics for Condition II
Variable Total Number of Draws Overall face concern: Other negative face Overall face concern: Self negative face scale Overall face concern: Other positive face scale Overall face concern: Self positive face scale Network size multiplied by strength of network influence Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other positive face Entrapment scale: self positive face
Mean 10.41 -0.02 0.03 -0.23 0.14 -0.39 -0.03 0.00 -0.63 0.38
Std. Deviation 5.13 0.98 1.04 0.91 0.99 40.79 1.10 0.94 1.05 0.68
-0.48
0.96
85
Table 14 Independent Variables: Factor Score Descriptive Statistics for Condition III Variable Mean Std. Deviation
Total Number of Draws Overall face concern: Other negative face Overall face concern: Self negative face scale Overall face concern: Other positive face scale Overall face concern: Self positive face scale Network size multiplied by strength of network influence Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other positive face Entrapment scale: self positive face
10.52 0.15 -0.01 -0.01 -0.07 5.80 -0.03 -0.09 .019 -0.49
4.72 1.04 1.00 1.00 1.04 35.89 0.86 0.97 1.00 1.06
0.26
0.95
86
Table 15 Independent Variables: Factor Score Descriptive Statistics for Condition IV Variable Mean Std. Deviation
Total Number of Draws Overall face concern: Other negative face Overall face concern: Self negative face scale Overall face concern: Other positive face scale Overall face concern: Self positive face scale Network size multiplied by strength of network influence Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other positive face
9.03 -0.01 -0.07 1.03 -0.13 1.52 -0.11 -0.12 0.41 0.43
5.49 0.99 1.07 1.16 1.07 42.61 1.08 1.06 0.72 0.71
87
Test of the Hypotheses Hypothesis 1 (H1) posited that entrapment will be more likely when a constituency is present then when it is not present. To test this hypothesis, a one-way ANOVA was performed, with constituency (present versus not present) as the independent variable, and entrapment—measured by the number of balls drawn — as the dependent variable. Presence of constituency did not affect entrapment, F(1, 235) = .07, n.s. Therefore, H1 was not supported. Hypothesis 2a (H2a) posited that internal and external justification would be greater when respondents are required to report their behavior. Two ANOVAs were performed with reporting conditions dichotomized (i.e., reporting yes or no) as the independent variable, and internal and external justification factor scores, derived from the entrapment scale, as the dependent variables. The results indicated no significant effect for internal self-justification, F(1, 228) = .45, n.s. However, a significant main effect was found for external self-justification, F(1, 228) = 22.11, p < .01. That is, although requiring an individual to report his or her behavior had no effect on the need to internally self-justify behavior, the requirement to report did affect individuals’ need to appear rational to others. Therefore, H2a was partially supported. Hypothesis 2 (H2b) stated that when individuals have to report their behavior, their need for internal and external self-justifications will lead to greater entrapment. To test this hypothesis, only cases in which reporting was required were selected (n = 116). A regression analysis was performed with internal and external self-justification factor scores derived from the entrapment scale serving as independent variables, and the total number of draws—or entrapment—as the dependent variable. The results showed that the
88
overall analysis was significant, F(2, 110) = 7.30, p < .01. However, consistent with the finding presented above, only the effect of internal justification was significant, with a negative relationship to entrapment, t(110) = 3.47,ß = -1.94, p < .01. That is, among the individuals who had to report their behavior, the greater need for internal selfjustification led to lower levels of entrapment. The external justification had no effect on entrapment, t(110) = .46, n.s. Therefore, H2b was partially supported. Hypothesis 3 (H3) posited that the presence of constituency would lead to greater internal and external justification. Two ANOVAs were performed with constituency dichotomized (i.e., present or not) as the independent variable, and internal and external justification as the dependent variables. The results demonstrated no significant main effects for either test (internal justification: F[1, 228] = .83, n.s.; external justification: F[1, 228] = 1.83, n.s.). Therefore, H3 was not supported. Hypothesis 4 (H4) stated that individuals with greater self- and other-positive face concerns would be more likely to become entrapped. Prior to testing this hypothesis, a regression analysis was performed using self- and other-positive and self- and othernegative face scale scores derived from the face scale instrument administered prior to the experimental task as the independent variable and total number of draws as the dependent variable. The regression analysis was performed to test if general concern for face affects entrapment. The overall relationship was non-significant, F(4, 232) = 1.24, n.s. However, the results provided marginal support for the relationship between other positive face goals and entrapment, which approached significance, t(114) = 1.86, p = .06, = -.71. That is, the results suggest that the greater the person’s concern for other’s
positive face, the less likely the person is to become entrapped.
89
Next, to test the hypothesis that individuals’ face-saving goals affect entrapment, two regression analyses were performed using self-positive and other-positive face goal (derived from the entrapment scale administered after participants completed the experimental task) as independent variables and the total number of draws as the dependent variable. It should be noted that the analysis for self-positive face was performed only on experimental conditions 1, 2, and 3. The control group (condition 4) was excluded from the analysis because the experimental design was such that individuals in this condition should not have experienced any threats to self-positive face. Because individuals in this condition should not have experienced any threats to selfpositive face, questions about self-positive face differed for this group and this analysis could not be run. The results showed that other-positive face goals were negatively associated with entrapment, F(1, 228) = 9.20, p < .01; = -1.03. This result is consistent with the finding
reported above regarding individuals’ general other-positive face concerns and entrapment. That is, the results indicated that the individuals were less likely to get entrapped if they had significant concerns for other-positive face. However, self-positive face goals did not affect entrapment, F(1, 168) = .35, n.s. Therefore, H4 was partially supported. Hypothesis 5 (H5) posited that concern for saving self- and other-positive face would be greater when the individuals have to report their behavior. To test this hypothesis, two ANOVAs were performed with reporting conditions dichotomized (yes/no) as the independent variable and self- and other-positive face factor scores for entrapment as dependent variables. In self-positive face analysis, only constituency
90
present and reporting conditions (i.e., conditions 1, 2, and 3) were used, because, in the control condition, the participants should have not experienced any threats to self-positive face. The results revealed a reliable main effect across experimental conditions, F(1, 168) = 4.30, p < .05, such that concern for saving self-positive face was greater when individuals had to report their behavior. However, a similar two-way ANOVA with reporting conditions dichotomized (yes/no) as the independent variable and other-positive face as the dependent variable, using participants in all of the conditions, demonstrated no significant main effect, F(1, 228) = .53, n.s. Therefore, H5 was partially supported. Hypothesis 6 (H6) posited that overall other- and self-positive face concerns (as opposed to face concerns related to the experimental manipulation) would lead to greater need to justify one’s action in a situation that could lead to entrapment. To test this hypothesis, four separate regression analyses were performed with self- and otherpositive face, derived from the face scale instrument administered prior to the experimental task, as the independent variables and internal and external justification as dependent variables. The results indicated that face concerns did not affect internal justification (self-positive face: F[1, 226] = .48, n.s.; other-positive face F[1, 225] = .96, n.s.). However, the relationships between external self-justification and other- and selfpositive face were significant (other-positive face F[1, 225] = 9.08, p < .01; self-positive face F[1, 226] = 5.67, p
In criminal law, entrapment is conduct by a law enforcement agent inducing a person to commit an offense that the person would otherwise have been unlikely to commit
ABSTRACT
Title of dissertation:
FACTORS AFFECTING ENTRAPMENT BIAS: JUSTIFICATION NEEDS, FACE CONCERNS AND PERSONAL NETWORKS Anna Karavanov, Doctor of Philosophy, 2006
Dissertation directed by:
Professor Deborah A. Cai Department of Communication
This study explores the link between the entrapment bias and the concept of face (self- and other-positive) and internal and external justification processes. It examines how face-saving concerns and justification needs moderate the entrapment bias in accountability condition (i.e., presence of constituencies and reporting requirements). In addition, this research examines whether the size and influence of personal networks is associated with face-saving behaviors that, in turn, affect entrapment. The research also explores whether overall face concerns have an effect on internal and external selfjustification. Finally, the study explored messages used by individuals in a scenario potentially leading to entrapment. Respondents in the study were 236 undergraduate students majoring in communication enrolled in a large East Coast university. Study participants were assigned to one of the four conditions: (1) constituency, reporting; (2) constituency, no reporting; (3) no constituency; reporting; (4) no constituency; no reporting. The current investigation did not support the findings from previous studies that suggest that justification processes and face concerns lead to entrapment. This study
found that only internal self-justification and other-positive face concerns are related to entrapment, but instead of contributing to entrapment, these aspects prevent individuals from becoming entrapped. Personal networks were demonstrated to have positive effect on both self- and other-positive face concerns, providing empirical support for the value of using personal networks as a predictor of face goals. However, personal networks did not contribute to entrapment. Finally, the study examined messages used by individuals in a situation leading to entrapment, suggesting that when individuals try to explain their behavior, they tend to use causal accounts. Overall, this study has made a contribution to the field of communication by identifying processes and conditions (e.g., concern for other-positive face, internal selfjustification, reporting requirement, no direct observation by constituency, keeping clear record of performance success or failure) that may prevent entrapment bias from occurring. These processes and conditions could potentially improve the outcomes of negotiation with the use of effective communication strategies.
FACTORS AFFECTING ENTRAPMENT BIAS: JUSTIFICATION NEEDS, FACE CONCERNS AND PERSONAL NETWORKS
by
Anna Karavanov
Dissertation submitted to the Faculty of theGraduate School of the University of Maryland, College Park in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2006
Advisory Committee: Professor Deborah A. Cai, Chair Professor Robert Gaines Professor Cynthia Stevens Professor Monique Turner Dr. Leah Waks
©Copyright by Anna Karavanov
2006
ii
DEDICATION
This dissertation is dedicated to the memory of my father, Alexander Karavanov.
iii
ACKNOWLEDGEMENTS
As with any milestone, there are always many people who make the accomplishment possible. It is unlikely that I can name all of the people who have contributed to the accomplishment. Thank you. As this experience comes to an end, I’m amazed that I still have a husband, family and friends who are all still talking to me. Thank you for seeing me through endless frustrations, grumpiness and bouts of irritability. And a very special “thank you” to Dr. Deborah Cai, my advisor. Thank you so much for your support and guidance.
iv
TABLE OF CONTENTS
List of Tables ..................................................................................................................... vi List of Figures .................................................................................................................. viii Chapter 1 Overview........................................................................................ 1
Theoretical Rationale .............................................................................................. 3 Entrapment Bias.......................................................................................... 3 Theoretical Frameworks ................................................................. 3 Self-justification.............................................................................. 8 Gaps in Self-Justification and Entrapment Research.................... 11 Entrapment and Negotiation ..................................................................... 14 Accountability and Constituency.............................................................. 17 Accountability and Constituencies in Negotiation Context.......... 17 Accountability, Constituency, and Entrapment Bias .................... 19 Face ........................................................................................................... 20 Face-Saving and Negotiation........................................................ 21 Face-Saving and Entrapment Bias ................................................ 24 Personal Networks .................................................................................... 28 Luck .......................................................................................................... 33 Summary of Research Hypotheses ........................................................... 35 Research Question .................................................................................... 37 Chapter 2 Method ......................................................................................... 39 Experimental Design................................................................................. 39 Study Sample ............................................................................................ 40 Recruiting Procedures and Research Assistants ....................................... 41 Data Collection and Experimental Procedures ......................................... 43 Task Description ........................................................................... 44 Constituency Manipulation........................................................... 45 Reporting Manipulation ................................................................ 48 Dependent Variable ...................................................................... 49 Final Materials and Development of Measures ............................ 50 Results.......................................................................................... 80 Data Preparation........................................................................................ 80 Test of the Hypotheses.............................................................................. 87 Luck and Entrapment.................................................................... 95 Research Question .................................................................................... 97 Coding scheme and procedures .................................................... 98 Analysis....................................................................................... 108 Findings....................................................................................... 111 Discussion................................................................................. 113
Chapter 3
Chapter 4
v
Summary of the Study Design and Gaps Addressed by the Research.... 113 Research Hypotheses .............................................................................. 117 Implications............................................................................................. 130 Limitations of the Study.......................................................................... 134 Experimental Design and Manipulation ..................................... 135 Personal Networks Measures...................................................... 137 Directions for Future Research ............................................................... 138 Significance of the Study........................................................................ 141 Appendix A: IRB: Approval Letter .............................................................................. 145 Appendix B: Participant Solicitation Announcement.................................................... 146 Appendix C: ID Number Instructions............................................................................ 147 Appendix D: Informed Consent Form ........................................................................... 148 Appendix F: Personal Networks Questionnaire............................................................. 150 Appendix G: Influence Scale ......................................................................................... 160 Appendix H: Face Scale................................................................................................. 162 Appendix I: Experimental Conditions: Participant Instructions.................................... 166 Appendix J: Participant Information Packet .................................................................. 174 Appendix K: Manipulation Check ................................................................................. 186 Appendix L: Luck Questionnaire................................................................................... 188 Appendix M: Outcome Narrative .................................................................................. 192 Appendix N: Entrapment Questionnaire........................................................................ 203 Appendix O: Debriefing Form....................................................................................... 215 References....................................................................................................................... 216
vi
LIST OF TABLES Table 1 2 Page Study Design…………………………………………………………...40 Network Questionnaire: Summary of Descriptive Statistics for Total Number of People Nominated in the Networks for Items 1 through 8………………………………………………………………52 3 4 5 6 Influence Scale: Summary of Descriptive Statistics…………………...55 Face Scale: Summary of Descriptive Statistics……………………….66 Belief In Good Luck Scale: Summary of Descriptive Statistics……….67 Entrapment Bias Scale: Summary of Descriptive Statistics Condition I……………………………………………………………72 7 Entrapment Bias Scale: Summary of Descriptive Statistics Condition II…………………………………………………………....74 8 Entrapment Bias Scale: Summary of Descriptive Statistics Condition III…………………………………………………………..76 9 Entrapment Bias Scale: Summary of Descriptive Statistics Condition IV…………………………………………………………..77 10 11 12 Scale Reliability Coefficients,…………………………………………79 Correlation Matrix: Independent and Dependent Variables…………81 Independent Variables: Factor Score Descriptive Statistics for Condition I……………………………………………….83 13 Independent Variables: Factor Score Descriptive Statistics for Condition II……………………………………………….84
vii
14
Independent Variables: Factor Score Descriptive Statistics for Condition III……………………………………………...85
15
Independent Variables: Factor Score Descriptive Statistics for Condition IV…………………………………………….86
16 17 18
Codebook…………………………………………………………….104 Condition by Qualitative Category ………………………………….109 Percent of People Entrapped or Not Who Used Specific Qualitative Themes To Explain Their Actions…...…………………...110
19
Interaction of Reporting (Yes/No) and Constituency (Yes/No) on External Justification: Mean Summary…………………………...121
viii
LIST OF FIGURES Figure 1 2 3 Page The proposed value function from prospect theory……………..…….5 Research hypotheses…………………………………………….……36 Research findings……………………………………………….…….118
1
CHAPTER 1 Overview A woman decides to wait for a bus rather than walk. With the decision to wait for the bus and time invested in doing so, the woman waits for such a long time that she could have walked to the destination and back again by the time the bus finally arrives. This situation exemplifies the entrapment bias, when individuals continue to incur costs to achieve their objectives instead of changing their behavior. The entrapment bias is the tendency of people to assume that the more resources are expended, the closer they are to attaining their desired goal (Rubin, Kim, & Peretz, 1990). In research, this phenomenon is referred to as sunk cost (Arkes & Blumer, 1985), entrapment or entrapment bias (Brockner, 1977), concord fallacy (Arkes & Ayton, 1999), and escalation of commitment (Staw, 1976, 1981). All of these terms refer to the same phenomenon, and in the research literature they are used interchangeably. The phenomenon of entrapment has been observed and studied by scholars on different levels (i.e., macro, interpersonal and intrapersonal) and in different types of situations (e.g., waiting, bidding at an auction, gambling, decision making) to uncover variables and circumstances underlying the entrapment (Brockner, 1977; Schelling, 1960; Shubik, 1971; Staw, 1976). Responsibility levels, social motivations, information ambiguity, and observed rate of loss have been identified as some of the variables influencing conflict escalation (Rubin & Brockner, 1975; Staw, 1981; Teger, 1980). The entrapment bias has been studied predominantly within business decisionmaking and organizational contexts, including negotiation (Neale & Bazerman, 1985). Negotiation is used in a wide variety of settings: policy and law formulation, fiscal
2
budgets and salary discussions, house settlements and establishment of children’s curfews (Wall & Blum, 1991; Ways, 1979). Although many negotiators are able to achieve their desired outcomes, others fail to do so even when the parties involved have a zone of agreement; that is, when the maximum price the buyer is willing to offer is larger than the minimum price acceptable to the seller (Raiffa, 1982). This logic is applicable to a wide variety of negotiation contexts—not just buyer-seller interactions (Neale & Bazerman, 1985). Bazerman (1986) suggests that the entrapment bias could be one of the reasons why negotiators fail to achieve settlements. This dissertation examines the link between the entrapment bias and the concept of face—the desire to create and sustain positive identity in front of others (Goffman, 1955)—and internal and external justification processes. Because negotiation often involves the presence of other people (i.e., constituency) who influence negotiators’ performance (Gelfand & Realo, 1999), this dissertation explores how face-saving concerns and justification needs moderate the entrapment bias in the presence of constituencies. In addition, this research examines whether the size and influence of personal networks is associated with face-saving behaviors that, in turn, affect entrapment. In this chapter the theoretical rationale for the study is provided and research hypotheses are outlined. Chapter 2 provides an overview of the research methods employed in the study. Chapter 3 presents the study results. The summary of the study, chapter 4 discusses the results and their implications, and identifies limitations of the research. In addition, directions for future research and its significance are provided in chapter 4.
