|
The
Impact of Risk and Benefit on Unethical Behavior
Jennifer
Huen
University
of California, Los Angeles
Abstract
To measure the impact of risk and benefit on the
self-reported likelihood of engaging in immoral behavior, participants
read four hypothetical scenarios and rated how likely three targets (i.e.,
self, friend, stranger) would be to commit an unethical act under high or
low levels of risk and benefit. Each
scenario was analyzed separately with four 2 x 2 MANOVAS (risk x benefit).
Under low risk conditions, participants reported that the friend
and stranger targets would be more likely than themselves to commit an
unethical act, which supports the self-enhancement bias theory.
However, benefits did not increase the likelihood to cheat.
The findings suggest that situational variables such as risk and
benefit may not govern moral decision-making.
The
Impact of Risk and Benefit on Unethical Behavior
Imagine
that a grocery store clerk accidentally gives you an extra five-dollar
bill because it was stuck to the rest of your change, but you do not
realize this until you step outside and count your money.
As soon as you decide to go back inside the market to return the
excess cash, a homeless person comes up to you and asks for money. Now, stop for one second and think, “Would you give this
person that extra change?”
Right
now, you probably feel good about your decision, regardless of whether you
gave money to the homeless because people generally have a positive
self-concept, and are motivated to maintain a high sense of self-worth. One way that people can maintain their self-worth is by
living up to their own moral standards (Bersoff, 1999).
However, if there are individuals who try to maintain high moral
standards, then why would anyone ever commit unethical acts, risk being
caught, and ultimately compromise their image of themselves as good,
honest individuals? This
raises the interesting question of, “What influences a person’s
decision to engage in immoral behavior?”
Oftentimes, situational factors (e.g., risk of detection,
sanctions) can decrease or increase dishonesty (Covey, Saladin, and
Killen, 1989). However,
research has yet to address the relationship between immoral behavior and
two situational variables, risk and perceived benefits.
In one study, Covey et al. (1989) found that
dishonest behaviors are influenced by situational factors (i.e.,
surveillance and incentives) and personality factors (i.e.,
self-monitoring). Covey et
al. defined high self-monitors as people who adjust their presentations of
self according to the demands of a social situation; low self-monitors are
less concerned with their self-image and find it unnecessary to change the
way they present themselves. Covey
et al. predicted that in the absence of incentives, both high and low
self-monitors will not cheat because high self-monitors are too concerned
about their self-image, and low self-monitors do not have the need to
cheat. However, since low
self-monitors are unconcerned about their self-image, they would be more
willing to risk detection and cheat when incentives are provided.
The results showed that surveillance reduced cheating and that low
self-monitors’ lack of concern for their self-image interacted with
incentives to increase dishonesty. Since this study demonstrated that self-monitoring interacts
with incentives to reduce cheating, we decided to investigate another
variable that might encourage or discourage unethical behavior.
Hill and
Kochendorfer (as cited in Leming, 1980) found that cheating is less likely
to occur when participants sense the threat of detection.
Consistent with this reasoning, Leming (1980) showed that cheating
was situationally specific: participants cheated more under low-risk
conditions than under high-risk conditions.
There were also gender effects, such that sanction threats in
high-risk conditions reduced cheating only for women.
Although academic ability did not determine cheating behavior, high
ability students cheated significantly less in the high-risk condition
than in the low-risk condition.
The studies of Covey et al. (1989) and Leming (1980) suggest that risk
and incentives each reduced the likelihood of cheating in academic
settings. To examine whether
risk and incentives would have the same effect in other decision-making
contexts, we manipulated the risk of detection to see if high risk would
decrease the self-reported likelihood of committing unethical acts (e.g.,
keeping cash from a lost wallet, cheating on a partner).
Then, we wanted to see if the type of incentive would interact with
low risk situations to increase the reported likelihood of cheating.
Based on Leming’s findings (1980), we hypothesized that
participants would be less dishonest under high-risk conditions if the
potential benefits were low. We
believed that high-risk conditions would be a situational constraint on
commission of unethical acts, regardless of the type of benefit
involved. Hence, we predicted
an overall main effect for risk level, such that high risk would reduce
cheating whereas low risk would increase the likelihood to cheat or do
something unethical. We also
expected a main effect for benefit such that high benefits would encourage
cheating and low benefits would discourage cheating.
Finally, we predicted a significant interaction such that when risk
of detection is high, only high benefits would increase cheating.
Under low risk conditions, cheating may occur regardless of the
type of benefits.
The decision to commit unethical acts is also dependent upon how
individuals view themselves in comparison to their friends.
People may justify their unethical behavior by describing
themselves as better than others, or by assuming that others would act in
the same way. Thus, we wanted
to see whether the self-enhancement bias (SEB) or the false consensus
effect (FCE) would better account for participants’ behaviors.
SEB states that individuals tend to think of themselves in more
favorable terms than they think of others, whereas FCE states that
individuals have the tendency to overestimate the commonness of their own
attitudes and behaviors.
