Description
This study investigates the interaction effects of locus of control, a personality variable, and ethical
reasoning on the behaviour of auditors in an audit conflict situation. Eighty experienced auditors from a
sample of Big Six and Non-Big Six CPA firms in Hong Kong were provided with a case study involving an
audit conflict situation and were asked to state the extent to which they would accede to the client’s
request. Subjects were also administered Rotter’s Locus of Control Scale and the Defining Issues Test
Pergamon Accountfug, Organfzations and Sockty, Vol. 21, No. 1. PP. 41-51, 1996
Copyright 0 1995 Elsevier Science Ltd
Printed in Great Britain. AU rights reserved.
0361-3682/% $15.00+0.00
03613682(95)00009_7
AUDITORS’ BEHAVIOUR IN AN AUDIT CONFLICT SITUATION: A RESEARCH
NOTE ON THE ROLE OF LOCUS OF CONTROL AND ETHICAL REASONING*
JUDY S. L. TSUI
City University of Hong Kong
and
FERDINAND A. GUL
The Chinese University of Hong Kong
Abstract
This study investigates the interaction effects of locus of control, a personality variable, and ethical
reasoning on the behaviour of auditors in an audit conflict situation. Eighty experienced auditors from a
sample of Big Six and Non-Big Six CPA firms in Hong Kong were provided with a case study involving an
audit conflict situation and were asked to state the extent to which they would accede to the client’s
request. Subjects were also administered Rotter’s Locus of Control Scale and the Defining Issues Test
(DITJ to measure ethical reasoning. Analyses of the data using multiple regression found that ethical
reasoning moderated the relationship between locus of control and the auditors’ responses to accede to
client’s request in an audit contlict situation. An implication of these results is that the explicit
recognition of both locus of control and ethical reasoning provides a better explanation for differences
in auditors’ ethical decision making
The issue of ethical behaviour of accountants
has recently been a source of much concern
and research interest (Cohen et al., 1993). A
discernable strand of thls research interest
attempts to unravel and evaluate the factors
that could affect ethical behaviour of accoun-
tants (Shaub, 1989; Ponemon, 1990; Ponemon
& Gabhart, 1993). These studies emphasized
the notion that the cognitive processes of indi-
viduals should not be treated as a “black box”,
but should be explicitly recognized since these
processes are expected to affect individuals’
ethical decision making (Ponemon, 1990;
1992a; 1992b; Ponemon & Gabhart, 1990).
More specifically, these studies “incorporate
or expressly consider a psychological frame-
work for the ethical reasoning process” in
order to answer the question what makes
accountants more or less ethical? (Jones &
Ponemon, 1993, p. 411).
The psychological framework for ethical rea-
soning which is based on Kohlberg’s (1969)
Theory of Cognitive Moral Development uses
the Defining Issues Test (Rest, 1979) to mea-
sure an individual’s level of ethical reasoning.
Using the P scores, ’ individuals are classified
l The authors would like to acknowledge the comments of Mark Young and Chee Chow, two anonymous reviewers and
the editor. The research instruments used in this study are available from the authors.
1 The P score measures the relative importance of stages five and six of ethical reasoning. Ponemon (1992b, p. 182)
suggests that “the higher P score implies a lower percentage of stages 1 through 4.”
41
42 J. S. L. TSUI and F. A. GUL
into different levels of ethical reasoning. Based
on this theory, several studies have found sup
port for the notion that the levels of ethical
reasoning are related to auditors’ professional
judgments (Ponemon, 1990; 1992a; 1992b;
Ponemon & Gabhart, 1990; Tsui, 1994). For
example, Ponemon (1992b) found that those
auditors with lower levels of ethical reasoning
tended to underreport their chargeable time
more severely than those with higher levels
of ethical reasoning.
Unfortunately, consideration of the cognitive
process by way of the ethical reasoning pro-
cess, by itself, is an inadequate characteriza-
tion of the “black box” since it fails to
recognize the role of other “individual” differ-
ences, such as personality which may be
defined as individuals’ attitudes and beliefs
(Pratt, 1980; Gul, 1984). Trevino (1986, p.
602) recognized this and argued that:
the individual’s cognitive moral development stage
determines how an individual thinks about ethical
dilemmas, his or her process of deciding what is right
or wrong in a situation. However, cognitions of right
and wrong are not enough to explain or predict ethical
decision-making behaviour. Additional individual and
situational variables interact with the cognitive compo
nent to determine how an individual is likely to behave
in response to an ethical dilemma.
Trevino (1986, p. 602) went on to suggest that
the individual differences variable, or speciti-
tally such personality variables as locus of con-
trol (Rotter, 1966), could interact with ethical
reasoning to affect individual behaviour in an
ethical dilemma. Locus of control is defined as
(MacDonald, 1976, p. 169):
the extent to which persons perceive contingency
relationships between their actions and their outcomes.
Persons who believe that they have some control over
their destinies are called ‘internals’; that is, they believe
that at least some control resides within themselves.
Xxtemafs’, on the other hand, believe that their out-
comes are determined by agents or factors extrinsic to
themselves, for example, by fate, luck, chance, power-
ful others or the unpredictable.