3
Theoretical Rationale Entrapment Bias Theoretical Frameworks Brockner (1992) defines escalation as “the tendency for decision makers to persist with failing courses of action” (p. 39). Fox and Hoffman (2002) outline five common characteristics that all escalation situations share. First, an individual is engaged in a goal-directed activity (e.g., problem-solving). Second, some type of resources (e.g., money, time, effort, or emotion) has been expended to achieve the goal. Third, expenditures have not brought the desired results. Fourth, a decision has to be made whether to continue or quit investing in the same course of action. And fifth, future prospects seem unlikely for making gains or even covering losses by continuing in the same path, yet the person continues the original course of action. A number of explanations have been provided for the escalation of commitment, such as self-justification, prospect theory, decision dilemma and persistence. Staw (1976, 1981) used Festinger’s (1957) and Aronson’s (1968) theories of cognitive dissonance to suggest that individuals become entrapped because they feel the need to provide justification for their actions. The sources of justification could be either internal or external. With internal justification, the decision maker justifies to himself or herself that the decision to pursue the course of action was rational; withdrawal would indicate that the decision was inappropriate. With external self-justification, individuals want to appear rational or do not want to expose their mistakes to others, such as bosses or stakeholders.
4
Empirical evidence supports the self-justification explanation of entrapment. For example, studies conducted by Staw (1976), Staw and Ross (1987), Brockner and Rubin (1985), and Arkes and Blumer (1985) suggest that individuals tend to make investment decisions and stick to the chosen course of action so as not to appear wasteful and to appear consistent in their decisions. For example, Arkes and Blumer (1985) conducted a study in which they manipulated the cost of theater tickets for season subscribers. Some subscribers received discounts and others did not, and the tickets they received were discretely marked accordingly. The researchers counted ticket stubs after performances and the results showed that those individuals who paid full price were more likely to attend all of the plays or at least more plays than those individuals who purchased the discounted tickets. Another theory used to explain the entrapment phenomenon is Kahneman and Tversky’s (1979) prospect theory, applied to escalation by Whyte (1986). The theory suggests that a biasing effect occurs when individuals frame situations as losses or gains relative to some neutral reference point. The value function (or utility), which is in a shape of “S,” is convex and steep in the domain of losses and concave in the domain of gains (see Figure 1). In other words, individuals are expected to be risk averse when they consider the situation (prospect) from the point of view of maximizing gains and risktaking when they view the situation from the frame of minimizing losses. According to Whyte (1986), when using a loss frame, individuals feel compelled to recover the cost that they have lost even at the risk of losing more, thus becoming entrapped.
5
Value +
Losses
Gains
_
Figure 1. The proposed value function from prospect theory (Kahneman & Tversky, 1979). Bazerman (1984) added responsibility to the prospect theory explanations of escalation. He indicated that individuals responsible for making an original decision that led to failure have a different frame of mind than those who are not responsible for the decision. The nonresponsible decision makers are at a neutral reference point of the curve, because they have not experienced any sunk costs. In contrast, the responsible decision makers tend to be on the losing end of the curve because they have expended resources with no return and must decide whether to withdraw or continue by risking further. The failure to reinvest is perceived as certain loss, therefore, the decision makers feel pressured to re-invest to try to avoid such certain loss. Davis and Bobko (1986) conducted a study in which 50% of the participants were personally responsible for the initial funding decision and 50% were not. The researchers then manipulated the decision-making frame by providing the participants with information about a program that either was framed negatively as failing (i.e., “after 2 years of operation the program has failed to place 60.1% of all participants in either part-
6
time or full-time jobs”), or framed positively as a success (i.e., “after 2 years of operation the program has placed 39.9% of all participants in either part-time or full-time jobs”). Participants were then asked to make a decision to continue or terminate the program. The results of the study indicated that participants in the negative frame with personal responsibility for initial allocation conditions were more prone to entrapment than participants in the other three conditions. In other words, those participants both in the negative frame and personally responsible made the decision to continue with the program. Bowen (1987) proposed that the entrapment phenomenon could be explained by the framework called difficult business decision dilemma. He suggested that because many studies do not provide clear negative feedback about initial allocation of resources (e.g., the project is a failure), the escalation demonstrated by research participants could be explained by a variety of motives such as their curiosity, desire to bring a project to fruition, or desire to learn about the problem. However, Brockner (1992) argued that some research that used self-report measures has shown that the research participants did think the feedback they received about the project was indeed negative (Brockner & Rubin, 1985). Fox and Hoffman (2002) used motivation theories developed by Lewin (1935) and Atkinson and Raynor (1974) to offer another explanation of the escalation: that individuals desire to be persistent. In other words, people are goal driven and are motivated to accomplish their goals. Lewin’s theory involves psychological regions of tension, goal valence, or desirability, and psychological distance of the path to goal obtainment (how long or difficult is it to achieve the goal). The interaction between these
7
elements is used to explain the strength of motivational force driving the individual to achieve a goal. Achievement motivation theory, suggested by Atkinson and Raynor, involves goal desirability (i.e., expectancy to achieve the goal) and movement by inertia (i.e., individuals continue to pursue goals just because this is what they have been doing). An individual’s persistence to attain a goal will be determined by the combination of inertial tendencies resulting from past experiences and the strength of current goal desirability. Some empirical evidence provides support for the persistence or goal-driven explanation. For example, Garland and Conlon (1998) manipulated investment decision and project completion by telling some participants that a project designed to revamp a company’s manufacturing capabilities was 20% complete and others that a project was 80% complete. Participants in the 80% completion cell were more likely to allocate more resources to the project, leading the researchers to conclude that the participants did so just to get the project over with. Those individuals in the 20% completion preferred withdrawal. Similar studies have been carried out by Garland (1990) and Arkes and Blum (1985). However, no self-reported data studies explain that persistence is in fact behind the cognitive processes that lead to escalation. Furthermore, if persistence by itself is an explanation for escalation, it is not clear why individuals who are not responsible for the original allocation decision, but given the task of making a follow-up allocation decision (i.e., are given a set goal), refuse to allocate further resources to a given project and thus do not become entrapped. Brockner (1992) argued that none of the theories presented (e.g., self-justification, decision dilemma, or prospect theory) could explain the escalation phenomena in its
8
entirety. All of them provide explanations of entrapment in some circumstances. However, because the self-justification explanation has received most support in past research, the current study will focus on the self-justification explanation of escalation phenomenon as it relates to the desire to save face and the need to justify one’s actions. Self-justification Self-justification has been systematically explored in compliance-gaining research. According to Kelman (1961), compliance can occur when a person accepts influence from another individual or from a group in hopes to achieve a favorable reaction from others. The individual is willing to be influenced because he or she wants to receive certain awards or avoid punishment that the other individual or group controls. Compliance gaining has been extensively studied using forced compliance experiments in which participants are induced to advocate a counter-attitudinal position to motivate them to advocate views that violate their own attitudes (Eagley & Chaiken, 1993). For example, Festinger and Carlsmith (1959) conducted an experiment in which participants were asked to complete a boring task. Afterward, half of the participants were offered $1 and the other half were offered $20 to mislead a confederate that the task was enjoyable and interesting. Those participants offered higher incentives did not change their attitudes, whereas, the participants paid just $1 came to believe that the boring task they completed was indeed enjoyable. In other words, participants offered low incentives had to justify spending time completing a boring task by perceptually biasing the task as enjoyable. Other experiments on self-justification have included writing attitude-discrepant essays (Scheier & Carver, 1980) and eating a disliked food (Zimbardo, Weisenberg,
9
Firestone, & Levy, 1965). These experiments have shown that individuals faced with personal negative consequences—such as engaging in a task they do not want to perform—as a result of counter-attitudinal acts, with no external rewards to compensate for the dissatisfying nature of the experimental task, tend to change their opinions on an attitudinal issue to cognitively reduce any negative outcomes resulting from their behavior. In short, people feel compelled to justify their previous behavior and defend themselves from negative consequences through the perceptual biasing of behavioral outcomes. In addition to negative consequences, two other factors affect the biasing of outcomes within forced-compliance situations. The first factor is the commitment to behavioral consequences, which are irrevocable or not easily changed (Brehm & Cohen, 1962). The second factor is a sense of having at least a moderate degree of choice in one’s behavior. In other words, an individual should feel personally responsible for the negative consequences of his or her behavior (Cooper, 1971). A number of studies have investigated factors influencing the nonrational escalation of commitment to a previous action. For example, Staw (1976) examined how responsibility for negative consequences affects nonrational escalation of commitment. The author examines the process of escalating commitment through conducting the simulation of making a business investment decision. The participants were divided into two groups, one with high responsibility and one with low. Those participants in the group with high responsibility were asked to allocate funds to one of two corporate divisions of an organization. The participants were then told that, after 5 years of the initial allocation of funds, their investment turned out to be either successful or
10
unsuccessful. Subsequently, the participants were asked to make a second allocation to either one of the divisions. The low-responsibility group did not have to make a prior decision as to which corporate division was most deserving of the funds. This group was presented with the entire financial decision case, including successful or unsuccessful results of the first allocation. The low-responsibility participants were then asked to make the second allocation decision. The study results indicate that participants in the highresponsibility condition, who were told their first allocation was unsuccessful, committed a significantly higher amount to the original division in the second allocation than the participants in the low-responsibility group. Thus, Staw concluded that personal responsibility for negative consequences leads to increased investment of resources in a previously chosen course of action. In another study, Rubin and Brockner (1975) investigated the passage of time as a factor influencing the entrapment bias. They demonstrated how closeness to achieving a goal affects escalation of commitment to the futile investment of resources. The scholars conducted an experiment in which the passage of time could be viewed as an investment or as an expense. The experiment participants were given an initial sum of money (the “initial stake”) and an opportunity to win a bigger some of money (the “jackpot”). To win the jackpot, the participants had to solve a series of crossword puzzles. Some of the crossword puzzles were so difficult that they required the use of a dictionary. To obtain the dictionary (i.e., scarce resource), participants had to wait in line until the resource became available, which it never did. As the time passed, the amount of money in both the jackpot and initial stake decreased. Thus, the longer the time spent waiting for the dictionary, the greater the expense to the participants, and the greater the investment.