Method
Participants
Thirty-two
undergraduate students (18 women and 14 men, mean age = 22 years) from
University of California, Los Angeles volunteered to participate.
Twenty participants were from a social psychology laboratory course
and the remaining participants were non-psychology majors.
Procedure
We conducted five pilot tests, during which none of the
participants successfully identified our hypothesis regarding risk and
benefit. Two psychology
students thought that we were studying the self-enhancement bias, but
since many of the participants would be non-psychology majors, we felt
that random assignment would eliminate any possible bias by psychology
students.
Participants read four hypothetical scenarios (i.e., car, midterm, wallet,
and relationship) under varying levels of risk and benefit (please see
Appendix). We used a Latin
Square to counterbalance the order in which the scenarios appeared and to
determine the level of risk and benefit for each scenario (e.g., one
combination was in the following order 1) high risk-high benefit car
scenario, 2) high risk-low benefit relationship scenario, 3) low risk-high
benefit wallet scenario, and 4) low risk-low benefit midterm scenario).
After reading each scenario, the participants rated how likely
three targets (i.e., self, friend, stranger) would be to commit an
unethical act (please see Appendix).
The three-level dependent variable was measured on a Likert scale
of 1-5 with labeled endpoints (please see Appendix).
In between the questions for each target, a filler item asked,
“Do you know of anyone who has encountered a similar situation?”
At the end of the experiment, participants were debriefed and given
an information sheet.
As a final note, we thought that it was interesting how the filler
question had amounted to useful information.
It allowed us to see whether a scenario was too hypothetical, and
suggested that participants made self-enhancing statements about
themselves.
Results
We
predicted that participants would be more likely to commit an unethical
act in low-risk situations than in high-risk situations, and would be most
likely to commit the unethical act if the benefits were high in the
low-risk condition. We tested
these predictions by analyzing each scenario separately, using a total of
four 2 x 2 MANOVAS (risk x benefit), with the dependent variable as target
type with three levels, self, friend, and stranger.
In the
midterm scenario (please refer to Table 1), there was a significant main
effect for risk when the target was stranger, F(1,
28) = 5.609, p < .05, meaning
that when risk was low (M = 3.94, SD = .21), participants reported that a
stranger would be more likely to cheat than when risk was high (M = 3.25,
SD = .21). When the target
was friend, there was a marginal main effect for risk, F(1.28)
= 3.80, p < .10, such that
when risk was low (M = 3.0, SD = .295), participants rated that the friend
would be more likely to cheat than when the risk was high (M = 2.19, SD =
.295). There were no main
effects for benefit, which showed that the incentive to cheat was not as
influential as the risk of being caught.
There was no interaction between risk and benefit.
In the
wallet scenario, there were no main effects for risk and benefit, and no
interaction. Hence, we
analyzed the frequencies for the filler item (i.e., Do you know of anyone
who has encountered a similar situation?) to examine whether the
commonness of the scenario had influenced participants to not take the
cash. The frequencies showed that more than half of the
participants knew of someone who encountered situations similar to the
midterm, car, and relationship scenarios (more participants answered yes
to the filler item in these scenarios—please see Table 3).
However, in the wallet scenario, fewer participants were familiar
with this type of situation, indicating that it is not as salient as other
situations for undergraduate students.
In the
relationship scenario, there were no main effects for risk and benefit,
and no interaction. These
data, however, should not be interpreted as risk and incentives having no
effect on infidelity. Rather,
these results suggest that the context was perhaps too rigid and needs to
be revised. A more compelling
circumstance (e.g., meeting an attractive person after an argument with
one’s partner) might yield results that are more conclusive.
In the
car scenario (please refer to Table 2), we expected that participants
would be most likely to leave a note for the owner of the car if they left
a large dent (high incentive) in a busy parking lot (high risk).
However, there was a significant main effect for risk when the
target was self, F(1,28) = 1.61,
p < .05, such that
participants rated themselves as more likely to leave a note for the owner
of the car under low risk conditions (M
= 3.19, SD = .31) than under high risk conditions (M
= 2.13, SD = .41). Although
this finding is counter-intuitive to our hypothesis, its implications will
be discussed later. There
were no main effects for benefit and no interaction.
Finally, we looked at the correlations between the self and friend targets
versus the self and stranger targets to see which theory, SEB or the FCE,
would better account for the behavior in this study.
According to SEB, there would be a high correlation between self
and a close friend with a low correlation between self and stranger;
according to FCE, the ratings for self, friend, and stranger would be
significantly correlated with each other. We found that self was significantly correlated with friend (r
= .70, p < .01), which suggests that participants identified themselves
with their close friend. The
fact that self was not significantly correlated with stranger, (r = .22, p = ns), means
that participants did not identify themselves with the stranger, and
demonstrates that the FCE did not account for the behavior in this study.
Thus, participants in this study rated the targets in a manner that
was consistent with what the SEB would predict.