Trevino (1986) suggested that since exter-
nals rely on fate, luck and chance, they are
more likely to behave unethically than inter-
nals who are able to rationalize and control
their behaviour. Similarly, internals who
believe that events are under their control
would take responsibility in their determina-
tion of right and wrong, thus behaving more
ethically than externals.
Rotter (1966) argued that the intemal-
external construct is a unidimensional person-
ality variable that is stable (Ashlcanasy, 1985, p.
1328). Thus, personality traits such as locus of
control are relatively more stable in an indivi-
dual (Roford & Penno, 1992, p. 128) than ethi-
cal reasoning levels since these levels may be
improved through ethical intervention and
formal education.
Thus, a more fruitful approach to the study
of how auditors make ethical decisions should
explicitly recognize that ethical reasoning
levels and locus of control would jointly influ-
ence ethical decision making (Trevino, 1986).
Such a model (Fig. 1) is likely to significantly
improve attempts at not only predicting and
understanding ethical behaviour, but also
developing ethical intervention strategies. The
present study which extends previous account-
ing ethics literature explores and tests the
notion that the effects of locus of control on
auditors’ decision making depend on an indivi-
dual’s level of ethical reasoning.
The particular decision-making task that is
selected for study is the auditors’ ability to
resist management pressure in an audit contlict
situation. Two reasons motivated this decision.
First, client pressure in an audit conflict situa-
tion is prevalent in the audit environment. Evi-
dence regarding the prevalence of
management pressure was, for example, pro-
vided by Finn et al . (1988) who found that
Ethical Reasoning
Locus of Control
*
b Auditor’s Response ID
Accede IO Client’s Request
Fig. 1
AUDITORS’ BEHAVIOUR IN AN
CPAs identified client pressure to alter tax
reports and con&t of interest/ Independence
situations as posing the “most difficult ethical
situations” (Glory et aZ., 1992, p. 286). Second,
there is no systematic empirical evidence link-
ing locus of control and ethical reasoning to
auditors’ behaviour in confIict situations.*
Audit conflict is thus a source of much con-
cern and research interest (Knapp, 1985; Gul,
1991). In particular, the problem regarding the
auditors’ ability to resist management pressure
in audit conflict situations was identified as
early as 1938 (Mckesson & Robbins, 1941)
and more recently by the Cohen Commission
(AICPA, 1978). Arising from this, prior studies
have concentrated on the effects of different
contextual variables, such as client financial
condition, size of audit fees and management
advisory services, on the auditors’ ability to
resist management pressures in audit conflict
situations (Knapp, 1985, 1987; GuI, 1991; Gul
& Tsui, 1992). No study has examined the ques-
tion of whether locus of control and ethical
reasoning will jointly affect auditors’ beha-
viour in an audit conflict situation.
In addition, the theoretical linkages between
personality, ethical reasoning and ethical beha-
viour need to be empirically demonstrated
before the accounting profession and educa-
tors attempt to undertake ethical interventions
to promote more ethical professional beha-
viour since such interventions may be more
effective for individuals with particular person-
ality traits. Evidence on these issues will pro-
vide a better understanding of the ethical
behaviour of accountants. In summary, this
study adds to the literature on auditors’ ethical
behaviour in emphasizing the “processes”3
involved in ethical decision making by consid-
ering two issues. First, it takes into account the
AUDIT CONFLICT SITUATION 43
joint influence of ethical reasoning and lndivi-
dual differences on auditors’ ethical behaviour,
and second, it sheds some light on the beha-
viour of auditors ln an audit conflict situation.
HYPOTHESIS DEVELOPMENT
The research value and endurance of locus of
control as a significant predictor of human
behaviour in a wide variety of situations is
welI documented (Jennings & Zeithmal, 1983;
Findley & Cooper, 1983). In accounting litera-
ture, it has been studied in different decision
contexts such as, for example, in budgetary
participation (Brownell, 1982; Frucot &
Shearon, 1991). Based on locus of control the-
ory, it is likely that auditors’ behaviour in an
audit conflict situation will be infhtenced by
whether he/ she is an internal or an external.
For example, since an “external” believes
that outcomes are beyond control and can be
attributed to fate, luck or destiny, he/ she is less
likely “to take personal responsibility for the
consequences of ethical/ unethical behaviour
and is more likely to rely on external forces”
(Trevino, 1986, p. 601). Thus, individuals who
are externals are more likely to accede to a cli-
ent’s request in an audit conflict situation. On
the other hand, since an “internal” is more
likely “to take responsibility for consequences
and rely on his or her internal determination of
right and wrong to guide behaviour” (Trevino,
1986, p. 601) he/ she is less likely to accede to
client pressure in an audit conflict situation.
This relationship between locus of control
and auditors’ behaviour, however, may be
dependent on the auditors’ ethical reasoning
since this cognitive variable has been found
to affect the behaviour of accountants in a
* Conflict theory (Thomas, 1992) suggests that auditor’s con&t with the client can pose an ethical dilemma for the
auditor.