11
The authors manipulated three variables: rate of decrease of the jackpot, availability of a chart providing information about the decrease (i.e., length of waiting time), and information concerning participants’ illusory position in line for the dictionary. The study results showed that entrapment in the waiting condition was high overall. Entrapment was particularly high when the jackpot decreased slowly, when the decrease chart was not available, and when participants thought that they were first in line for the dictionary instead of third. This study illustrates that goal completion could be used as a reason to self-justify a commitment to failing action. Overall, the discussion of self-justification research illustrates that there are five conditions that could lead to an entrapment bias: negative consequences (i.e., failure of the original decision), commitment to behavior, perceived degree of freedom in commitment to this behavior, responsibility for the negative consequences, and desire for goal completion. Gaps in Self-Justification and Entrapment Research Although a number of studies have provided support for self-justification theory (see Brockner [1992] for review), only a few studies take into consideration social factors and the way these social factors could affect escalation. Brockner and Rubin (1985) identify four types of social variables that can influence escalation: group influence, behavioral modeling, presence of an audience, and competition against a social (labor vs. management) or nonsocial (waiting for a bus) entity in an attempt to achieve the goal. One factor of particular interest to the current research is the presence of an external audience, but only a few studies (Brockner, Rubin, Fine, et al., 1982; Brockner, Rubin, & Lang, 1981; Fox & Staw, 1979; Staw & Hoang, 1995) have examined the presence of an
12
audience and its effect on entrapment. Further, no studies have explored the role of cultural variables (e.g., distinguishing between dimensions of face, personal networks) in entrapment situations. Another gap in entrapment research and the self-justification explanation of the escalation phenomena, is a lack of studies that examine cognitive processes that may explain escalation. Three types of research have been conducted to provide support for the self-justification explanation of the escalation phenomena. The most popular type includes studies that “operationalize feedback from prior resource allocations and decision makers’ needs to justify those prior resource allocations; the typical finding is that escalation is greatest when both feedback is negative and justification needs are high” (Brockner, 1992, p. 49). But these studies do not examine cognitive processes— that is, what led the participants to make the decision. These studies measure commitment to a previous decision, which is often operationalized as the amount of money allocated to the previously chosen course of action. In addition to the study described earlier by Staw (1976), studies by Davis and Bobko (1986), Fox and Staw (1979) and Schoorman, Mayer, Douglas, and Hetrick (1994) also operationalized entrapment as the amount of money invested and did not examine cognitive processes leading to the decision. The second type of research providing support for the self-justification explanation of entrapment explores other manifestations of behaviors (i.e., besides escalated commitment) resulting from the self-justification motive (Brockner, 1992). For example, Conlon and Parks (1987) found that individuals, to provide self-justification, tend to look for retrospectively focused information to make a decision regarding
13
subsequent reallocation. The researchers argue that retrospective focusing results, in part, because exoneration and justification necessitate a plausible explanation of how and why a negative consequence occurred. The third type of studies uses self-report data of psychological states to establish that behavioral escalation is indeed related to individuals’ self-justification needs (Brockner & Rubin, 1985; Strube & Lott, 1984). For example, findings from the study by Brockner and Rubin suggest that individuals who manifest escalation behavior also produce self-reports that indicate a need for justification of prior allocation (e. g., “I had already invested so much, it seemed foolish not to continue,” or “Once I had invested a certain amount, I had to keep going; otherwise all of that previous investment would have been a waste” [p. 148]). Ross and Staw (1986) cite the results of an unpublished study by Bazerman, Schoorman, and Goodman (1980) who investigated the relationship between participants’ behavioral manifestation of escalation and self-reports of other measures related to self-justification. The findings indicated that escalation was “associated with the perceived importance of a decision, the extent of disappointment with initial losses, and the perceived interconnectedness of current and past decisions in the situation” (Ross & Staw, 1986, p. 276). These studies focus on psychological states, not cognitive processes associated with escalation. None of the escalation studies examined focused specifically on cognitive processes. In addition, no studies have examined the escalation of commitment phenomenon from a communication point of view, namely focusing on messages used in situations in which individuals persist with continued commitment to the failing course of action. Drummond (1994) analyzed a case study of hiring an incompetent manager. The author
14
used case study research by participant observation that employed “direct observation, questioning, diary keeping and unobtrusive methods” (Drummond, 1994, p. 46). The scholar used quotes to document the case study, however, no systematic analysis of the messages was employed. Overall, the discussion above illustrates four gaps and weaknesses in the study of the entrapment phenomenon that the current research will address. First, there are only a few studies that examine the presence of external audience and its potential to affect entrapment. Second, there are no studies that have examined cultural variables and their potential influence on the commitment to the failing course of action. Third, there are no studies that have analyzed the cognitive processes involved in an escalation of commitment. Fourth, there is no research that analyzes messages communicated in an escalation situation. This research will address these four gaps: audience effect, cultural variables (i.e., types of face concern and personal networks), cognitive processes (i.e., internal and external justification needs), and messages used to justify one’s decision to escalate. Entrapment and Negotiation Entrapment is a cognitive bias that can have a have direct effect on negotiation. According to Putnam and Jones (1982), negotiation is a process in which two or more parties that hold or believe they hold incompatible goals try to reach a mutually acceptable solution by engaging in a give and take interaction. Neale and Bazerman (1985) argue that adopting a view of negotiation as a business decision-making process could help explain failure to reach agreements. The authors propose that cognitive or judgment biases result in reduced negotiator effectiveness in reaching best outcomes and
15
reduced likelihood of attaining agreement. The authors refer to cognitive biases (including entrapment bias) as heuristics that are helpful as shortcuts in an effort to reduce the amount of information to be processed. At the same time, these heuristics can also bias outcomes in systematic ways. In the case of entrapment bias, negotiators tend to escalate their commitment to justify their earlier bids, offers, and proposals, and to avoid the financial and ego losses of coming in second. Neale and Bazerman (1985) provide a specific discussion outlining ways in which escalation and negotiation are related. The scholars suggest that escalation is likely to lead negotiators “to stand firm on their initial offers through the course of successive negotiations” (p. 48). One reason for development of such rigid negotiation positions is the presence of an external audience or constituency. The researchers maintain: The pressure from the constituency may lead to the escalation of commitment [entrapment bias], which impedes the ability of the negotiator to represent their best interests. Further, as both sides accrue losses (e.g., during a strike), both sides are likely to increase their propensity to “hold out” in order to justify their initial positions. (p. 48) Bazerman (1986) offers three additional reasons for the entrapment bias in negotiation. The first reason is that negotiators tend to look for information that is salient with their initial commitment to a position. The second reason is that the negotiators’ judgment is biased to interpret what they observe at the negotiation table in a way that justifies or supports their initial position. The third reason is that the competitive context of a negotiation situation fuels the likelihood of entrapment. The negotiators perceive the notion of unilateral surrendering of a previously stated position or even making smaller
16
demands as defeat; on the other hand, escalation of commitment or entrapment “leaves the future uncertain” (p. 57). Consistent with prospect theory, Bazerman argues that this uncertain future is perceived as more desirable by negotiators than a prospect of certain loss. Bazerman (1986) recommends that, to avoid an entrapment bias in a negotiation context, negotiators have to realize their tendency to justify past actions. They have to remember to evaluate the benefits and costs of persisting with a chosen course of action. He also suggests that awareness of the tendency to escalate can help predict the opponents’ actions and thought processes. When individuals are too psychologically invested in their positions, they tend to increase their demands or hold out on making the concessions for too long. As a result, “a negotiator should avoid pushing opponents into a corner, getting them angry or otherwise making them feel that they can’t afford to give up the struggle” (p. 57). Other approaches to reducing cognitive biases in a negotiation include providing unequivocal feedback regarding one’s performance, improving negotiator selection criteria (i.e., identifying an individual’s tendency to be vulnerable to decisional biases), and developing training protocols to help negotiators eliminate decisional biases. Because the entrapment bias in a negotiation can lead to failure to achieve desired outcomes, this issue is important to understand and empirically explore by examining how entrapment can occur during negotiation. The next sections examine variables (i.e., accountability and face) that may affect entrapment in a negotiation situation.
17
Accountability and Constituency Normative social influence describes the force that compels people to conform to the positive expectations of another. Normative influence implies “real or imagined group pressure, and a person may conform to avoid sanctions, to gain approval or simply to avoid the embarrassment of being different” (Nail, 1986, p. 202). The desire for social approval is related to social normative influence that is associated with compliance (Asch, 1951) and consequently with the self-justification process. One condition that demonstrates normative influence is accountability, or “the extent to which representatives are required to justify their actions, and are going to be evaluated and rewarded by their constituents” (Gelfand & Realo, 1999, p. 721). A constituency can exert normative influence, which can lead to entrapment bias. Accountability and Constituencies in Negotiation Context According to Gelfand and Realo (1999), within a context of negotiation, accountability is activated when negotiators have to provide justification for their actions after the negotiation. Lewicki, Saunders, and Minton (1985) maintain that constituencies control individuals by administering rewards and punishments. Wall and Blum (1991) argue that constituencies tend to exert as strong an influence on negotiators as opponents do. These researchers explain, “The reasons are quite clear; constituents are powerful and will use their power, bringing pressures to bear that keep the negotiator committed to the constituents’—group’s or organization’s position” (p. 282). Lewicki, Saunders, and Minton (1985) posit that accountability occurs under two conditions: (a) when the constituency can observe and judge the individual’s performance, and (b) when the
18
individual is aware that the constituency depends on him or her to achieve positive outcomes. Research suggests that accountability to a constituency significantly influences negotiators’ behaviors; negotiators accountable to a constituency tend to behave more competitively and bargain tougher than those in low- or no-accountability conditions (Bartunek, Benton, & Keys, 1975; Carnevale, Pruitt, & Britton, 1979; Gruder, 1971; Gruder & Rosen, 1971; Organ, 1971). For example, findings from Neale (1984) indicate that when constituents evaluate negotiators’ bargaining and determine payoffs, the negotiators are more likely to reach impasses and less likely to concede. Lewicki, Saunders, and Minton (1985) maintain that the individual’s desire for consistency is often amplified by a desire to save face and maintain an illusion that he or she is in control in front of an audience. The authors suggest that such behavior is a result of unwillingness on the part of the individuals to admit an error or failure, particularly when the other party might interpret doing so as a weakness. They state, “The mere presence of an audience, particularly one that can easily observe the negotiator, motivates a negotiator to seek a favorable evaluation from that audience and to avoid an unfavorable evaluation” (p. 295). Neale and Bazerman (1985) and Rubin, Kim, and Peretz (1990) argue that the pressure from the audience may lead to the escalation of conflict (entrapment) and impede the negotiator’s ability to represent the best interests of this audience. Furthermore, they state that the pressure generated by constituencies toward adherence to a certain position may not be in the best interest of either the constituency or the
19
negotiator and may impede the negotiators’ ability to represent the constituency’s best interests. Accountability, Constituency, and Entrapment Bias A number of studies suggest that accountability contributes significantly to entrapment (Fox & Staw, 1979; Staw & Hoang, 1995). Staw (1981) attributes this effect to external justification. For example, Staw and Hoang (1995) studied National Basketball Association (NBA) draft choices and found that coaches gave greater playing time to players who were selected earlier in the draft and who were paid more money independent of the individual’s performance. In this case, owners, fans, and media could be considered as the external audiences to whom coaches are responsible. In another study, Fox and Staw (1979) conducted a simulation in which business students played the role of business executives who had to make funding allocation decisions under high or low conditions of job security. The scholars manipulated the popularity of a recently implemented policy at the time that the initial allocation decision was made. Fox and Staw hypothesized that those participants whose job was threatened or who implemented an unpopular policy would be motivated to protect themselves against failure. Judgment by the board of directors regarding allocation simulated the accountability condition. The participants in the job insecurity condition were informed that they were temporarily assigned the role of a vice president; those participants in the job security condition were told that their job was permanent. Half of the respondents were told that the board of directors was dissatisfied with their initial funding decision (resistance condition) and the remaining half were informed that the board was pleased with the initial resource allocation (no resistance condition). The participants were asked
20
to reallocate their funds. The findings indicated that when a course of action led to negative results (resistance condition), the participants in the low job security condition were more likely to escalate their commitment to the losing course of action by making greater subsequent resource allocations. In other words, negative input from the constituency (the board of directors) in combination with being dependent on this constituency for job security resulted in greater resource allocation. These studies did not examine the possibility that one of the underlying reasons for entrapment in the presence of constituency is an individual’s need to save face. Based on the discussion presented in the sections above, this research puts forth the following hypotheses: Hypothesis 1 (H1): Entrapment will be more likely when a constituency is present. Hypothesis 2a (H2a): Justification would be greater when respondents are required to report their behavior. Hypothesis 2b (H2b): When individuals have to report their behavior, they will have greater need for internal and external self-justifications leading to entrapment. Hypothesis 3 (H3): Presence of constituency is likely to lead to greater internal and external justification. Face Goffman (1955, 1959) introduced the concept of face or the desire to create and sustain positive identities in the eyes of significant others. This desire motivates individuals to appear strong and capable, and to avoid situations in which they could be publicly embarrassed in front of an audience. Goffman argued that people will try to prevent loss of face even if they have to incur costs. Facework is communication aimed at
21
enhancing or repairing face that has been damaged. Brown (1968, 1970) adopted Goffman’s definition of face and facework and conceptualizes facework as face-saving and face-restoration. Face-saving is defined as an attempt to prevent another from causing an individual to appear foolish or incapable to significant others and facerestoration is conceived as an individual’s attempt to seek revenge from another after the other has already damaged face. Brown and Levinson (1987) discuss two types of face: negative face and positive face. Negative face is conceived as an individual’s desire to maintain his or her autonomy (i.e., the need not to be imposed upon), whereas positive face is conceptualized as the need to seek inclusion or approval from significant others. In addition, Brown and Levinson (1987) charge that, because of the interdependent nature of social relations, individuals can threaten or support other people’s face and protect their own face. TingToomey (1988) adds another dimension to the concept of face: self and other. Self-face implies concern for one’s own image and self-interest, whereas other-face implies concern for another’s image and other-interest. Ting-Toomey (1988) proposes four types of face maintenance. First, self-negative face is associated with one’s need to protect one’s autonomy from other’s infringements. Second, other-negative face implies the need to demonstrate respect for other person’s need for autonomy. Third, self-positive face is concerned with the need to defend one’s need for inclusion. Fourth, other-positive face is defined as the need to support the other person’s need for association. Face-Saving and Negotiation The concept of face has been addressed by scholars examining conflict in the negotiation context. Wilson and Putnam (1990) state that face goals exert strong
22
influence on the negotiation process. These scholars state that, during negotiation, negotiators have a certain image of themselves that they would like to preserve. Scholars (e.g., Thompson, Nadler, & Kim, 1999; White, Tynan, Galinsky, & Thompson, 2004; Wilson, 1992) have examined the role of face threats in negotiation. According to White et al. (2004), positive face threats include situations in which one of the parties is embarrassed, disrespected or criticized. Negative face threats, on the other hand, are associated with situations in which one party desires to avoid imposition. White et al. (2004) argue that both positive and negative aspects of face could be threatened during a negotiation. However, they charge that research conducted by Cupach and Carson (2002) and Cupach and Messman (1999) suggests that positive face is more important for relationships, and therefore is more likely to be associated with integrative outcomes. Some direct face threats include nonnegotiable offers (Tjosvold, 1977), criticizing an opponent’s position (Brown & Levinson, 1987), pressure (Thompson, Nadler & Kim, 1999), and resisting making a concession (Tjosvold & Huston, 1978). In addition to direct threats to face by an opponent, certain situational factors increase face threat in a negotiation (White et al., 2004). One of these factors is constituency. The presence of constituency is likely to heighten face concerns. The discussion in the previous section illustrates that the presence of an audience can exert significant influence on an individual in a variety of contexts. Brown and Garland (1971) suggest two reasons why presence of an audience increases face-saving behavior. First, audiences can provide evaluative feedback directly to an individual. Second, an audience could also communicate its evaluation to others. Brown (1968) maintains that negotiators seek to communicate a positive image of themselves not only to the counterparts
23
involved in a bargaining process but also any other audience interested in the outcome. According to Brown (1968), “The latter may include the group the bargainer represents or in which he holds membership; they may be ‘real’ or imagined, physically present at or absent from the setting in which the bargaining occurs” (p. 109). Similarly, Wilson and Putnam (1990) argue that face-maintenance goals become more salient when negotiators receive feedback from and feel highly accountable to constituents. Stevens (1963) links face saving to concession making. He states that bargainers face a dilemma when they have to make concessions to reach an agreement. He maintains that the act of making the concession itself can be perceived by others as a sign of weakness, which can cause face loss and possibly increase attempts of exploitation by the other party. Therefore, the facesaving and economic motives may require mutually incompatible responses by individuals. The study conducted by Brown (1968) showed that bargainers are willing to sacrifice their own economic gain just to inflict worse losses on their counterparts, especially when they have been made to appear foolish in front of a salient audience. Brown conducted an experiment in which the face saving was induced by informing the participants that they would be observed by an audience while participating in a bargaining task. The task was based on a two-person trucking game in which each player runs a trucking company and has to move his or her truck over a road system to a final destination. The faster one reaches the final destination, the more money the player will earn. The game is set up in such a way that for each player to win the most money, they have to cooperate. In addition, each player has control over a tollgate through which the
