Professional Documents
Culture Documents
a r t i c l e
i n f o
Keywords:
Affect
Tolerance of ambiguity
Ambiguity tolerance
Ethics
Moral reasoning
Ethical decision making
Auditing
Accounting
PANAS
DIT
Accounting education
a b s t r a c t
This research note extends the research on ethical decision making in accounting by examining the effect of
two variables, affect and tolerance of ambiguity, on ethics-related decisions made in two contexts: both
personal and organizational settings. I conduct an experimental study among undergraduate accounting
students and nd that tolerance of ambiguity signicantly inuences ethical decisions in both personal and
organizational settings. The results also reveal a signicant interactive effect between negative affect and
tolerance of ambiguity.
2009 Elsevier Ltd. All rights reserved.
58
59
2. Methodology
To test these hypotheses I conduct an experiment using 157 upperdivision undergraduate accounting students from a large public
university in the southwest. Undergraduate accounting students are
relevant subjects because this study, while relevant to the general
practice of accounting, is particularly relevant to accounting educators
attempting to incorporate ethics into the university accounting
curriculum. Subjects completed the experiment during class time,
and were told that they were participating in an important study that
would assist the college in designing a curriculum to meet accreditation requirements (which was a truthful statement). No course credit
or nancial incentives were given for participation.
The experiment consisted of a number of brief ethics scenarios,
followed by questions designed to measure personal attributes of the
subjects, namely, tolerance of ambiguity and negative and positive affect.
Subject responses to the ethics scenarios serve as the dependent variables,
while subject personal attributes serve as the independent variables. The
instrument used in the experiment has been provided in Appendix A.
60
Table 3
Univariate ANOVA results.
Std. Dev.
% Responses 3c
1
3
4
5
6
Overall
Source
0.969
0.768
0.988
0.914
1.088
0.972
27.00%
11.70%
30.10%
18.90%
30.60%
23.70%
Std. Dev.
% Responses 3c
1
2
3
4
5
Overall
1.186
0.976
0.85
0.983
1.101
1.052
33.70%
25.40%
11.40%
39.10%
35.20%
28.90%
2.077
1.454
2.036
1.699
2.041
1.861
2.197
1.938
1.44
2.177
2.166
1.983
Notes:
a
See Appendix A for descriptions of each scenario.
b
Subjects provide their answer for each personal scenario using a 5-point scale
ranging from 1I denitely would not agree to do this, to 5I would feel comfortable
doing this. I would denitely do this.
c
This column reports the percentage of subject responses greater than or equal to
three. A response greater than or equal to three on any ethical scenario indicates that a
subject is either unsure of how to respond to the scenario, or that the subject would be
willing to violate ethical norms.
d
Subjects provide their answer for each reporting scenario using a 5-point scale ranging
from 1I denitely would report do this, to 5I denitely would not report this.
Table 2
Cell sizes of between-subjects factors.
Variable
61
N
Low
High
74
81
78
83
76
79
73
80
76
82
75
79
Note:
This table reports the number of subjects in each cell after splitting each variable at its
mean. Low (High) indicates that subject responses were lower (higher) than the
mean response.
dF
Mean sq.
Type III SS
7
1
1
1
1
1
1
1
1
149
157
156
18.351
12235.533
54.606
7.876
14.073
5.265
43.048
3.974
1.331
9.597
1.912
1274.972
5.690
0.821
1.466
0.549
4.486
0.414
0.139
0.071
0.000
0.018
7
1
1
1
1
1
1
1
1
147
155
154
40.124
14294.863
98.945
14.299
60.811
19.006
56.943
34.296
0.105
13.469
2.979
1061.351
7.346
1.062
4.515
1.411
4.228
2.546
0.008
0.366
0.228
0.460
0.036
0.521
0.710
0.006
0.000
0.008
0.305
0.035
0.237
0.042
0.113
0.930
Notes:
Indicates signicance at the 10% level.
Indicates signicance at the 5% level.
Indicates signicance at the 1% level.
Table 4
Estimated marginal means of interaction term.
Panel A: PERSONAL ethical scenarios
(dependent variable: PERSONAL)
TOA
TOA
Low
High
NegAffect
Low
High
8.500 (0.503)
8.633 (0.490)
8.050 (0.517)
10.286 (0.476)
Low
High
NegAffect
Low
High
8.823 (0.604)
9.211 (0.588)
8.864 (0.613)
11.686 (0.563)
62
inuence ethical decision making, but that TOA has a more dominant
inuence. Future research should explore this interaction further.
3.3. Ancillary analysis
In addition to the main analysis, I also examined the inuence of
affective arousal on ethical decision making. In Gaudine and Thorne's
(2001) cognitiveaffective model of ethical decision making, the
authors theorize that both affective valence (positive or negative) and
affective intensity, which they refer to as affective arousal, inuence
ethical decision-making. Gaudine and Thorne's model states that
arousal will interact with positive affect to induce people to act on
their positive intentions and behave more ethically. Gaudine and
Thorne have no hypothesis for how arousal will inuence people
experiencing negative affect, noting that certain types of negative
affect, such as anger, can have different inuences on decision making
than other types of negative affect, such as depression. I examine the
effect of arousal generally.
Because I measure both the positive and negative affective state of
each subject independently, I am able to measure the joint affective
intensity of both types of affect in each subject. To do this, I create a
binary variable, AROUSAL, which I use to compare highly aroused
subjects, those who are experiencing both high positive and high
negative arousal, against low arousal subjects who experience neither
high positive nor negative affect. This analysis eliminates subjects who
scored highly on positive or negative affect but not both, leaving me
with a sample of 40 low arousal and 41 high arousal subjects. Interestingly, the mean responses of high arousal subjects were higher
(indicating a greater chance of violating ethical norms) than those of
low arousal subjects, contrary to Gaudine and Thorne's model. The
effect of arousal was marginally signicant (p = .058) in the personal
scenarios and signicant in the reporting scenarios (p = .039). It is
difcult to interpret these results beyond noting that the relationship
between affect and ethical decision making is complex and may not be
adequately explained by current models.
4. Conclusion
While ethics is clearly an important research area in accounting,
prior research on the connection between moral reasoning and ethical
behavior has yielded mixed results, indicating that we do not yet fully
understand how people make ethical decisions. One explanation for the
mixed results of past research is the existence of omitted (or noncontrolled) correlated variables. In an effort to bridge this gap, this paper
examines two personality attributes, affect and TOA, which prior
literature has indicated may inuence ethical decision making. While
the results of this study are largely inconclusive, they shed light on an
important area of accounting that has not been adequately researched.
Furthermore, the student responses to the ethical scenarios indicate that
a signicant portion of accounting students are either unaware of ethical
norms in the profession, or unsure of whether they would choose to
adhere to those norms when presented with ethical dilemmas.
I do not nd evidence of the relationship between positive affect
and ethical decision making predicted by Gaudine and Thorne (2001),
however, the results of the study indicate that in some circumstances,
negative affect may inuence ethical behavior, but in others it may not.
Future research should further explore the inuence of negative affect
and ethical decision making so that we can identify situations where
negative affect may increase the likelihood of unethical behavior.
I also nd a signicant relationship between tolerance of ambiguity
and ethical decision making. This relationship may explain the cultural
differences in ethical decision making documented in prior literature
(Wong-on-Wing & Lui, 2007). Furthermore, I nd a signicant
interaction between TOA and negative affect that has not yet been
explored in the literature. Generally, all of my results imply that
educators and practitioners may want to consider personality traits
2.
3.
4.
5.
63