3 Noreen (1988, p. 359) in the context of agency theory, concluded that “. an agreement (i.e. ethical code) to abstain
from opporhinistic behaviour cannot be enforced effectively by external rewards or sanctions; instead, the sanctions for
unethical behaviour must be internaked.” Flory el al. (1992, p, 288) pointed out that “the intemahzation of these
sanctions calls for better understanding of ethical behaviour.”
44 J. S. L. TSLJI and F. A. GUL
variety of ethical decision-making situations.
For example, Ponemon & Gabhart (1990) in
developing an experimental case to examine
auditors’ ethical judgment in an independence
context, found that auditors at lower levels of
moral cognitive development are more sensi-
tive to factors relating to penalty than affilia-
tion in making independence judgments.
Ponemon’s (1992b) study also found that audi-
tor underreporting is signiIicantly related to the
level of ethical reasoning. Those auditors with
lower DIT scores underreported their charge-
able time much more severely than those with
higher DIT scores. The evidence suggests that
individuals with higher DIT scores are more
likely to resist management pressure in an
audit con&t situation.
Thus, following Trevino’s (1986) suggestion,
it is likely that locus of control and levels of
ethical reasoning are important components
of the “black box” and would jointly influence
individuals’ behaviour in an audit conflict situa-
tion. Locus of control is treated as the primary
independent variable with ethical reasoning as
a moderating variable since, as pointed out
earlier, the literature suggests that locus of con-
trol is a relatively more stable characteristic
(Ashkanasy, 1985; Koford & Penno, 1992)
than ethical reasoning levels which could be
improved by education and other forms of
intervention (Ponemon & Gabhart, 1993).
Based on this reasoning, for example, an exter-
nal who has low DIT scores is more likely to
accede to a client’s request in an audit conflict
situation than an external who has high DIT
scores. This reasoning and evidence provided
above suggest the following hypothesis:
Hl: The effects of locus of control on the extent to
which an auditor would respond as acceding to client’s
request in an audit conflict situation wIII depend on the
level of ethical reasoning.
METHODS
The research instrument was administered in
the premises where the subjects work. This
ensures an adequate response rate in which
subjects were requested to complete the
research instrument in the presence of the
researcher. It is also an advantage for the
researcher to be able to observe important
behaviours such as the conscientiousness of
the subjects and the time taken in completing
the research instrument. A pilot test of the
research instruments with 11 academics who
are professional accountants and had prior
auditing experience as well as 10 experienced
auditors was conducted prior to administering
the questionnaires.
Subj ects
Si nce the subjects in this study are required
to make a decision on an auditor-client conflict
situation, auditors with professional qualifica-
tion and at least four years of audit experi-
ence* were selected as respondents. Qualified
subjects were selected by contact managers/
partners in each of their firms to participate
in this study. The subjects came from four Big
Six and five Non-Big Six CPA tirms in Hong
Kong. A sample size of 80 senior experienced
auditors was obtained for analysis (18 subjects
were dropped from the analysis due to failure
to satisfy the consistency check in the DIT).
The mean age was 31.56 years (range from
26-47) and mean auditing experience was 8.4
years (range from 4-28). Of the 80 respon-
dents, 51 came from the “Big Six” and ranged
in positions from partners (14), managers (46)
supervisors (16) to seniors (4).
Measurement of vari abl es
Locus of control was measured using Rotter’s
Social Reaction Inventory (1966),5 a 23 item
’ The contact partners/managers of CPA tirms who participated in this study confirmed that auditors with four years of
audit experience in Hong Kong would normally handIe auditor-cIient con%? situations.
5 See Rotter (1966) for test-retest reIiabiIity, convergent validity, discrhninant vaklity and other internal consistency tests.
AUDITORS’ BEHAVIOUR IN AN
forced-choice scale. It consists of 23 paired
statements. Each pair statement is character-
ized by an internal statement and an external
statement. One point is awarded for each exter-
nal statement. Scores can range from 0 to 23.
Low scores on the scale were classified as inter-
nals while those with high scores were classi-
fied as externals.
The widely used Defining Issues Test6 (Rest,
1979) was employed to measure the subjects’
cognitive moral development levels.’ Even
though the P score from the DIT is not exactly
equivalent to Kohlberg’s Moral Judgment Inter-
view (Colby & Kohlberg, 1987) it is consid-
ered an appropriate surrogate to measure the
relative importance on stages five and six ethi-
cal reasoning.
The DIT measure employed in this study
requires participants to analyze four hypotheti-
cal moral dilemmas.’ For each dilemma, the
participants have to rank order, from a list of
twelve items, four that are most significant in
influencing the resolution of each dilemma.
From the most important ranked question to
the least important one, a weighting of the
scale from 4 points being the most important
to 1 as least important is given. The P score is
calculated from the sum of the weighted total
of the four most important ranked items. It is
expressed as a continuous variable from 0 to
100%.
The DIT scoring procedure includes an M
L
AUDIT CONFLICT SITUATION 45
(meaningless) score which acts as a consis-
tency and social desirability check on the
data. The M score indicates the extent to
which the subjects select items that are lofty
sounding but meaningless. Rest (1993, p. 13)
suggested that data obtained from participants
with an M score equal to or greater than eight
for the six-story version should be discarded
from analysis. Since a four-story version of the
DIT is used, an M score of five would propor-
tionately be the cutoff point for elimination
from the analyses. The data were also checked
for consistency and those who failed the test (a
total of 18 subjects) were also eliminated from
the sample for purposes of analysis.