24
counterpart’s truck has to pass. Players are required to either charge the opponent a specified toll or grant him or her free passage. During the first ten trials, a confederate controlled the tollgate and systematically charged the participant high tolls resulting in significant monetary losses for the participant. Then, the participants received feedback from the audience members, who were supposedly, but not actually, observing the interaction from behind a one-way mirror. The participants received feedback that either said that they looked foolish and weak (e.g., “Bolt was out to beat Acme and he really made Acme look like a sucker”) or that they looked good (e.g., “Bolt made Acme pay a lot of high tolls but Acme looked good because he tried hard and played fair). The control group received no feedback. In the second round, the participants had control over the tollgate. The participant had two choices: He or she could either retaliate against the confederate or increase his or her own winnings. The results indicated that those participants who received negative feedback from the audience were much more likely to retaliate against their counterparts than participants who were provided with positive feedback. Further, participants in the negative feedback condition were willing to lose money to restore face. Those in the positive feedback condition proceeded to maximize their profits. The results from the post-experimental questionnaire showed that the negative feedback participants were more concerned with looking strong than were the other participants. Face-Saving and Entrapment Bias Only two studies have explored the relationship between face-saving behavior and the entrapment bias. Brockner, Rubin, and Lang (1981) conducted two experiments to explore how the presence of an audience affects entrapment. In the first experiment, the
25
participants were told that they are about to participate in a study on decision making. The participants were asked to make an investment decision. Half of the participants were told about the virtues of investing conservatively (cautious condition) and the other half was instructed about advantages of investing a larger amount (risky condition). To investigate the role of face saving, experimenters assigned half of the participants to the large-audience condition (experimenter plus two confederates who were supposedly working for psychology professors, and who were interested in observing the procedures because they were considering using them in their own experiments) and the second half to the small audience condition (just the experimenter). The researchers also hypothesized that, because social anxiety is related to self-presentation, the participants with low social anxiety would be less influenced by the experimenter’s instructions than would those with high social anxiety. The results showed that investments were less than half in the cautious condition than in the risky condition. The face-saving analysis showed that (1) the instructions had a greater effect on subjects with high rather than low social anxiety, and (2) individuals with high social anxiety who participated in front of a large audience were more influenced by the instructions than were individuals with low social anxiety who participated in front of a small audience. (p. 68) In the second experiment, the procedures were the same as the ones in the first experiment, except Brockner, Rubin, and Lang (1981) orthogonally varied the importance of costs and rewards. In the high-cost-importance condition, the participants were given a chart with information regarding their costs at various points in the
26
experiment, and they were asked to plot their progress toward the jackpot. In the lowcost-importance condition, the participants were not given a chart, and they did not have to track their progress. The results showed that the reward importance variable did not have any effect. The effect of perceived costs was mediated by participants’ concern about how their investment would make them look in front of others. Participants in the high-cost-importance condition quit earlier and stated that they became less entrapped to make a desirable self-presentation (e.g., “I thought that it would look good to quit”). The authors concluded, “Individuals will become more or less entrapped to the extent that doing so will portray them in a more favorable light” (p. 78). The second study exploring face and entrapment was conducted by Brockner, Rubin, Fine, et al. (1982). These researchers varied decision makers’ face-saving concerns and the point in time at which these concerns arose. The scholars conducted two experiments. The first experiment dealt with perceived importance of costs and rewards associated with continued investment and did not involve face-saving concerns. In the second experiment, the face-saving manipulation was present and was operationalized through the presence of an evaluative audience. The participants were told that the purpose of the experiment was to simulate gasline-waiting procedures. The participants were asked to make a decision about whether to stop or continue waiting in a gas line. Half of the participants were told that experts in decision making would be observing and evaluating their behavior (evaluative condition) and the other half were told that they would be watched by a non-evaluative audience (non-evaluative condition). In the evaluative condition, the participants were provided with a form that the observers would supposedly use to evaluate them. The form included
27
such items as how much participants “appeared to have the situation under control,” “were being made to look foolish,” and “were using their money wisely” (p. 258). Brockner, Rubin, Fine, et al. also manipulated the timing of observations so that half of the participants were observed by the audience from the start of the experiment (early condition) and the other half were observed after a significant portion of resources was already invested (late condition). The results showed that the presence of an evaluative audience had no effect on participants’ resource allocation early in the process, but it did have a significant effect on investment behavior later in the process. More specifically: When the audience was described as “experts in decision making,” subjects high in public self-consciousness (or social anxiety) became less entrapped than those low on these dimensions. When the audience consisted of individuals who “wished to simply observe the experimental procedure,” however, high public self-consciousness (or social anxiety) individuals were significantly more entrapped than lows. Moreover, these interaction effects occurred when the audience was introduced late, but not early, into the entrapment situation. (pp. 247–248) Given the small number of studies related to, but not directly testing, face-saving behavior in the entrapment situation, further investigation is needed of how the desire to save face, particularly in the presence of constituency, can lead to entrapment. The current research examines the effect of face on entrapment. Because scholars have indicated (White et al., 2004) that concerns for positive face are more important to negotiators than the desire to avoid an imposition (concern for negative face), this study
28
will primarily focus on concerns for self- and other-positive face. This research puts forth the following hypotheses: H4: Individuals with greater self- and other-positive face concerns will be more likely to become entrapped. H5: Concern for saving face will be greater when the individuals have to report their behavior. H6: Overall other- and self-positive face concerns that an individual has will lead to greater need to justify one’s actions in a scenario leading to entrapment. H7: Concerns for saving face will be greater when constituency is present. Personal Networks During the last decades, network analysis has emerged as a way to examine social structures (Emirbayer & Goodwin, 1994). According to Valente (1995), networks are patterns of support, advice, friendship, and communication that are shared by members of a social system. Personal networks include an individual, people who are in contact with this individual, social relationships between the individual and other people, and social relationships between the people in contact with the individual. Emirbayer and Goodwin (1994) argue that personal network analysis is “one of the most promising currents in sociological research” (p. 1412). In fact, the authors argue that network analysis “offers a more powerful way of describing social interactions than do other structural perspectives that focus solely on the categorical attributes of individuals and collective actors” (p. 1413). According to Wellman (1983), network analysis explains social behavior as the result of individuals’ involvement in structured social relations as opposed to common attributes and norms they possess. Network
29
analysis looks at patterns of relations (Burt, 1986) as opposed to such individual attributes as political affiliation, gender, social status, or ethnicity. The study of personal networks benefits not only sociological research; it can also be applied to the field of communication. Over the past decades, individualismcollectivism and independent and interdependent self-construal have dominated crosscultural research as explanations for cultural differences in types of relationships. Both sets of constructs have been widely used to explain the relationship between individuals and their relevant others (Hofstede, 1980; Markus & Kitayama, 1991). The concepts of collectivism and inter-dependent self-construal emphasize the importance of and dependence on relevant others. On the other hand, the concepts of individualism and independent self-construal emphasize self-reliance and focus on oneself. The concept of face has been closely related to individualism-collectivism. TingToomey (1988) posits that members of individualistic cultures and people with independent self-construals are more concerned with self-face maintenance than members of collectivistic cultures or people with interdependent self-construals. On the other hand, individuals in collectivistic cultures are more concerned with mutual-face and other-face maintenance. Further, Ting-Toomey argues that members of individualistic cultures tend to use autonomy-preserving strategies, and members of collectivistic cultures tend to use approval-seeking strategies when managing conflict. However, Fiske (2002) criticized individualism-collectivism research, and Levine et al. (2003) offered criticism of the independent and interdependent self-construals pointing out that both sets of constructs have significant operationalization and measurement drawbacks. Massett (1999) suggested that personal networks provide more
30
accurate representation of the relevant others than the two sets of constructs. Massett outlined three advantages of using personal networks as a framework for examining individuals’ differences. First, personal networks allow for more accurate representation of one’s social interactions with relevant others within a society. Second, examination of the individual’s personal networks, when he or she can identify his or her own set of significant others, reduces the problems associated with arbitrary definitions of groups. Third, examination of personal networks can provide insight into quality and frequency of communication among individuals in a given society or culture. Networks not only allow for more careful description of relevant others but also allow for the measure of range and strength of the relationship with relevant others (Granovetter, 1973; Marsden & Campbell, 1984). According to Emirbayer and Goodwin (1994), network range is the number of an individual’s ties to other people and strength of ties is “the relative frequency, duration, emotional intensity, reciprocal exchange and so on which characterize a given tie or set of ties” (pp. 40-41). Granovetter (1982) provided an overview of studies that have examined tie strength. Among other applications, tie strength was a good indicator of social mobility, in which different degrees of tie strength were positively related to the outcomes of job search efforts. The most common and best indicator of tie strength is the closeness of a relationship (Marsden & Campbell, 1984). Massett (1999) conducted a study comparing the effects of culture and otherorientation on personal communication networks and behavioral intentions in the United States and Mexico. The study focused on individuals’ health networks and diabetesrelated behaviors. Massett found that Americans were significantly more independent and
31
that Mexican respondents were significantly more interdependent. In addition, the comparison of personal networks indicated that, although U.S. participants nominated more people in their networks, Mexican respondents nominated more family members, communicated with members of networks more often, reported more high-context communication with network members and indicated greater degree of closeness with network members than did the U.S. respondents. These findings suggest support for using personal networks to investigate the relationships between individuals and their significant others. More specifically, the results of the network comparison parallel closely the characteristics traditionally associated with interdependent (Mexico) and independent (U.S.) self-construals; that is, individuals with interdependent self - construal are more likely to have strong, close tie networks and individuals with independent selfconstrual are more likely to have weak, loose-tie networks. Given a plausible relationship between self-construals and networks (Massett, 1998), and taking into consideration that self-construals have an effect on face concerns (Oetzel & Ting-Toomey, 2003; Ting-Toomey, 1988), it is possible that personal networks affect face concerns. In addition, as the previous section established a connection between face concerns and entrapment, an indirect relationship may exist between personal networks and entrapment bias with face concerns as a moderating variable. Further, because a concept of a network range is at the core of network analysis (Granovetter, 1973), the size of an individual’s network along with the strength of ties within the network is expected to affect face concerns and entrapment bias. This research argues that the range and strength of one’s personal networks will predict face-saving behavior, which in turn will affect the entrapment bias. The greater
32
the size of the network and the stronger the ties in the network, the more the individual should be concerned with protecting self-positive face because of his or her desire to look good to the members of the network in order to maintain his or her inclusion in that network. In addition, individuals with stronger ties and larger personal networks are expected to be more concerned with protecting other-positive face, because they feel compelled to protect the other party’s interests. Further, individuals with larger, strong-tie personal networks are expected to be more prone to entrapment in the presence of a constituency, because members of the constituency are likely to be members of the negotiator’s professional network. Therefore, the negotiator may feel that his or her reputation, as well as reputation of the constituency, are at stake and persist with a failing course of action. Along the same lines of reasoning, these individuals are also more likely to feel more compelled to justify their behavior than individuals with smaller, loose tie networks. Based on the discussion presented above, this research will test the following hypotheses: H8: Individuals with more expansive and strong ties in their personal networks will be more concerned with their self-positive face. H9a: When participants have expansive personal networks, entrapment will be positively associated with protecting the constituency’s positive face. H9b: In the presence of constituency, individuals with larger, strong-tie personal networks will be more prone to entrapment than the individuals with smaller, weak-tie networks.
33
H10: In the presence of constituency, individuals with larger personal networks and strong ties would report greater justification for making their decisions. Luck Although unrelated to the theoretical rationale of this research, the concept of luck is discussed here, because luck was used in the method to foil the nature of the study; that is, the participants were told that the study is about luck. It was necessary not to disclose the nature of the research so as not to bias the outcomes. To identify measures of luck suitable to be employed in this research, a literature review of studies examining luck was conducted. Summarized below is a brief overview of luck research. Luck measures used in this dissertation are discussed in chapter 2. Traditionally, luck is discussed within the framework of conditions related to expectations of control and success (Darke & Freedman, 1997a). According to Darke and Freedman (1997a), “luck is a random, uncontrollable factor which should have little effect on future expectations” (p. 487). Luck is typically considered to be an external, unstable factor that can explain achievement outcomes or social events (Darke & Freedman, 1997b). Social learning theory of personality developed by Rotter (1955) posits that perceptions of control decrease if events are attributed to luck or other people (external locus of control) and increase when events seem to be a result of a person’s own actions (internal locus of control). Overall, individuals deceive themselves as having less control if they believe that luck is involved. An attributional model explaining the origin of perceived control identifies four causal factors to which failure or success is usually attributed: luck, task difficulty, effort, and ability (Weiner et al., 1972). In addition to the locus of control aspect specified by
34
the social learning theory, attributional theory introduces stability—a degree of consistency across time and situations. According to the attributional model, luck is considered to be an external and unstable factor. That is, the model predicts that any success attributable to luck should be regarded as uncontrollable; in addition, future successes cannot be predicted on the basis of luck. Both the social learning theory and the attributional model assume that individuals have rational views about causes of luck; they see luck as unstable and external. However, studies conducted by Darke and Freedman (1997a) suggest that not all people have rational views about luck, some hold irrational beliefs that luck is a stable factor that influences events in their favor. Individuals who hold the latter view of luck tend to have positive expectations for the outcome of future events. This result supports findings from another study conducted by Darke and Freedman (1997b) that demonstrates that those who believe in luck are more confident and positive about future success. Further, a study conducted by Wohl and Enzle (2002a) shows that individuals who believe they are lucky also believe that they can use their luck intentionally to influence the outcome of chance events. For example, research participants acted “as though luck could be transmitted from themselves to a wheel of fortune and thereby positively affect their perceived chance of winning” (p. 1388). Another study conducted by Wohl and Enzle (2002b) indicates that individuals’ differences in self-perceived luck influence their future behaviors. More specifically, Wohl and Enzle (2002b) conducted an experiment in which participants were assigned to either near win or near loss conditions. The experimental task consisted of playing a computerized slot-machine style wheel of fortune game. In the near-loss condition, the wheel appeared to almost stop at
35
the Bankrupt section but inched forward towards a small payoff section. In the near-win condition, the wheel appeared ready to stop at the Jackpot section, but came to a complete stop at the small payoff section. The near escape of big loss led participants to deem themselves lucky. As a result, they were more prone to gamble more in comparison to participants who experienced a near big win. The discussion above suggests that luck is a plausible disguise to be used in this research. Specifics of the experimental procedures are detailed in the chapter 2: Methods. Summary of Research Hypotheses Figure 2 provides visual representation of study hypotheses and relationships among study variables. Circles represent constructs that incorporate more than one variable. For example, personal networks are comprised of the size of the network and influence of network members have on the person or tie strength.
36
Personal Networks (size and influence)
Self-positive face
Constituency Other-positive face Entrapment Reporting Internal Justification
Overall selfand otherpositive and negative face concerns
External Justification
Figure 2. Research Hypotheses. Overall, this research attempts to provide support for the following hypotheses: Hypothesis 1 (H1): Entrapment will be more likely when a constituency is present. Hypothesis 2a (H2a): Justification would be greater when respondents are required to report their behavior. Hypothesis 2b (H2b): When individuals have to report their behavior, they will have greater need for internal and external self-justifications leading to entrapment. Hypothesis 3 (H3): Presence of constituency is likely to lead to greater internal and external justification.