The extent to which the auditor was willing
to accede to client pressure in an auditor-client
conflict situation was measured by using an
adapted version of a short case developed by
Knapp (1985) and used by GUI (1991) (see
Appendix). In the case,’ the auditor disagrees
with management that certain unrecorded
liabilities are immaterial and should not be
recorded in the financial statements. The sub
ject was requested to role play as the auditor of
the company and indicate the likelihood that
he/ she will ignore the unrecorded liabilities.
A reasonable inference can then be made
regarding the extent to which the auditor is
willing to resist client pressure in an audit con-
flict situation and it is therefore an indication of
unethical behaviour. lo
” See Rest (1979) for validation of the DIT in terms of predictive validity, face validity, criterion group validity, longitudinal
change studies, convergent-divergent validity, experimental enhancement studies and discriminant validity For test-retest
reliability and Alpha Cronbach, see Rest (1986).
’ The Moral Judgment Interview developed by Colby & Kohlberg (1987) is also recognized as one of the very reliable and
valid instruments for measuring cognitive moral development. However, the DlT is adopted because it is an objective
recognition instrument which is considered more scientihc, valid and reliable.
* The original DIT has six hypothetical moral dilemmas. However, a four-story version is used in this study because the
other two stories are considered too specific for Western cultures (Ma, 1988). The subjects were administered the English
version of the DIT even though their tirst language (ii most cases) would not be English. It is considered appropriate in
these circumstances because these subjects are qualified accountants and are well versed in English.
9 Preliminary discussions with experienced auditors on the conduct of the study indicated that a short case was preferable
in view of the severe constraints on auditors’ time.
i0 Knapp (1985) argued that an inference can also be made regarding auditor independence levels.
46 J. S. L. TSUI and F. A. GUL
A continuous scale from 0 to 100% was used
to measure the likelihood that the auditor wiii
ignore the unrecorded liabilities as recom-
mended by management. A lower Rkellhood
that the auditor will ignore the unrecorded
liabilities is indicative of the extent to which
the auditor will accede to a client’s request
and therefore it is suggested that through the
auditor’s response, one can infer that he/she is
more/less independent. The higher the iikeli-
hood that the auditor will ignore the unrec-
orded liabilities, the greater is the extent to
which the auditor will accede to the client’s
request. Such a decision may reflect the extent
of the auditor’s independence. This interpreta-
tion is consistent with Ponemon’s (1992b, p.
176) viewpoint that the behaviour of the audi-
tor underreporting of time reflects “the ethical
values that the individual brings to the task.” In
the present study, the auditor’s response in
acceding to the client’s request should similarly
reflect some aspects of an individual’s beliefs
on independence and ethical inclinations.
As part of the debriefing questions, respon-
dents were asked whether they bad encoun-
tered similar conflict cases and to describe
the most memorable one. Most of the respon-
dents indicated that they had encountered
similar cases before in their auditing careers.
Data analysis
Multiple regression was used to analyze the
data (Pedhazur, 1982). To test for twoway
interaction effects, the foilowing multiplicative
model is employed:
Y = o + PlXl + PZX, + P;u1X** (1)
where: Y is the auditors’ response in acceding
to client’s request; X1 is ethical reasoning; Xa is
the locus of control; X,X, is the interaction
term; and H1
s # 0.
According to Southwood (1978, p. 1155) the
interaction term is expressed as the cross-pro
duct “. . . as occurring between independent
variables in their effect on the dependent vari-
able.” If the fi for the interaction term does not
” X1 = -b/P3
= -5.04/-0.10
= 50.4.
equal 0, this implies that the interaction is sig-
nificant. The interaction term in itself does not
tell us much about the nature and form of the
interaction. To obtain this information, the par-
tial derivative of equation (1) needs to be exam-
ined (Schoonhoven, 1981; Govindarajan &
Fisher, 1990; Gul & Chia, 1994). Table 1 pre-
sents the descriptive statistics for the variables
in the model.
RESULTS AND DISCUSSION
The results reported in Table 2 showed that
Hi:j33 # 0, may be accepted. Since there is a
significant interaction between ethical reason-
ing and locus of control on the dependent vari-
able, the analysis was pushed to the second
stage by studying the partial derivative of equa-
tion (1). The following partial derivative equa-
tion depicts this relationship:
GY/sx, = p* + &Xi.
The point of inflection calculated to be
50.411 is within the range of scores obtained
TABLE 1. Descriptive statistics
Standard
Variables Mean deviation
RaQw
Ethical reasoning (RR) XI 29.72 13.90 2.5-67.5
Locus of control (lx) x, 10.10 3.98 2.0-18.0
Auditor’s response (AR) Y 24.06 23.62 0.0-95.0
TABLE 2. Results of multiple regression analysis
Variables
Ethical reasoning @R) X1 0.7P
(0.45)
Locus of control (LC) X, 5.04”
This study investigates the interaction effects of locus of control, a personality variable, and ethical
reasoning on the behaviour of auditors in an audit conflict situation. Eighty experienced auditors from a
sample of Big Six and Non-Big Six CPA firms in Hong Kong were provided with a case study involving an
audit conflict situation and were asked to state the extent to which they would accede to the client’s
request. Subjects were also administered Rotter’s Locus of Control Scale and the Defining Issues Test
Pergamon Accountfug, Organfzations and Sockty, Vol. 21, No. 1. PP. 41-51, 1996
Copyright 0 1995 Elsevier Science Ltd
Printed in Great Britain. AU rights reserved.
0361-3682/% $15.00+0.00
03613682(95)00009_7
AUDITORS’ BEHAVIOUR IN AN AUDIT CONFLICT SITUATION: A RESEARCH
NOTE ON THE ROLE OF LOCUS OF CONTROL AND ETHICAL REASONING*
JUDY S. L. TSUI
City University of Hong Kong
and
FERDINAND A. GUL
The Chinese University of Hong Kong
Abstract
This study investigates the interaction effects of locus of control, a personality variable, and ethical
reasoning on the behaviour of auditors in an audit conflict situation. Eighty experienced auditors from a
sample of Big Six and Non-Big Six CPA firms in Hong Kong were provided with a case study involving an
audit conflict situation and were asked to state the extent to which they would accede to the client’s
request. Subjects were also administered Rotter’s Locus of Control Scale and the Defining Issues Test
(DITJ to measure ethical reasoning. Analyses of the data using multiple regression found that ethical
reasoning moderated the relationship between locus of control and the auditors’ responses to accede to
client’s request in an audit contlict situation. An implication of these results is that the explicit
recognition of both locus of control and ethical reasoning provides a better explanation for differences
in auditors’ ethical decision making
The issue of ethical behaviour of accountants
has recently been a source of much concern
and research interest (Cohen et al., 1993). A
discernable strand of thls research interest
attempts to unravel and evaluate the factors
that could affect ethical behaviour of accoun-
tants (Shaub, 1989; Ponemon, 1990; Ponemon
& Gabhart, 1993). These studies emphasized
the notion that the cognitive processes of indi-
viduals should not be treated as a “black box”,
but should be explicitly recognized since these
processes are expected to affect individuals’
ethical decision making (Ponemon, 1990;
1992a; 1992b; Ponemon & Gabhart, 1990).
More specifically, these studies “incorporate
or expressly consider a psychological frame-
work for the ethical reasoning process” in
order to answer the question what makes
accountants more or less ethical? (Jones &
Ponemon, 1993, p. 411).
The psychological framework for ethical rea-
soning which is based on Kohlberg’s (1969)
Theory of Cognitive Moral Development uses
the Defining Issues Test (Rest, 1979) to mea-
sure an individual’s level of ethical reasoning.
Using the P scores, ’ individuals are classified
l The authors would like to acknowledge the comments of Mark Young and Chee Chow, two anonymous reviewers and
the editor. The research instruments used in this study are available from the authors.
1 The P score measures the relative importance of stages five and six of ethical reasoning. Ponemon (1992b, p. 182)
suggests that “the higher P score implies a lower percentage of stages 1 through 4.”
41
42 J. S. L. TSUI and F. A. GUL
into different levels of ethical reasoning. Based
on this theory, several studies have found sup
port for the notion that the levels of ethical
reasoning are related to auditors’ professional
judgments (Ponemon, 1990; 1992a; 1992b;
Ponemon & Gabhart, 1990; Tsui, 1994). For
example, Ponemon (1992b) found that those
auditors with lower levels of ethical reasoning
tended to underreport their chargeable time
more severely than those with higher levels
of ethical reasoning.
Unfortunately, consideration of the cognitive
process by way of the ethical reasoning pro-
cess, by itself, is an inadequate characteriza-
tion of the “black box” since it fails to
recognize the role of other “individual” differ-
ences, such as personality which may be
defined as individuals’ attitudes and beliefs
(Pratt, 1980; Gul, 1984). Trevino (1986, p.
602) recognized this and argued that:
the individual’s cognitive moral development stage
determines how an individual thinks about ethical
dilemmas, his or her process of deciding what is right
or wrong in a situation. However, cognitions of right
and wrong are not enough to explain or predict ethical
decision-making behaviour. Additional individual and
situational variables interact with the cognitive compo
nent to determine how an individual is likely to behave
in response to an ethical dilemma.
Trevino (1986, p. 602) went on to suggest that
the individual differences variable, or speciti-
tally such personality variables as locus of con-
trol (Rotter, 1966), could interact with ethical
reasoning to affect individual behaviour in an
ethical dilemma. Locus of control is defined as
(MacDonald, 1976, p. 169):
the extent to which persons perceive contingency
relationships between their actions and their outcomes.
Persons who believe that they have some control over
their destinies are called ‘internals’; that is, they believe
that at least some control resides within themselves.
Xxtemafs’, on the other hand, believe that their out-
comes are determined by agents or factors extrinsic to
themselves, for example, by fate, luck, chance, power-
ful others or the unpredictable.