37
Hypothesis 4 (H4): Individuals with greater self- and other-positive face concerns will be more likely to become entrapped. Hypothesis 5 (H5): Concern for saving face will be greater when the individuals have to report their behavior. Hypothesis 6 (H6): Overall other- and self-positive face concerns that an individual has will lead to greater need to justify one’s actions in a scenario leading to entrapment. Hypothesis 7 (H7): Concerns for saving face will be greater when constituency is present. Hypothesis 8 (H8): Individuals with more expansive and strong ties in their personal networks will be more concerned with their self-positive face. Hypothesis H9a (H9a): When participants have expansive personal networks, entrapment will be positively associated with protecting the constituency’s positive face. Hypothesis 9b (H9b): In the presence of constituency, individuals with larger, strong-tie personal networks will be more prone to entrapment than the individuals with smaller, weak-tie networks. Hypothesis 10 (H10): In the presence of constituency, individuals with larger personal networks and strong ties would report greater justification for making their decisions. Research Question Finally, because currently there are no known studies that have examined the entrapment phenomenon from a communication point of view—that is, no studies have examined messages used by individuals in situations potentially leading to entrapment—
38
this study poses the following research question: In a situation leading to entrapment, are individuals likely to use social accounts to explain their behavior? Bies (1987) defines social account as “a verbal strategy employed by a person to minimize the apparent severity of the predicament or convince the audience that the wrongful act is not a fair representation of what the actor is ‘really like’ as a person” (p. 294). Sitkin and Bies (1993) state that social accounts are used to influence an individual’s perception of responsibility for an action and motivation for an action, and to paint the unfavorability of an action. Sitkin and Bies (1993) identify three broad categories of accounts: (a) mitigating responsibility, (b) legitimizing the action by appealing to some higher-order values or norms, and (c) reframing outcomes. Mitigating, or causal, accounts suggest that the situation forced the individual to take the action. Sitkin and Bies state, “by suggesting that the offending party had no other alternatives to the chosen action, a social account claiming mitigating circumstances should reduce the amount of blame attributed to the party” (p. 350). When individuals use the second type, exonerating accounts, they try to explain their actions by placing them within a broad normative framework that will give legitimacy to their motives. The third type, the reframing account, attempts to put the actions in the best possible light by suggesting to the offended party the appropriate context for interpretation. To this end, this research will examine the messages produced by the research participants to determine whether they are likely to be social accounts and explore which of the social accounts are used most often.
39
CHAPTER 2 Method This chapter describes the process of developing research questionnaires and data collection procedures. The first section presents describes the experimental design. The second section provides a description of the study sample. The third section describes in detail the recruitment procedures. The fourth section discusses the experimental procedures and manipulations. Finally, the fifth section provides a detailed discussion of the development of study measures. Experimental Design One of the purposes of this research is to examine whether the presence of constituency leads to entrapment. The presence of constituency is expected to pose face threats that lead to entrapment. The research also hypothesizes that accountability, in this case reporting in person to the constituency, will pose greater face threat than nonaccountability. In addition, face concerns are expected to be moderated by the size and strength of personal networks. There are three conditions essential to simulate accountability (Gelfand & Realo, 1999). First, constituents need to have control over rewards. Second, individuals must justify their performance. Third, individuals must be evaluated by the constituency. For the purposes of this study, accountability is manipulated using two requirements in the accountability condition: (a) presence of a constituency who has control over rewards and evaluates performance, which meets the requirements of the first and third conditions; and (b) reporting, so that individuals have to explain their behavior to the constituency in person, meeting the requirement of the second condition. To this end, two experimental
40
conditions are employed: presence of constituency and reporting on the outcomes of the experiment. Overall, the respondents were assigned to one of the four conditions (see Table 1): 1. Working in groups and reporting in-person to the group (COND I: constituency; reporting). 2. Working in groups, with no reporting (COND II: constituency; no reporting). 3. Working individually and reporting in-person to the researcher (COND III: no constituency; reporting). 4. Working individually (COND IV: no constituency; no reporting). The research design is 2X2 (constituency [yes/no] by reporting [yes/no]), with face and justification as mediating variables, personal networks as a moderating variable, and entrapment as a dependent variable. Table 1. Study Design (N = 236) Constituency Yes Yes Reporting No n = 58 n = 62 n = 57 No n = 59
Study Sample The study sample consisted of 236 undergraduate students majoring in communication enrolled in a large East Coast university (N = 236). The overwhelming majority of the participants (97.4%) were full-time undergraduate students. The remaining 2.6% were part-time undergraduate students. Of the participating students
41
2.5% were freshmen, 33.9% were sophomores, 39.4% were juniors, and 23.7% were seniors. Ethnic composition of the sample was as follows: White, non-Hispanic or Caucasian (64.4%), African American or Black (14%), Hispanic (6.4%), Asian or AsianAmerican (6.8%). In addition, 6.8% of the respondents reported their ethnicity as “Other.” Participants were between 17 and 33 years old (M = 20.3 years, SD = 1.86, median = 20). Female participants constituted 80.9% of the sample and male participants constituted the remaining 19.1%. Such an imbalance in female to male student ratio is not unusual, as this is a growing trend in communication classes, where the majority of majors are female. A slight majority of the participants (53%) reported being employed. The most often-cited categories of employment were sales, administrative/clerical (e.g., secretary, administrative assistant, account clerk), and service industry (e.g., waiter, nanny, chef) positions. Recruiting Procedures and Research Assistants Participants volunteered to take part in the experiment. They received extra credit in their communication courses for their participation. In addition to extra credit, each participant received ten raffle tickets for four raffle drawings; each drawing was worth $50. Each ticket was worth $1. Participants were not allowed to purchase tickets outside of the study. When the study was completed, the researcher administered one lottery and awarded $50 to the winner. The study was announced by each class instructor (see Appendix B). The announcement was made several times during the course of the semester. The prospective participants were told that if they would like to receive extra credit for the class, one option available to them was to participate in an experiment investigating an individual’s
42
luck and decision-making processes. An alternative written assignment was offered to those students who wanted extra credit but did not want to participate in research. The instructor passed around the sign up sheet. Those individuals who chose to participate in the study received an informed consent form (see Appendix D) and a number of questionnaires that they were asked to fill out prior to arriving at the experimental laboratory. The questionnaires included demographic information (see Appendix E), a personal network instrument (see Appendix F), a network influence instrument (see Appendix G), and a face scale instrument (see Appendix H). Two female undergraduate students were recruited to help the researcher and act as the “confederate constituency” in conditions in which constituency was present. Several hours were spent training the undergraduate students on how to act and “exert pressure” on a participant. The confederates were encouraged to put full responsibility for deciding to continue or withdraw from the task on the research participant. Sample statements used by the confederates included the following: “It is all up to you, but it sure would be nice to win some money,” “Win us some money,” “I hope you win, I can use some extra cash” and “If you think it is time to withdraw, it is your call, but extra cash for Christmas presents would be nice.” In addition, on several occasions, the confederates also acted as research assistants, administering instruments to the participants in the noconstituency conditions. During their training the undergraduate assistants also received instructions on how to use the instruments, and they practiced administering them. In addition, the researcher sat in during the first four experiments conducted by the assistants. The researcher was available for questions when the research assistants
43
administered the study. The undergraduate students helping the researcher received three independent-study credits for their involvement in this research project. Data Collection and Experimental Procedures Data were collected between June and December 2004. Prior to starting the experimental task, the participants received an information packet that included details of the task (see Appendix J) and an informed consent form (see Appendix D). In addition, they were asked to fill out the following questionnaires: manipulation check and luck (see Appendices K and L, respectively). After participating in the experiment, the participants were asked to provide a narrative explaining the experiment outcomes and complete the entrapment bias questionnaire and the face-work scale instrument (see Appendices M, N, and O). At the beginning of each experiment, the researcher explained the procedures (see Appendix I) and distributed the information packet (see Appendix J). The participants were informed that the experiment was designed to study luck. Prior to participation in the study, the respondents were told that the purpose of the experiment was to study luck “but some features of the research will not be revealed until the research is completed. This will ensure that study results are not biased” (see Appendix I). During the pre-experiment instructions, the researcher also mentioned that results of many past studies investigating luck indicate that how well one performs in this particular experimental task is “a strong indicator of how lucky one is in other areas of life, such as finding a good job, finding a partner for life, etc.” After participants completed the pre-task instruments, the researcher mentioned, “Even though some people feel unlucky, their performance on this task does not depend on it. And, furthermore,
44
even if you feel that you are unlucky, there is still about one in ten chances that you will be successful in accomplishing the task.” This statement was designed to preclude participants from quitting too early because they consider themselves unlucky. In social research, not fully disclosing the nature of the study is not uncommon and sometimes is necessary to prohibit biases and confounding variables (Schrag, 2003). According to Schrag (2003), this type of deception is justified when validity of research is at stake, when risk to participants is minimal, and when the researcher has a debriefing plan. In this research, all three criteria were met. First, the researcher could not disclose that the investigation concerns entrapment, because then the participants would have been aware of the potential for escalation of commitment and would be unlikely to become entrapped. Second, the risk to the participants was indeed minimal, and they were not likely to object once they were told about how they were deceived. Third, at the end of the experiment the researcher debriefed the participants and explained to them the true purpose of the experiment, defined entrapment, and identified the experimental conditions and manipulated factors. The debriefing also stressed that the experiment did not “have anything to do with luck and your [participant’s] performance is not indicative of how lucky you are as a person.” The debriefing form can be found in Appendix O. Task Description The participants were told that the task would consist of drawing ping-pong balls out of a box. The experimenter explained that the box contained 100 ping-pong balls, of which 90 were white and 10 were red, and that the participant’s luck would be measured by the number of red balls he or she drew from the box. In fact, there were no red balls in the box. The participants were told that the goal of the task is to draw five out of the ten
45
red balls that are mixed in with white balls. The participants were informed that they could make up to 20 attempts to draw five red balls. It was also emphasized that the experiment was structured so that only drawing five red balls counted as a successful completion of the task; in other words, drawing one, two, three, or four red balls constituted failure to complete the task. All of the participants received extra credit for participation regardless of their performance and were reminded that their performance would not affect receiving extra credit. All of the participants started the experiment with ten raffle tickets. The participants were told that if they successfully completed the task (i.e., drew 5 red balls) they would receive an additional 20 raffle tickets, bringing the total number of raffle tickets to 30. However, because there were no red balls to draw, it was impossible for anyone to receive 30 tickets. Constituency Manipulation Half of the participants were told that they would be working on their own (no constituency). In the no constituency condition, the respondents were told that after they received instructions, they would be escorted into the experiment room where they would perform the drawing task. They received a pay-off sheet reflecting potential gains and losses (see Appendix J). More specifically, the pay-off sheet detailed that for every white ball drawn, the participant would lose one raffle ticket (of the ten tickets he or she started with) and for every red ball drawn he or she would win two raffle tickets. Each participant was informed that if he or she were to get “in the negative,” he or she would have to pay for the number of negative points accumulated, such that one point was worth one dollar. Because the participants started with ten raffle tickets each and could make up to 20 draws, the maximum penalty possible was $10. However, no money was actually
46
collected in the end. The participants were also informed that regardless of how many points he or she lost, if all 5 red balls were drawn by the end of the task, the person would receive 20 raffle tickets with no penalties. For example, if the respondent had negative points at the time of drawing all 5 balls, she or he would not have to pay and would receive all 20 raffle tickets for successfully completing the task. The other half of the participants were told that they would be working in groups of three participants (constituency condition). These groups actually consisted of two confederates (constituency) and one participant. By the time the participant arrived at the lab, the confederates were already waiting. The researcher pretended that she did not know them and took down their names and classes they were enrolled in along with the participant’s information for the purposes of providing this information to his or her instructor to notify the instructor that the student should receive extra credit for participating in the study. In the constituency condition, the participants were told that, although they were working as a group, only one person would be performing the task, and that person would be decided by number drawing. The participants in the constituency condition were told that the experimenter was interested in investigating “how an individual’s luck affects groups; particularly, because there have been studies that have shown that not only individuals could be lucky, but also groups.” To vote on the person to perform the task, the participants were asked to draw a number out of a hat; the person who got number “3” had to perform the drawing. To ensure that the participant was the one who would be performing the task, all of the folded pieces of paper had the number “3.” The confederates had an agreement that one
47
would always say that she had “1” and the other one would always say that she had “2.” Because of these arrangements the participant was always the one to perform the task. Each participant in the constituency condition was instructed in front of the confederates that, for every round in which the participant failed to draw a red ball, the constituency would lose one raffle ticket per person and the participant would also lose one raffle ticket. On the other hand, for every round in which the participant did draw a red ball, the constituency would win two tickets each, and the participant would win two tickets. If the participant got “in the negative,” losing more raffle tickets than the number held by each person, the number of points that the constituency would lose would be the same as the number of points lost by the participant. For example, if the participant drew eleven white balls—that is, one negative point—he or she would lose $1, and each member of the constituency would also lose $1. The maximum penalty for each member of the constituency and the participant was $10. Furthermore, to meet the requirement that the constituency control rewards, the participants were instructed that, at the conclusion of the task, the constituency would determine an amount (from $0 to $5) to be awarded to the participant for his or her performance (this award is in addition to the extra credit that all participants would receive regardless of their performance). The experimenter told the participants that she would pay him or her the amount indicated by the constituency. Modeling after the procedures used by Organ (1971), the participant was told that the constituency would also receive compensation, but the amount or payment structure was “left unspecified so as not to bias their decisions on how much to award.”