Trevino (1986) suggested that since exter-
nals rely on fate, luck and chance, they are
more likely to behave unethically than inter-
nals who are able to rationalize and control
their behaviour. Similarly, internals who
believe that events are under their control
would take responsibility in their determina-
tion of right and wrong, thus behaving more
ethically than externals.
Rotter (1966) argued that the intemal-
external construct is a unidimensional person-
ality variable that is stable (Ashlcanasy, 1985, p.
1328). Thus, personality traits such as locus of
control are relatively more stable in an indivi-
dual (Roford & Penno, 1992, p. 128) than ethi-
cal reasoning levels since these levels may be
improved through ethical intervention and
formal education.
Thus, a more fruitful approach to the study
of how auditors make ethical decisions should
explicitly recognize that ethical reasoning
levels and locus of control would jointly influ-
ence ethical decision making (Trevino, 1986).
Such a model (Fig. 1) is likely to significantly
improve attempts at not only predicting and
understanding ethical behaviour, but also
developing ethical intervention strategies. The
present study which extends previous account-
ing ethics literature explores and tests the
notion that the effects of locus of control on
auditors’ decision making depend on an indivi-
dual’s level of ethical reasoning.
The particular decision-making task that is
selected for study is the auditors’ ability to
resist management pressure in an audit contlict
situation. Two reasons motivated this decision.
First, client pressure in an audit conflict situa-
tion is prevalent in the audit environment. Evi-
dence regarding the prevalence of
management pressure was, for example, pro-
vided by Finn et al . (1988) who found that
Ethical Reasoning
Locus of Control
*
b Auditor’s Response ID
Accede IO Client’s Request
Fig. 1
AUDITORS’ BEHAVIOUR IN AN
CPAs identified client pressure to alter tax
reports and con&t of interest/ Independence
situations as posing the “most difficult ethical
situations” (Glory et aZ., 1992, p. 286). Second,
there is no systematic empirical evidence link-
ing locus of control and ethical reasoning to
auditors’ behaviour in confIict situations.*
Audit conflict is thus a source of much con-
cern and research interest (Knapp, 1985; Gul,
1991). In particular, the problem regarding the
auditors’ ability to resist management pressure
in audit conflict situations was identified as
early as 1938 (Mckesson & Robbins, 1941)
and more recently by the Cohen Commission
(AICPA, 1978). Arising from this, prior studies
have concentrated on the effects of different
contextual variables, such as client financial
condition, size of audit fees and management
advisory services, on the auditors’ ability to
resist management pressures in audit conflict
situations (Knapp, 1985, 1987; GuI, 1991; Gul
& Tsui, 1992). No study has examined the ques-
tion of whether locus of control and ethical
reasoning will jointly affect auditors’ beha-
viour in an audit conflict situation.
In addition, the theoretical linkages between
personality, ethical reasoning and ethical beha-
viour need to be empirically demonstrated
before the accounting profession and educa-
tors attempt to undertake ethical interventions
to promote more ethical professional beha-
viour since such interventions may be more
effective for individuals with particular person-
ality traits. Evidence on these issues will pro-
vide a better understanding of the ethical
behaviour of accountants. In summary, this
study adds to the literature on auditors’ ethical
behaviour in emphasizing the “processes”3
involved in ethical decision making by consid-
ering two issues. First, it takes into account the
AUDIT CONFLICT SITUATION 43
joint influence of ethical reasoning and lndivi-
dual differences on auditors’ ethical behaviour,
and second, it sheds some light on the beha-
viour of auditors ln an audit conflict situation.
HYPOTHESIS DEVELOPMENT
The research value and endurance of locus of
control as a significant predictor of human
behaviour in a wide variety of situations is
welI documented (Jennings & Zeithmal, 1983;
Findley & Cooper, 1983). In accounting litera-
ture, it has been studied in different decision
contexts such as, for example, in budgetary
participation (Brownell, 1982; Frucot &
Shearon, 1991). Based on locus of control the-
ory, it is likely that auditors’ behaviour in an
audit conflict situation will be infhtenced by
whether he/ she is an internal or an external.
For example, since an “external” believes
that outcomes are beyond control and can be
attributed to fate, luck or destiny, he/ she is less
likely “to take personal responsibility for the
consequences of ethical/ unethical behaviour
and is more likely to rely on external forces”
(Trevino, 1986, p. 601). Thus, individuals who
are externals are more likely to accede to a cli-
ent’s request in an audit conflict situation. On
the other hand, since an “internal” is more
likely “to take responsibility for consequences
and rely on his or her internal determination of
right and wrong to guide behaviour” (Trevino,
1986, p. 601) he/ she is less likely to accede to
client pressure in an audit conflict situation.
This relationship between locus of control
and auditors’ behaviour, however, may be
dependent on the auditors’ ethical reasoning
since this cognitive variable has been found
to affect the behaviour of accountants in a
* Conflict theory (Thomas, 1992) suggests that auditor’s con&t with the client can pose an ethical dilemma for the
auditor.