48
Reporting Manipulation In each accountability condition, half of the participants were asked to report on their performance in the task (reporting condition). The other half of the participants were told that, after they finished the experimental task, they would complete some questionnaires and would be free to go (no reporting condition). Thus, prior to the experiment, half of the participants were informed that they would report in person either to the constituency (one-fourth of the participants) or to the researcher (one-fourth of the participants) at the end of the task. All participants were asked to write a paragraph justifying his or her strategy and statements he or she would use to explain the success or failure in performing the task (outcome narrative). Respondents in the reporting condition were told that they would have to complete the outcome narrative before facing the constituency or the researcher. In fact, the participants did not have to report to the constituency or researcher. As soon as the respondents completed the post-experimental questionnaires they were debriefed by the researcher and were free to go. The information packet for the reporting/constituency condition read: “As a group representative you will perform the drawing task on behalf of your group. At the conclusion of the task, you will write a short essay (one or two paragraphs) summarizing the outcome of the task and explaining your strategy and the results. You will use this essay to report to the group members after you are finished with the drawing. After the explanation is presented to the group members, they will carefully weigh the information and decide how much money to award you for your performance.” The information packet for the reporting/no constituency condition read: “You are to perform the experimental task on your own. At the conclusion of the task, you will write a short essay
49
(one or two paragraphs) summarizing the outcome of the task and explaining your strategy and the results. You will use this essay to report to the researcher after you are finished with the drawing.” In the no reporting/constituency condition, each participant was told that the experimenter would inform the group members of the outcomes of the task and the group members would make a decision regarding the monetary reward to be received by the participant. In the no reporting/no constituency condition, the respondent was told that after completion of the task and the post-experimental questionnaires the participant would receive a debriefing and would be free to go. The statements produced by the participants were collected by the experimenter and analyzed for the types of messages most often used by the respondents in an attempt to explain their behavior. Dependent Variable The dependent variable in this study was the entrapment bias. For the purposes of this study, the entrapment bias was conceived of as a participant’s commitment to continuing the task. It was operationalized by the number of drawing rounds (the greater number of rounds, the greater the entrapment bias) the participant chose to conduct. Analysis of descriptive statistics of the total number of draws shows M = 9.78, SD = 5.17 and median = 9.00. Number of draws ranged from 0 (2.5% of the respondents) to 20 (12.7% of the sample). About 75% of the participants withdrew by the 10th round. Approximately 8% of the sample lost all 10 tickets, and about 31% stopped at round nine and thus, were left with just one raffle ticket. The remaining 25% continued the task
50
incurring the monetary penalty, with approximately 18% (of the total sample) incurring a penalty over $5. Final Materials and Development of Measures For anonymity purposes, each participant selected a unique four-digit number that he or she wrote on all study instruments (see Appendix C). When the respondents handed back their questionnaires, the researchers made sure that all surveys had the matching identification numbers. The respondents filled out the materials in the following order: (1) prior to arriving to the lab, the respondents signed an informed consent form and completed the demographics questionnaire, personal network questionnaire, network influence instrument, and face scale; (2) after receiving the experimental instructions, the respondents signed another informed consent form and completed pre-task questionnaires including the manipulation check and luck scale; (3) after finishing the experimental task, the participants completed a series of post-task questionnaires, which included an outcome narrative, an entrapment scale, and face scale (identical to the one they completed prior to participating in the research study); and (4) after the respondents turned in their post-task questionnaires they read a debriefing form and received explanations about the study from the researcher. Institutional Review Board (IRB) approval was received to conduct this study. The IRB approval letter can be seen in Appendix A. The discussion of measures used in the study (presented below) follows the order in which each instrument was administered. Table 3 summarizes reliability coefficients (Cronbach’s ) for the scales used in this study. Informed consent form. The informed consent form assured the respondents of the confidentiality of their performance, informed them that there were no long-term effects
51
associated with this research, and explained that they were free to terminate the session at any time without penalty (see Appendix D). In other words, if a participant chose to withdraw from the task, he or she would still receive extra credit for participation. The respondents were asked to sign the form twice. The first time, the participants received the form along with other questionnaires when they signed up for the study. They were asked to bring the signed form and completed questionnaires to the lab on the day of their participation in the study. The second time, the participants were given the form and asked to sign it after they received experimental task instructions. Demographic questionnaire. The demographics questionnaire contained questions regarding participant gender, age, employment status, student status and standing, and race (Appendix E). Personal networks questionnaire. Prior to arriving at the lab, the participants were asked to complete a personal networks questionnaire (see Appendix F). The questionnaire used items developed by McAllister and Fischer (1978). The questionnaire included a total of 11 items. Items 1 through 8 covered various aspects of an individual’s life. Examples of questions included in the instrument are “Who would care for your home if you were to go out of town?”; “With whom do you talk about school or work decisions?”; “With whom do you engage in social activities?”; and “From whom would or could you borrow a large sum of money?” Descriptive statistics for these questions are summarized in Table 2. Item 9 asked respondents to provide the number of all adult members in their household. As the respondents identified members of various networks, they were asked to indicate either the name of the network member or the relationship of the person to the participant (e.g., my father). The questionnaire allowed the respondents
52
to nominate up to 15 network members for each of the questions. However, in case there was anyone else important to them that they would have liked to nominate, item 10 offered them an opportunity to list those people. The final question asked the respondents to count the total number of people they named on all of the lists and asked them not to count a person multiple times. However, only 79% of the sample provided answers to this question, reporting 15 people as an average number of network members (M = 14.64, SD = 6.69 and median = 14). The number of members in the network ranged from 1 to 38. Because about 20% of the respondents failed to indicate the total number of people nominated in their networks, a new variable NETWORK was created in which a total number of people nominated in the network was calculated such that every person nominated, even if they were nominated more than once, were counted (Massett, 1999). The range of the network size for this new variable was 78, with minimum of 9 and maximum of 87 people and with the mean of 39.94 (M = 39.94, SD = 15.18, median = 39). In addition, for analysis purposes the NETWORK variable was dichotomized, with the network size between 1 and 39.9 considered as small and network size between 40 to100 was considered large. This variable was called NETSIZE.
53
Table 2 Network Questionnaire: Summary of Descriptive Statistics for Total Number of People Nominated in the Networks for Items 1 through 8 Question (Total number of people listed) Who would care for your home if you were to go out of town? With whom do you talk about school or work decisions? Who, if anyone, has helped with household tasks in the last three months? With whom do you engage in social activities (e.g., going to a movie, having dinner)? Whom do you talk with about your interests or hobbies? With whom do you talk about personal worries? Whose advice do you consider in making important decisions? From whom would or could you borrow a large sum of money?
Mean (M) 4.46
Standard Deviation (SD) 2.90
Median 4.00
4.75
2.41
5.00
2.85
1.84
3.00
6.70
3.63
6.00
5.39
3.12
5.00
4.36 3.87
2.43 1.98
4.00 4.00
2.60
2.00
1.36
Because closeness is an important predictor of strong ties, three items were added to the questionnaire (Marsden & Campbell, 1984). Two of these questions, “Please indicate how well do you know each of the people you mentioned?” and “How close do you feel to the person?”, were used by Massett (1999). Responses to these questions were not used in the final analysis; instead, the influence scale (see section below) was used to measure strength of the relationships within the network. The third question, used by
54
Marsden and Campbell (1984), read, “How would you characterize this person?” In other words, the respondents were asked to indicate whether each person named was a relative, an acquaintance, a good friend, or a very close friend. Overall, the respondents nominated mostly their close friends (M = 18.01, SD = 10.09, median = 17) and relatives (M = 16.75, SD = 7.85, median = 16) as members of their networks. The friends category (M = 2.76, SD = 3.58, median = 1) received just a few nominations, and almost no one nominated acquaintances (M = .71, SD = 1.57, median = 0) as part of their network. Influence scale. In addition to the personal network questionnaire, an influence scale was also administered. The scale was developed to measure how close the participants were to the members of their network by asking them how much influence people nominated in their network have on the respondent’s life. This 12-item instrument used a 7-point Likert scale in which 1 indicated “no influence” and 7 indicated “strong influence.” The respondents were asked to indicate how much influence people nominated in the networks would have on their job choice, education choice, social life, personal life decisions, social habits, outward appearance, interests and hobbies, neighborhood choice, ways to resolve conflict, political position, and decision to undergo a serious medical procedure. Reliability analysis indicated that the scale had strong reliability (Cronbach’s items. = .82). Table 3 summarizes descriptive statistics for the scale
55
Table 3 Influence Scale: Summary of Descriptive Statistics Standard Deviation (SD) 1.36
Question Overall how much influence do people you listed in Q1 through 10 have on a type of job you might choose (e.g., type of company, position)? Overall how much influence do people you listed in Q1 through 10 have on your decision to pursue education beyond college? Overall how much influence do people you listed in Q1 through 10 have on your social life (e.g., choice of friends)? Overall how much influence do people you listed in Q1 through 10 have on personal development (e.g., books you read, music you listen to)? Overall how much influence do people you listed in Q1 through 10 have on your personal life decisions (e.g., whom to date, sex conduct)? Overall how much influence do people you listed in Q1 through 10 have on your social habits (smoking, alcohol consumption)? Overall how much influence do people you listed in Q1 through 10 have on your outward appearance (e.g., clothing)
Mean (M) 4.83
Median 5.00
5.08
1.57
5.00
5.23
1.41
6.00
4.61
1.52
5.00
4.78
1.58
5.00
4.85
1.63
5.00
4.27
1.54
5.00
56
Question Overall how much influence do people you listed in Q1 through 10 have on your interests and hobbies? Overall how much influence do people you listed in Q1 through 10 have on your neighborhood choice? Overall how much influence do people you listed in Q1 through 10 have on ways you choose to resolve conflict? Overall how much influence do people you listed in Q1 through 10 have on political position you might take (e.g., support a candidate, party, voting behavior)? Overall how much influence do people you listed in Q1 through 10 have on your decision to undergo a serious medical procedure (e.g., surgery)?
Mean (M) 4.27
Standard Deviation (SD) 1.48
Median 5.00
4.26
1.75
5.00
4.49
1.58
5.00
3.69
1.84
4.00
4.93
1.78
5.00
According to Levine (2005), in communication research confirmatory factor analysis (CFA) is preferred to exploratory factor analysis (EFA) because in comparison to exploratory factor analysis, “CFA provides stronger evidence for dimensionality than EFA because EFA can under-factor correlated constructs and because model fit is typically tested with CFA” (p. 337). Therefore, throughout this research CFA was used to confirm scales employed by the study. CFAs were performed using principal components extraction and varimax rotation. To this end, confirmatory factor analysis (CFA) was used to confirm a composite measure of influence that people nominated in the networks have on the participants’ lives (INFLUENC). This initial factor analysis
57
yielded three factors. However, loadings for the two of the factors were weak. To create a factor score, the items were forced into one factor. The resulting factor score was used in analyses discussed in the chapter 3, Results. The overall combined network size (NETWORK) was then multiplied by the overall influence measure (INFLUENC) to create an overall measure of the size and strength of the nominated network (NTWRKINF). Face scale. The participants were asked to complete a face scale developed specifically for this experiment (see Appendix H). The scale was administered twice, once prior to arriving to the lab and the second time after the experiment. The scale was adapted from the instrument developed by Wilson and Kunkel (2000), who examined face threats in relationship issues using a series of scenarios with follow-up questions. For the purposes of this research, instead of scenarios, one-sentence items were created. The instrument was developed using four dimensions: self-positive face, other-positive face, self-negative face, and other-negative face. The 24-item instrument used a 7-point Likert scale in which 1 indicated “strongly disagree” and 7 indicated “strongly agree.” Among items examining self-positive face were questions such as “I worry about how other people judge me when it comes to my physical appearance” and “In general, it is important to me that people do not think that I am nosy.” The reliability analysis of items measuring this dimension yielded strong reliability (Cronbach’s = .73). In
addition for the purposes of analysis, confirmatory factor analysis was used to create a composite measure of self-positive face (SELFPOS). The initial factor analysis yielded one factor with all of the loadings above .50. The resulting factor score was used in analyses using self-positive face discussed in the Results chapter.
58
Examples of items looking at other-positive face included “Overall, when I talk to people, it is important to me that what I say does not make them look inadequate” and “I don’t like to get into arguments with people because it might make them look uncooperative.” The reliability analysis of items measuring this dimension yielded strong reliability (Cronbach’s = .74). Review of the reliability of specific scale items
suggested that one item, “When I talk to people, I want to make them feel comfortable discussing issues with me,” should be deleted. The new reliability analysis yielded stronger reliability (Cronbach’s = .75). Confirmatory factor analysis was used to create
a composite measure of other-positive face (OTHERPOS). The initial factor analysis yielded one factor with all of the loadings above .50. The resulting factor score was used in analyses discussed in the Results chapter. Items “In general, it is important to me that others feel that I’m an independent person” and “When my close friend does me a favor, I worry about the fact that I will be obliged to return the favor in the future” illustrate examples of questions measuring selfnegative face. Review of the reliability of specific scale items measuring this dimension suggested that one item, “Overall, I dislike when people give me an advice when I did not ask for it,” should be deleted, resulting in an improved reliability coefficient (Cronbach’s = .71). Confirmatory factor analysis was used to create a composite measure of selfnegative face (SELFNEG). The initial factor analysis yielded two factors. However, loadings for one of the two factors were weak. To create a factor score, another confirmatory factor analysis was performed in which the items were forced into one factor. All of the new loadings were above .50. The resulting factor score was used in analyses discussed in the Results chapter.
59
Finally, examples of items examining other-negative face included “When I ask my close friend for a favor, I worry that he/she might find it hard to say ‘no’” and “When I ask my close friend to do something for me, I worry that he/she might feel pushed into agreeing with what I want.” Original reliability analysis of items measuring this dimension revealed that item “In general, I do not feel comfortable sharing my problems with other people because I’m afraid I’m going to overburden them” should be deleted, resulting in an improved reliability (Cronbach’s = .96). Confirmatory factor analysis
was used to confirm a composite measure of other-negative face (OTHENEG). The initial factor analysis yielded one factor solution with strong loadings, all above .70. The resulting factor score was used in analyses discussed in the Results chapter. Overall, for the purposes of analysis, four new variables were created: SELFPOS, OTHERPOS, SELFNEG, and OTHERNEG. Table 4 summarizes descriptive statistics for the scale items. Table 4 Face Scale: Summary of Descriptive Statistics Standard Deviation (SD) 1.66
Question When I ask my close friend for a favor, I worry that she/he might feel obliged to comply with my request When I ask my close friend for a favor, I am concerned that he/she might feel like they have to say "yes" to my request When I ask my close friend for a favor, I worry that she/he might find it hard to say "no"
Mean (M) 3.84
Median 4.00
3.69
1.67
3.00
3.68
1.68
3.00
60
Question When I ask my close friend to do something for me, I'm concerned that she/he might feel like she/she has to go along with my request When I ask my close friend to do something for me, I worry that she/he might fee pushed into agreeing with what I want In general, I do not feel comfortable sharing my problems with other people because I'm afraid I'm going to overburden them In general, it is important to me that others feel that I'm an independent person When I ask somebody for a favor, I worry about the fact that I will feel indebted to the person In general it is important to me to be self-sufficient because I do not like owing anything to other people In general it is important to me that people in my life realize that I'm capable of making decisions myself When my close friend does me a favor, I worry about the fact that I will be obliged to return the favor in the future Overall, I dislike when people give me an advice when I did not ask for it
Mean (M) 3.54
Standard Deviation (SD) 1.61
Median 3.00
3.30
1.60
3.00
3.00
1.79
3.00
5.24
1.36
5.00
3.49
1.65
3.00
5.13
1.53
5.00
5.73
1.17
6.00
2.98
1.59
2.98
3.85
1.64
4.00
61
Question Overall, when I talk to people, it is important to me that what I say does not make them look inadequate In general, when I give advice to a person, it is important to me that she/he does not feel like I'm implying that she/he has difficulty handling the situation Overall, when I give an advice to my friend, it is important to me that my friend does not feel like I'm implying that she/he does not understand the consequences of his/her actions In general, when my friend inadvertently breaks promise, I don't want him/her to feel like I think that she/he is person who never honors his/her commitments I do not like to get into arguments with people because it might make them look uncooperative When I talk to people, I want to make them feel comfortable discussing issues with me I worry about how other people judge me when it comes to my physical appearance I want people to find me physically attractive
Mean (M) 4.71
Standard Deviation (SD) 1.41
Median 5.00
4.83
1.41
5.00
4.66
1.50
5.00
4.17
1.47
4.00
3.18
1.37
3.00
6.04
1.08
6.00
4.76
1.59
5.00
5.85
1.16
6.00
62
Question In general, when a group of colleagues is having a conversation, I like to be included in the conversation In general, it is important to me that people do not think that I am nosy In general, it is important to me to make a positive impression on people In general, it is important to me that people think that I am smart and capable
Mean (M) 5.66
Standard Deviation (SD) 1.20
Median 6.00
4.86
1.50
5.00
6.29
.80
6.00
6.29
.89
6.00
Manipulation check. Manipulations were checked through questions in the information packet given to the participants prior to performing the experimental task. After reading the materials, the participants were asked questions to evaluate their understanding of the instructions (see Appendix K). The questionnaire included such items as “Will you be evaluated?” (1 = yes, 2 = no); “How closely do you believe your actions will be examined?” (1 = not at all closely to 7 = very closely); “How much will you be required to justify your outcomes and strategy?” (1 = not at all required to 7 = very much required); “To receive 20 extra raffle tickets at the end of the task, how many red balls do you have to draw?” (“please circle the appropriate answer: 1, 2, 3, 4, 5”); “What is the maximum number of drawing rounds you can conduct?” (“please circle the
appropriate answer: 3, 7, 10, 15, 17, 19, 20, 22, or as many as I want”); “If you receive
two negative points, what will be the monetary penalty you will incur?” (“please circle the
appropriate answer: $1, $2, $3, $4, $5, $6, $7, $8, $9, $10”), and only for the constituency
63
condition “If you receive three negative points, what will be the monetary penalty incurred by each group member?” (“please circle the appropriate answer: 1, 3, 4, 6, 9, 10”). Two of the experimental manipulations, reporting Yes/No and constituency Yes/No, were checked using an independent-samples t-test. The t-test was performed on items designed to show meaningful differences based on the assigned conditions. Question one (“Will you be evaluated?”) contrasted the difference between the conditions in which the participants had to report in person to either constituency or the researcher (M = 1.07, SD = .26) with the conditions with no reporting (M = 1.38, SD = .49). Results indicated a significant difference between these two conditions t(231) = 6.17, p < .01. In addition, significant differences were also observed in reporting (M = 4.96, SD = 1.43) versus no reporting (M = 4.28, SD = 1.75) conditions on question 2, “How closely do you believe your actions will be examined?” (t[233] = 3.25, p < .01). Furthermore, the results showed that those participants in the no reporting conditions (M = 3.36, SD = 1.78) indicated that they believed that they would not have to justify their outcomes and strategy (question 3) in comparison to those in reporting conditions (M = 4.90, SD = 1.48; t[232] = 7.14, p < .01). For the constituency (M = 1.05, SD = .22) versus no constituency conditions (M = 1.39, SD = .49), only one item (“Will you be evaluated?”) was hypothesized to show significant differences between the conditions. The null hypothesis was rejected (t[231] = 6.70, p < .01), such that those participants in the constituency conditions knew that they would be evaluated. The majority (94.40%) of the individuals assigned to the constituency/reporting condition accurately reported that they would be evaluated by the group members. All (100%) individuals assigned to the no constituency/reporting condition correctly
64
indicated that they would be evaluated by the researcher. In addition, 98.70% of the participants correctly indicated that they would need to draw five red balls to receive 20 extra raffle tickets. Approximately 94% of the respondents reported that they could conduct a maximum of 20 rounds of drawing, as they were instructed. Furthermore, 83% of the participants accurately pointed out that if they were to lose 12 raffle tickets, they would incur a $2 monetary penalty. The results also suggested that the majority of the participants (89.30%) understood that if, on the first four tries, they were to draw only white balls, they would be left with 6 raffle tickets. The data indicated that 81.60% of the individuals assigned to the constituency conditions understood that if they were to lose 13 raffle tickets, each group member would incur a $3 monetary penalty. Luck scale and scenarios. Because the participants were told that the experiment was about luck, to make the situation more believable the participants were asked to fill out a number of luck scales and scenarios borrowed from studies investigating luck (Darke & Freedman, 1997a; Wohl & Enzle, 2002a). The items from these studies were placed into one instrument called “Luck Questionnaire” (see Appendix L). Although luck was not part of the experimental manipulation, reliability of the scales was assessed. The first scale included in the Luck Questionnaire was called Belief in Good Luck (BIGL), a 12-item-scale developed by Darke and Freedman (1997a). This scale was designed to measure the belief that luck is a stable and personal trait. The questionnaire consisted of 6-point Likert scale items ranging from 1 (strongly agree) to 6 (strongly disagree). For example, items 1, 4, and 10 respectively asked, “Luck plays an important part in everyone’s life,” “I believe in luck” and “Even the things in life I can’t control tend to go my way because I’m lucky.” Darke and Freedman reported strong reliability
65
(Cronbach’s
= .85). The reliability analysis using data from the current research study = .78). The review of the scale items
also yielded good reliability (Cronbach’s
suggested that item 7, “It’ s a mistake to base any decisions on how lucky you feel,” and item 12, “Luck is nothing more than random chance,” should be deleted. The reliability analysis was conducted without items 7 and 12 resulting in higher reliability (Cronbach’s = .88). To make the case that this was a study about luck more convincing, three short scenarios about luck were also included in the experimental materials. The scenarios were borrowed from Darke and Freedman (1997a). Originally, the scenarios were used to establish the external validity of the BIGL scale. For example, scenario number 1 read, “If you were walking down a street that was full of people and someone dropped a $20 bill in the middle of the crowd, do you feel that you would: (1) most certainly find it; (2) probably find it; (3) have a slightly better than even chance of finding it; (4) have no feeling one way or the other; (5) have a slightly better than even chance of not finding it; (6) probably not find it; (7) most certainly not find it.” Reliability analysis of these scenarios, however, showed that they were not reliable (Cronbach’s = .26).
In addition to the BIGL scale and luck scenarios, a questionnaire developed by Wohl and Enzle (2002a) was adopted and slightly modified for this study. This questionnaire consisted of four 7-point scale items, which included such items as “When it comes to games of chance (gamble), usually my chances of winning are” (1 = very bad to 7 = very good); “How often do you play games of chance (gamble)?” (1 = never to 7 = more than once a week); “To what extent do you feel that luck is a quality of the person or a quality of the situation?” (1 = quality of situation to 7 = quality of the person), and
66
“If you were to say that you were ‘lucky’ in terms of some event, to what extent do you mean that you a lucky type of person or that a lucky thing happened to you?” (1 = A lucky thing happened to me and 7 = I am a lucky type of person). These items were poorly interrelated (Cronbach’s = .50).
Low reliabilities for the scenarios and Wohl and Enzle’s (2002a) questionnaires had no implications for the results of the study, because luck was not the focus of this research and there were no research hypotheses associated with luck. However, because the BIGL scale did have strong reliability, for exploratory research purposes, a luck factor score was computed on that scale (LUCKSCAL) using confirmatory factor analysis. First, the analysis yielded two components. However, only two items in the second factor had a loading above .50, so another principal component analysis was performed forcing items to load as one factor. All of the new loadings were above .50, and the factor score from this analysis was used for analyses performed for the current research. Table 5 summarizes descriptive statistics for the BIGL scale items. Table 5 Belief in Good Luck Scale: Summary of Descriptive Statistics Standard Deviation (SD) 1.55 1.55 1.40 1.54 1.41 1.29
Question Luck plays an important part in everyone's life Some people are consistently lucky, and others are unlucky I consider myself to be a lucky person I believe in luck I often feel like it's my lucky day I consistently have good luck
Mean (M) 4.30 4.30 3.94 4.34 3.45 3.37
Median 4.00 4.00 4.00 5.00 3.00 3.00
67
Question It's a mistake to base any decision on how lucky you feel Luck works in my favor I don't mind leaving things to chance because I'm a lucky person Even the things in life I can't control tend to go my way because I'm lucky There is such a thing as luck that favors some people, but not others Luck is nothing more than a random chance
Mean (M) 4.67
Standard Deviation (SD) 1.54
Median 5.00
3.58 2.77
1.25 1.23
4.00 3.00
3.11
1.27
3.00
3.78
1.60
4.00
4.76
1.62
5.00
Outcome narrative. This instrument asked participants to summarize the results of the task and explain the st rategy that he or she chose to pursue while performing the drawing (see Appendix M). The respondents were first asked, “What was the color of the ball you drew?” Then, they were asked, “What did you choose to do next?” to which they could respond that they chose to withdraw from the task or continue drawing. If the respondents chose to continue to draw, they were asked in an open-ended question to explain why they made this decision. Participants were asked the same set of questions for each round of drawing. At the end of the questionnaire the respondents were asked to identify some specific statements that they might used to explain what happened during the drawing task. These data were collected to explore whether individuals are likely to use social accounts to explain their behavior. Coding procedures and analysis of the open-ended questions will be discussed in the Results chapter.
68
Entrapment bias scale. The scale measuring cognitive processes contributing to conflict escalation was constructed specifically for this study. It was administered after the participants completed the experimental task (See Appendix N). The instrument used a 7-point Likert scale in which 1 indicated “strongly disagree” and 7 indicated “strongly agree.” The questionnaire was administered to participants in all conditions. The instrument consisted of 20 items for the constituency conditions and no constituency/reporting condition and 18 items for the control condition. For the control condition, two items were omitted because they were not applicable to the respondents. The scale was developed using four factors: internal self-justification, external self-justification, and other- and self-positive face goals. The internal self-justification measured the need to appear to be a rational decision maker (Aronson, 1968). This portion of the scale consisted of six items (i.e., 1, 2, 3, 4, 5, and 16) including “I feel that my strategy choice was rational,” “I made my decision to continue drawing unemotionally,” “I weighted potential risks and gains carefully prior to making each round of drawing,” “I calculated my odds prior to each drawing,” “I felt that my investment in the process will be justified by the eventual pay off,” and “It was important for me to do well on this particular task.” Review of the reliability coefficients (Cronbach’s = .60) for this dimension revealed that one item, “I made my decision to
continue (or stop) drawing unemotionally,” should be omitted, which improved the reliability for this factor (Cronbach’s = .67). Confirmatory factor analysis was used to
create a composite measure of internal self-justification (INTRNJS). The initial factor analysis yielded one factor with all factor loadings above .50.
69
The external self-justification measured the desire to appear rational to others and to demonstrate to them that persisting in a failing course of action was a correct decision in the long term (Staw, 1976). The external self-justification items (i.e., 6, 7, 8, 9, 10) included five questions, among them were such items as “In general, it was important to me that the group members see me as a rational person” and “I wanted the other group members to understand the logic used in choosing my course of action.” Items administered to the participants in the no constituency condition were adapted for relevancy. For example, “I wanted my group members to feel that my choices were wellcalculated” was adapted to read “If I were to perform this task in a group in which I was a group representative, I would want the group members to feel that my choices were wellcalculated.” To improve reliability coefficients (Cronbach’s = .79) on this dimension,
two items, “Under the circumstances, I believe that the course I pursued had the best potential for earning the most extra raffle tickets for my group members” and “I felt it was important to perform well inn this task for my group members,” were omitted resulting in stronger reliability score (Cronbach’s = .83). Confirmatory factor analysis
was used to create a composite measure of external self-justification (EXTRNJS). The initial factor analysis yielded one factor with strong factor loadings, all of them above .50. In addition, the discussion in the preceding sections regarding face argues that participants who have expansive personal networks are likely to be concerned with protecting group image (other-positive face). Five items (i.e., 11, 12, 13, 14, 15) were developed to measure other-positive face goals. Sample items included “I felt it was important to perform well in this task for my group members” and “I was concerned for
70
my group’s needs.” Again, items were adapted to be applicable to all conditions. For example, the item “I wanted to help my group” was administered to the participants in constituency present conditions; for the no constituency conditions the item was adapted to read, “When working in a team, I want to help my group.” Reliability analysis resulted in strong reliability (Cronbach’s = .87), but also suggested that item 15, “I was worried
what the group members will think about me after the task was completed,” should be removed, which resulted in even greater reliability (Cronbach’s = .91). Confirmatory
factor analysis was used to create a composite measure of other-positive face goals (OTHRPENT). The initial factor analysis yielded one factor with strong factor loadings, all of them above .50. Items measuring the self-positive face dimension (i.e., 17, 18, 19, 20, 21, 22) included “I was worried what [the researcher/group members] will think about me after the task was completed,” “It was important for me to do well on this task,” “I wanted to perform well to make a positive impression on [the researcher/group members],” “I don’t want my group members to be mad at me for loosing some of their raffle tickets and/or money,” “I think this was an unfair task” and “I didn’t want [the researcher/group members] to think that I’m unable to calculate the odds.” However, because the items were largely inapplicable to the control group—that is, by definition, this condition (no reporting/no constituency) did not include any threats to self-positive face—this condition was excluded from the analysis for the control group. Reliability analysis on items for conditions I, II, and III revealed that reliability of the scale for this dimension could be improved (Cronbach’s = .67) if items 19, 21 and 22 were omitted (Cronbach’s
= .80). Confirmatory factor analysis was used to create a composite measure of self-
71
positive face goals (SELFPENT). The initial factor analysis yielded one factor with strong factor loadings, all of them above .50. Overall, to measure how much internal and external self-justification and otherand self-positive face goals contributed to the entrapment bias and to distinguish which of the factors contributed to the entrapment, four new variables were created: INTRNJS (measure for internal justification); EXTRNJS (measure for external justification); OTHRPENT (measure for other positive face), and SELFPENT (measure for own positive face). Again, the self-positive face factor score (SELFPNET) did not include COND IV. Tables 6 through 9 summarize descriptive statistics for the entrapment bias scale items by each condition. Table 6 Entrapment Bias Scale: Summary of Descriptive Statistics for Condition I Standard Deviation (SD) 1.60 1.89