3 Noreen (1988, p. 359) in the context of agency theory, concluded that “. an agreement (i.e. ethical code) to abstain
from opporhinistic behaviour cannot be enforced effectively by external rewards or sanctions; instead, the sanctions for
unethical behaviour must be internaked.” Flory el al. (1992, p, 288) pointed out that “the intemahzation of these
sanctions calls for better understanding of ethical behaviour.”
44 J. S. L. TSLJI and F. A. GUL
variety of ethical decision-making situations.
For example, Ponemon & Gabhart (1990) in
developing an experimental case to examine
auditors’ ethical judgment in an independence
context, found that auditors at lower levels of
moral cognitive development are more sensi-
tive to factors relating to penalty than affilia-
tion in making independence judgments.
Ponemon’s (1992b) study also found that audi-
tor underreporting is signiIicantly related to the
level of ethical reasoning. Those auditors with
lower DIT scores underreported their charge-
able time much more severely than those with
higher DIT scores. The evidence suggests that
individuals with higher DIT scores are more
likely to resist management pressure in an
audit con&t situation.
Thus, following Trevino’s (1986) suggestion,
it is likely that locus of control and levels of
ethical reasoning are important components
of the “black box” and would jointly influence
individuals’ behaviour in an audit conflict situa-
tion. Locus of control is treated as the primary
independent variable with ethical reasoning as
a moderating variable since, as pointed out
earlier, the literature suggests that locus of con-
trol is a relatively more stable characteristic
(Ashkanasy, 1985; Koford & Penno, 1992)
than ethical reasoning levels which could be
improved by education and other forms of
intervention (Ponemon & Gabhart, 1993).
Based on this reasoning, for example, an exter-
nal who has low DIT scores is more likely to
accede to a client’s request in an audit conflict
situation than an external who has high DIT
scores. This reasoning and evidence provided
above suggest the following hypothesis:
Hl: The effects of locus of control on the extent to
which an auditor would respond as acceding to client’s
request in an audit conflict situation wIII depend on the
level of ethical reasoning.
METHODS
The research instrument was administered in
the premises where the subjects work. This
ensures an adequate response rate in which
subjects were requested to complete the
research instrument in the presence of the
researcher. It is also an advantage for the
researcher to be able to observe important
behaviours such as the conscientiousness of
the subjects and the time taken in completing
the research instrument. A pilot test of the
research instruments with 11 academics who
are professional accountants and had prior
auditing experience as well as 10 experienced
auditors was conducted prior to administering
the questionnaires.
Subj ects
Si nce the subjects in this study are required
to make a decision on an auditor-client conflict
situation, auditors with professional qualifica-
tion and at least four years of audit experi-
ence* were selected as respondents. Qualified
subjects were selected by contact managers/
partners in each of their firms to participate
in this study. The subjects came from four Big
Six and five Non-Big Six CPA tirms in Hong
Kong. A sample size of 80 senior experienced
auditors was obtained for analysis (18 subjects
were dropped from the analysis due to failure
to satisfy the consistency check in the DIT).
The mean age was 31.56 years (range from
26-47) and mean auditing experience was 8.4
years (range from 4-28). Of the 80 respon-
dents, 51 came from the “Big Six” and ranged
in positions from partners (14), managers (46)
supervisors (16) to seniors (4).
Measurement of vari abl es
Locus of control was measured using Rotter’s
Social Reaction Inventory (1966),5 a 23 item
’ The contact partners/managers of CPA tirms who participated in this study confirmed that auditors with four years of
audit experience in Hong Kong would normally handIe auditor-cIient con%? situations.
5 See Rotter (1966) for test-retest reIiabiIity, convergent validity, discrhninant vaklity and other internal consistency tests.
AUDITORS’ BEHAVIOUR IN AN
forced-choice scale. It consists of 23 paired
statements. Each pair statement is character-
ized by an internal statement and an external
statement. One point is awarded for each exter-
nal statement. Scores can range from 0 to 23.
Low scores on the scale were classified as inter-
nals while those with high scores were classi-
fied as externals.
The widely used Defining Issues Test6 (Rest,
1979) was employed to measure the subjects’
cognitive moral development levels.’ Even
though the P score from the DIT is not exactly
equivalent to Kohlberg’s Moral Judgment Inter-
view (Colby & Kohlberg, 1987) it is consid-
ered an appropriate surrogate to measure the
relative importance on stages five and six ethi-
cal reasoning.
The DIT measure employed in this study
requires participants to analyze four hypotheti-
cal moral dilemmas.’ For each dilemma, the
participants have to rank order, from a list of
twelve items, four that are most significant in
influencing the resolution of each dilemma.
From the most important ranked question to
the least important one, a weighting of the
scale from 4 points being the most important
to 1 as least important is given. The P score is
calculated from the sum of the weighted total
of the four most important ranked items. It is
expressed as a continuous variable from 0 to
100%.
The DIT scoring procedure includes an M
L
AUDIT CONFLICT SITUATION 45
(meaningless) score which acts as a consis-
tency and social desirability check on the
data. The M score indicates the extent to
which the subjects select items that are lofty
sounding but meaningless. Rest (1993, p. 13)
suggested that data obtained from participants
with an M score equal to or greater than eight
for the six-story version should be discarded
from analysis. Since a four-story version of the
DIT is used, an M score of five would propor-
tionately be the cutoff point for elimination
from the analyses. The data were also checked
for consistency and those who failed the test (a
total of 18 subjects) were also eliminated from
the sample for purposes of analysis.
The extent to which the auditor was willing
to accede to client pressure in an auditor-client
conflict situation was measured by using an
adapted version of a short case developed by
Knapp (1985) and used by GUI (1991) (see
Appendix). In the case,’ the auditor disagrees
with management that certain unrecorded
liabilities are immaterial and should not be
recorded in the financial statements. The sub
ject was requested to role play as the auditor of
the company and indicate the likelihood that
he/ she will ignore the unrecorded liabilities.
A reasonable inference can then be made
regarding the extent to which the auditor is
willing to resist client pressure in an audit con-
flict situation and it is therefore an indication of
unethical behaviour. lo
” See Rest (1979) for validation of the DIT in terms of predictive validity, face validity, criterion group validity, longitudinal
change studies, convergent-divergent validity, experimental enhancement studies and discriminant validity For test-retest
reliability and Alpha Cronbach, see Rest (1986).
’ The Moral Judgment Interview developed by Colby & Kohlberg (1987) is also recognized as one of the very reliable and
valid instruments for measuring cognitive moral development. However, the DlT is adopted because it is an objective
recognition instrument which is considered more scientihc, valid and reliable.
* The original DIT has six hypothetical moral dilemmas. However, a four-story version is used in this study because the
other two stories are considered too specific for Western cultures (Ma, 1988). The subjects were administered the English
version of the DIT even though their tirst language (ii most cases) would not be English. It is considered appropriate in
these circumstances because these subjects are qualified accountants and are well versed in English.
9 Preliminary discussions with experienced auditors on the conduct of the study indicated that a short case was preferable
in view of the severe constraints on auditors’ time.
i0 Knapp (1985) argued that an inference can also be made regarding auditor independence levels.
46 J. S. L. TSUI and F. A. GUL
A continuous scale from 0 to 100% was used
to measure the likelihood that the auditor wiii
ignore the unrecorded liabilities as recom-
mended by management. A lower Rkellhood
that the auditor will ignore the unrecorded
liabilities is indicative of the extent to which
the auditor will accede to a client’s request
and therefore it is suggested that through the
auditor’s response, one can infer that he/she is
more/less independent. The higher the iikeli-
hood that the auditor will ignore the unrec-
orded liabilities, the greater is the extent to
which the auditor will accede to the client’s
request. Such a decision may reflect the extent
of the auditor’s independence. This interpreta-
tion is consistent with Ponemon’s (1992b, p.
176) viewpoint that the behaviour of the audi-
tor underreporting of time reflects “the ethical
values that the individual brings to the task.” In
the present study, the auditor’s response in
acceding to the client’s request should similarly
reflect some aspects of an individual’s beliefs
on independence and ethical inclinations.
As part of the debriefing questions, respon-
dents were asked whether they bad encoun-
tered similar conflict cases and to describe
the most memorable one. Most of the respon-
dents indicated that they had encountered
similar cases before in their auditing careers.
Data analysis
Multiple regression was used to analyze the
data (Pedhazur, 1982). To test for twoway
interaction effects, the foilowing multiplicative
model is employed:
Y = o + PlXl + PZX, + P;u1X** (1)
where: Y is the auditors’ response in acceding
to client’s request; X1 is ethical reasoning; Xa is
the locus of control; X,X, is the interaction
term; and H1
s # 0.According to Southwood (1978, p. 1155) the
interaction term is expressed as the cross-pro
duct “. . . as occurring between independent
variables in their effect on the dependent vari-
able.” If the fi for the interaction term does not
” X1 = -b/P3
= -5.04/-0.10
= 50.4.
equal 0, this implies that the interaction is sig-
nificant. The interaction term in itself does not
tell us much about the nature and form of the
interaction. To obtain this information, the par-
tial derivative of equation (1) needs to be exam-
ined (Schoonhoven, 1981; Govindarajan &
Fisher, 1990; Gul & Chia, 1994). Table 1 pre-
sents the descriptive statistics for the variables
in the model.
RESULTS AND DISCUSSION
The results reported in Table 2 showed that
Hi:j33 # 0, may be accepted. Since there is a
significant interaction between ethical reason-
ing and locus of control on the dependent vari-
able, the analysis was pushed to the second
stage by studying the partial derivative of equa-
tion (1). The following partial derivative equa-
tion depicts this relationship:
GY/sx, = p* + &Xi.
The point of inflection calculated to be
50.411 is within the range of scores obtained
TABLE 1. Descriptive statistics
Standard
Variables Mean deviation
RaQw
Ethical reasoning (RR) XI 29.72 13.90 2.5-67.5
Locus of control (lx) x, 10.10 3.98 2.0-18.0
Auditor’s response (AR) Y 24.06 23.62 0.0-95.0
TABLE 2. Results of multiple regression analysis
Variables
Ethical reasoning @R) X1 0.7P
(0.45)
Locus of control (LC) X, 5.04”