Question
I feel that my strategy was rational I made my decision to continue (or stop) drawing unemotionally I weighed potential risks and gains carefully prior to making each round of drawing I calculated my odds prior to each drawing I felt that my investments in the process will be justified by the eventual pay off Under the circumstances, I believe that the course I pursued had the best potential for earning most extra raffle tickets for my group members
Mean (M) 5.38 5.05
Median 6.00 6.00
3.87
1.96
4.00
2.91 3.96
1.87 1.90
3.00 4.00
4.29
1.69
4.00
72
Question
I wanted the other group members to understand the logic used in choosing my course of action I wanted my group members to feel that my choices were well-calculated In general, it was important to me that the group members see me as a rational person I felt it was important to perform well in this task for my group members I was concerned for my group’s needs I wanted to help my group I did not want to let my group down When performing the task it was important to me to consider the consequences my behavior might have for my group I was worried what the group members will think about me after the task was completed It was important for me to do well on this task I wanted to perform well to make a positive impression on my group members I don’t want my group members to be mad at me for loosing some of their raffle tickets and/or money I think this was an unfair task I didn’t want my group members to think that I’m unable to calculate the odds
Mean (M) 4.91
Standard Deviation (SD) 1.58
Median 5.00
4.80
1.48
5.00
4.65
1.76
5.00
4.94
1.56
5.00
4.75 5.40 5.24 5.04
1.52 1.24 1.42 1.52
5.00 5.00 5.00 5.00
3.73
1.60
4.00
4.44 4.27
1.60 1.72
4.00 4.00
5.35
1.51
5.00
3.93 4.04
1.84 1.68
4.00 4.00
73
Question
I don’t think I understood the instructions well I believe that any person in my position would have behaved the same way I did
Mean (M) 3.00 4.89
Standard Deviation (SD) 1.89 2.00
Median 3.00 5.00
Table 7 Entrapment Bias Scale: Summary of Descriptive Statistics for Condition II Standard Deviation (SD) 1.49 1.45
Question
I feel that my strategy was rational I made my decision to continue (or stop) drawing unemotionally I weighed potential risks and gains carefully prior to making each round of drawing I calculated my odds prior to each drawing I felt that my investments in the process will be justified by the eventual pay off Under the circumstances, I believe that the course I pursued had the best potential for earning most extra raffle tickets for my group members I wanted the other group members to understand the logic used in choosing my course of action I wanted my group members to feel that my choices were well-calculated In general, it was important to me that the group members see me as a rational person
Mean (M) 5.55 5.40
Median 6.00 6.00
4.43
2.04
5.00
3.29 4.48
1.95 1.72
3.00 4.00
4.28
1.79
4.00
5.02
1.63
5.00
4.83
1.71
5.00
5.14
1.33
5.00
74
Question
I felt it was important to perform well in this task for my group members I was concerned for my group’s needs I wanted to help my group I did not want to let my group down When performing the task it was important to me to consider the consequences my behavior might have for my group I was worried what the group members will think about me after the task was completed It was important for me to do well on this task I wanted to perform well to make a positive impression on my group members I don’t want my group members to be mad at me for loosing some of their raffle tickets and/or money I think this was an unfair task I didn’t want my group members to think that I’m unable to calculate the odds I don’t think I understood the instructions well I believe that any person in my position would have behaved the same way I did
Mean (M) 5.09
Standard Deviation (SD) 1.48
Median 5.00
4.86 5.45 5.55 5.21
1.56 1.38 1.37 1.39
5.00 6.00 6.00 5.00
4.29
1.68
4.00
4.57 4.33
1.54 1.46
5.00 4.00
5.07
1.58
5.00
4.14 4.03
1.89 1.73
4.00 4.00
2.60 5.03
1.85 1.57
2.00 5.00
75
Table 8 Entrapment Bias Scale: Summary of Descriptive Statistics for Condition III Standard Question Mean (M) Deviation (SD) Median
I feel that my strategy was rational I made my decision to continue (or stop) drawing unemotionally I weighed potential risks and gains carefully prior to making each round of drawing I calculated my odds prior to each drawing I felt that my investments in the process will be justified by the eventual pay off Under the circumstances, I believe that the course I pursued had the best potential for earning most extra raffle ticket I wanted the researcher to understand the logic used in choosing my course of action I wanted the researcher to feel that my choices were wellcalculated In general, it was important to me that the researcher sees me as a rational person In general, when performing in a group/team environment, it is important to me to perform well for my group members When working in a team, I’m concerned for my group’s needs When working in a team, I want to help my group When working in a group, I do not want to let my team down
5.13 4.95
1.63 1.89
5.00 5.00
4.36
1.93
5.00
3.32 4.13
1.98 1.61
3.00 4.00
3.63
1.90
4.00
3.89
1.74
4.00
3.91
1.71
4.00
3.75
1.70
4.00
6.04
0.97
6.00
5.95
.90
6.00
6.11 6.25
1.06 1.06
6.00 6.50
76
Question
When working in a group, it is important to me to consider the consequences my behavior might have for my group When working in a group, I worry what the group members will think about me after the project is completed It was important for me to do well on this particular task I wanted to perform well to make a positive impression on the researcher I don’t want the researcher to think less of me for losing raffle tickets and/or money I think this was an unfair task I didn’t want the researcher to think that I’m unable to calculate the odds I don’t think I understood the instructions well I believe that any person in my position would have behaved the same way I did
Mean (M) 6.20
Standard Deviation (SD) .92
Median 6.00
5.25
1.47
6.00
4.11 3.57
1.65 1.72
4.00 4.00
3.34
1.67
3.00
3.41 3.34
1.91 1.62
3.00 3.00
2.14 4.20
1.45 1.35
2.00 4.00
Table 9 Entrapment Bias Scale: Summary of Descriptive Statistics for Condition IV Standard Deviation (SD) 1.66 1.61
Question
I feel that my strategy was rational I made my decision to continue (or stop) drawing unemotionally
Mean (M) 5.60 5.57
Median 6.00 6.00
77
Question
I weighed potential risks and gains carefully prior to making each round of drawing I calculated my odds prior to each drawing I felt that my investments in the process will be justified by the eventual pay off Under the circumstances, I believe that the course I pursued had the best potential for earning most extra raffle ticket In general, when I do something it is important to me that others understand the logic I used in choosing my course of action. If I were to perform this task in a group in which I was a group representative, I would want the group members to feel that my choices were well-calculated. In general, when working in groups, it is important to me that group members see me as a rational person In general, when performing in a group/team environment, it is important to me to perform well for my group members When working in a team, I’m concerned for my group’s needs When working in a team, I want to help my group When working in a group, I do not want to let my team down
Mean (M) 4.15
Standard Deviation (SD) 2.29
Median 5.00
2.93 3.82
1.91 1.70
2.00 4.00
4.00
1.77
4.00
4.47
1.73
4.00
5.63
1.21
6.00
5.88
1.04
6.00
6.13
0.89
6.00
5.95
1.05
6.00
6.27 6.38
0.86 0.86
6.00 7.00
78
Question
When working in a group, it is important to me to consider the consequences my behavior might have for my group When working in a group, I worry what the group members will think about me after the project is completed It was important for me to do well on this particular task When working in a group, it is important to me to perform well to make a positive impression on my team members I think this was an unfair task I don’t think I understood the instructions well I believe that any person in my position would have behaved the same way I did
Mean (M) 6.08
Standard Deviation (SD) 0.94
Median 6.00
4.43
1.57
5.00
3.90 5.72
1.89 1.20
4.00 6.00
3.48 2.47 4.10
1.89 1.64 1.56
4.00 2.00 4.00
79
Table 10 Scale Reliability Coefficients Scale Influence scale Face scale Self-positive face Other-positive face Self-negative face Other-negative face Belief in good luck Entrapment bias scale Internal self-justification External self-justification Other-positive face Self-positive face (only conditions I, II and III) 0.80 0.67 0.83 0.91 0.73 0.75 0.71 0.91 0.88 Cronbach’s 0.82
80
CHAPTER 3 Results This chapter presents the results of the study. This research explores a number of factors affecting entrapment bias. The discussion below presents the findings for each of the hypotheses and research question. Data Preparation A number of transformations were performed in attempts to correct skewness in the data while maintaining variance homogeneity. However, no transformation was found that would eliminate skewness. Overall, most variables had no or less than three missing values. Missing values were a result of respondents leaving the items blank, apparently because they either refused to or forgot to answer the questions. Table 11 presents a correlation matrix for the independent and dependent variables.
81
Table 11 Correlation Matrix: Independent and Dependent Variables
Overall face Overall face Overall face Overall face concern: concern: concern: Other concern: Self Other positive Self positive negative face negative face face scale face scale
Total number of draws Constituency (present or not) Reporting (yes or no) Overall face concern: Other negative face Overall face concern: Self negative face Overall face concern: Other positive face scale Overall face concern: Self positive face scale Network size multiplied by strength of network influence Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other Positive Face Entrapment scale: Self positive face (Conditions 1,2 and 3) 0.02 0.13* -0.01 0.04 -0.10 0.07 -0.03 0.07 -0.23** -0.23** -0.20** -0.05
Constituency (present or not)
Reporting (yes or no)
0.01 0.01 0.02 0.10 0.02 0.01 0.08 0.06 0.09 -0.41** 0.34** 0.06 0.01 -0.12 0.04 0.05 -0.03 -0.04 -0.30** -0.05 -0.16* 0.47** 0.26** 0.15* 0.07 0.07 0.05 0.10 0.07 0.12 0.33** 0.30** 0.06 0.08 0.08 0.10 0.04 0.20** 0.34** 0.03 0.13* 0.07 0.20** 0.13* 0.30** 0.13 0.02 0.05 0.16* 0.25** 0.24** .
82
Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other Positive Face Entrapment scale: Self positive face (Conditions 1,2 and 3)
Network size multiplied by strength of network influence 0.03 0.10 0.13 0.13 0.15*
Luck
Entrapment scale: Internal justification
Entrapment scale: External justification
Entrapment scale: Other Positive Face
0.17* 0.06 0.09 0.15* 0.50** 0.31** 0.41** 0.42** 0.60** 0.33**
* Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed).
83
Table 12 Independent Variables: Factor Score Descriptive Statistics for Condition I Variable Mean Std. Deviation
Total Number of Draws Overall face concern: Other negative face Overall face concern: Self negative face scale Overall face concern: Other positive face scale Overall face concern: Self positive face scale Network size multiplied by strength of network influence Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other positive face Entrapment scale: self positive face
10.52 0.15 -0.01 -0.01 -0.07 5.80 -0.03 -0.09 .019 -0.49
4.72 1.04 1.00 1.00 1.04 35.89 0.86 0.97 1.00 1.06
0.26
0.95
84
Table 13 Independent Variables: Factor Score Descriptive Statistics for Condition II
Variable Total Number of Draws Overall face concern: Other negative face Overall face concern: Self negative face scale Overall face concern: Other positive face scale Overall face concern: Self positive face scale Network size multiplied by strength of network influence Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other positive face Entrapment scale: self positive face
Mean 10.41 -0.02 0.03 -0.23 0.14 -0.39 -0.03 0.00 -0.63 0.38
Std. Deviation 5.13 0.98 1.04 0.91 0.99 40.79 1.10 0.94 1.05 0.68
-0.48
0.96
85
Table 14 Independent Variables: Factor Score Descriptive Statistics for Condition III Variable Mean Std. Deviation
Total Number of Draws Overall face concern: Other negative face Overall face concern: Self negative face scale Overall face concern: Other positive face scale Overall face concern: Self positive face scale Network size multiplied by strength of network influence Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other positive face Entrapment scale: self positive face
10.52 0.15 -0.01 -0.01 -0.07 5.80 -0.03 -0.09 .019 -0.49
4.72 1.04 1.00 1.00 1.04 35.89 0.86 0.97 1.00 1.06
0.26
0.95
86
Table 15 Independent Variables: Factor Score Descriptive Statistics for Condition IV Variable Mean Std. Deviation
Total Number of Draws Overall face concern: Other negative face Overall face concern: Self negative face scale Overall face concern: Other positive face scale Overall face concern: Self positive face scale Network size multiplied by strength of network influence Luck Entrapment scale: Internal justification Entrapment scale: External justification Entrapment scale: Other positive face
9.03 -0.01 -0.07 1.03 -0.13 1.52 -0.11 -0.12 0.41 0.43
5.49 0.99 1.07 1.16 1.07 42.61 1.08 1.06 0.72 0.71
87
Test of the Hypotheses Hypothesis 1 (H1) posited that entrapment will be more likely when a constituency is present then when it is not present. To test this hypothesis, a one-way ANOVA was performed, with constituency (present versus not present) as the independent variable, and entrapment—measured by the number of balls drawn — as the dependent variable. Presence of constituency did not affect entrapment, F(1, 235) = .07, n.s. Therefore, H1 was not supported. Hypothesis 2a (H2a) posited that internal and external justification would be greater when respondents are required to report their behavior. Two ANOVAs were performed with reporting conditions dichotomized (i.e., reporting yes or no) as the independent variable, and internal and external justification factor scores, derived from the entrapment scale, as the dependent variables. The results indicated no significant effect for internal self-justification, F(1, 228) = .45, n.s. However, a significant main effect was found for external self-justification, F(1, 228) = 22.11, p < .01. That is, although requiring an individual to report his or her behavior had no effect on the need to internally self-justify behavior, the requirement to report did affect individuals’ need to appear rational to others. Therefore, H2a was partially supported. Hypothesis 2 (H2b) stated that when individuals have to report their behavior, their need for internal and external self-justifications will lead to greater entrapment. To test this hypothesis, only cases in which reporting was required were selected (n = 116). A regression analysis was performed with internal and external self-justification factor scores derived from the entrapment scale serving as independent variables, and the total number of draws—or entrapment—as the dependent variable. The results showed that the
88
overall analysis was significant, F(2, 110) = 7.30, p < .01. However, consistent with the finding presented above, only the effect of internal justification was significant, with a negative relationship to entrapment, t(110) = 3.47,ß = -1.94, p < .01. That is, among the individuals who had to report their behavior, the greater need for internal selfjustification led to lower levels of entrapment. The external justification had no effect on entrapment, t(110) = .46, n.s. Therefore, H2b was partially supported. Hypothesis 3 (H3) posited that the presence of constituency would lead to greater internal and external justification. Two ANOVAs were performed with constituency dichotomized (i.e., present or not) as the independent variable, and internal and external justification as the dependent variables. The results demonstrated no significant main effects for either test (internal justification: F[1, 228] = .83, n.s.; external justification: F[1, 228] = 1.83, n.s.). Therefore, H3 was not supported. Hypothesis 4 (H4) stated that individuals with greater self- and other-positive face concerns would be more likely to become entrapped. Prior to testing this hypothesis, a regression analysis was performed using self- and other-positive and self- and othernegative face scale scores derived from the face scale instrument administered prior to the experimental task as the independent variable and total number of draws as the dependent variable. The regression analysis was performed to test if general concern for face affects entrapment. The overall relationship was non-significant, F(4, 232) = 1.24, n.s. However, the results provided marginal support for the relationship between other positive face goals and entrapment, which approached significance, t(114) = 1.86, p = .06, = -.71. That is, the results suggest that the greater the person’s concern for other’s
positive face, the less likely the person is to become entrapped.
89
Next, to test the hypothesis that individuals’ face-saving goals affect entrapment, two regression analyses were performed using self-positive and other-positive face goal (derived from the entrapment scale administered after participants completed the experimental task) as independent variables and the total number of draws as the dependent variable. It should be noted that the analysis for self-positive face was performed only on experimental conditions 1, 2, and 3. The control group (condition 4) was excluded from the analysis because the experimental design was such that individuals in this condition should not have experienced any threats to self-positive face. Because individuals in this condition should not have experienced any threats to selfpositive face, questions about self-positive face differed for this group and this analysis could not be run. The results showed that other-positive face goals were negatively associated with entrapment, F(1, 228) = 9.20, p < .01; = -1.03. This result is consistent with the finding
reported above regarding individuals’ general other-positive face concerns and entrapment. That is, the results indicated that the individuals were less likely to get entrapped if they had significant concerns for other-positive face. However, self-positive face goals did not affect entrapment, F(1, 168) = .35, n.s. Therefore, H4 was partially supported. Hypothesis 5 (H5) posited that concern for saving self- and other-positive face would be greater when the individuals have to report their behavior. To test this hypothesis, two ANOVAs were performed with reporting conditions dichotomized (yes/no) as the independent variable and self- and other-positive face factor scores for entrapment as dependent variables. In self-positive face analysis, only constituency
90
present and reporting conditions (i.e., conditions 1, 2, and 3) were used, because, in the control condition, the participants should have not experienced any threats to self-positive face. The results revealed a reliable main effect across experimental conditions, F(1, 168) = 4.30, p < .05, such that concern for saving self-positive face was greater when individuals had to report their behavior. However, a similar two-way ANOVA with reporting conditions dichotomized (yes/no) as the independent variable and other-positive face as the dependent variable, using participants in all of the conditions, demonstrated no significant main effect, F(1, 228) = .53, n.s. Therefore, H5 was partially supported. Hypothesis 6 (H6) posited that overall other- and self-positive face concerns (as opposed to face concerns related to the experimental manipulation) would lead to greater need to justify one’s action in a situation that could lead to entrapment. To test this hypothesis, four separate regression analyses were performed with self- and otherpositive face, derived from the face scale instrument administered prior to the experimental task, as the independent variables and internal and external justification as dependent variables. The results indicated that face concerns did not affect internal justification (self-positive face: F[1, 226] = .48, n.s.; other-positive face F[1, 225] = .96, n.s.). However, the relationships between external self-justification and other- and selfpositive face were significant (other-positive face F[1, 225] = 9.08, p < .01; self-positive face F[1, 226] = 5.67, p