Discussion
The
purpose of this study was to examine how situational variables such as
risk of detection and incentives could affect the likelihood of committing
unethical acts. We predicted
that high-risk conditions would reduce the likelihood to cheat, regardless
of the incentives or benefits. In
low-risk conditions, we speculated that cheating would increase if there
were high benefits involved. Results
for the midterm and car scenarios supported our hypothesis, participants
rated that people would be more likely to cheat in low-risk conditions
than in high-risk conditions. Furthermore,
participants rated that a stranger would be more likely than themselves to
cheat in the low-risk conditions, which reveals that the self-enhancement
bias better accounted for people’s decision making patterns regarding
immoral behavior. Had
the participants rated all three targets as equally likely to cheat, then
the false consensus effect would have accounted for their behavior.
Our results should be interpreted in light of a few limitations.
First, several participants completed the experiment individually,
as opposed to being in a group, which may have reduced the effect of
anonymity in our study. Consequently,
participants may not be as honest as they would be if they had completed
the questionnaire among other students. Our sample size was also too small, but this problem could be
easily corrected in future studies. There
was also some concern that measuring the dependent variable on a Likert
scale would not produce any meaningful variability.
However, these kinds of self-report measures on 5- or 7-point
scales analyzed by ANOVA are the most common dependent measures employed
in studies of this kind. The
significant results in the midterm and car scenarios confirm that the
participants had responded to the IVs according to our hypotheses.
In the
debriefing sessions, a few participants mentioned that they had
interpreted the risk and benefit levels differently from what we had
expected, particularly in the car scenario.
We believed that the presence of bystanders would curtail unethical
behaviors (i.e., hit and run accidents), and thus create a higher risk for
driving away without reporting the accident to the owner.
However, these participants felt that bystanders would be too busy
to notice the accident or would mind their own business, thus reducing the
probability of being apprehended in a busy area.
If we were to replicate this study, we would include a pilot test
for each scenario to ensure that the risk and benefit levels had
comparable effects across all scenarios.
The incentives to report the accident may have also been ambiguous.
Assuming that the self-monitoring effect would drive the high-self
monitors in a random sample to do what is morally right (Covey et al.,
1989), we chose to use the large
dent as a high incentive to report the accident.
However, the results showed that participants would not report the
accident, regardless of the incentive to maintain a positive self-image,
which contradicts Covey et al.’s findings for self-monitoring.
Our findings are important because they suggest that
self-monitoring may not occur in situations where there is a greater sense
of personal risk (i.e., higher damage costs, legal problems) involved.
The fact that there were no significant results for benefit has two
implications; first, incentives may not have the same effect on unethical
behaviors across all situations and second, benefits alone may not be the
only factor that influences immoral decisions.
It is also possible that if the scenarios were too hypothetical or
uncommon, participants would not be able to base their decisions on their
own or others’ experiences. Hence,
participants would be predicting their own behaviors in the hypothetical
situations, and it is shown that people are bad at predicting their own
behaviors (Gilbert, 2002). Self-enhancement
bias may also lead participants to describe themselves more
positively than others would in reality.
Thus, our measurement of moral decisions may not necessarily
reflect participants’ actual behaviors.
Paper-and-pencil attitude measures in general are incapable of
engendering all of the complex psychological reactions that occur when one
is actually in the situation (Bersoff, 1999).
The implications from these results provide valuable insight to
understanding the influence of situational factors on unethical behavior.
Since these scenarios pertain to real life experiences, results may
suggest that not all moral decisions are governed by risk and reward as is
regular decision making. Although
benefits did not appear to affect immoral choices, high risks may serve as
effective deterrents for cheating in numerous situations.
Thus, further research on risk may yield practical advice to anyone
from a professor trying to prevent cheating to a law enforcer trying to
reduce crime. Our findings
are promising because previous studies have not applied the
self-enhancement bias to ethical situations dealing with risks and
rewards. More importantly,
this study has broadened the opportunities for future research regarding
deterrents and motivations for unethical behavior.
References
Bennet,
R., Price, R.D., & Lindskold, S.
(1973). Is the moral revolution an illusion?
Decisions for self and others in situations involving values.
Proceedings of the Annual Convention of the American Psychological
Association, 81, 311-312.
Covey,
M.K., Saladin, S., & Killen, P.J.
(1989). Self monitoring, surveillance, and incentive effects on
cheating. The
Journal of Social Psychology, 129(5),
673-679.
Ebert,
J., & Gilbert, D. (2002). Decisions and revisions: The affective
forecasting of changeable outcomes. Journal
of Personality and Social Psychology, 82,
503-515.
Leming, J.S. (1980).
Cheating behavior, subject variables, and components of the
internal-external scale under high and low risk conditions.
Journal of Educational Research, 74(2), 83-87.
Author
Note
I thank
my research collaborators, Maggy Athanasious, Ryan A. Cabautan, Miwa Hatakeyama,
and Christine T. Vu for their group effort in conducting the study,
especially Maggy for her advice and dedication to making this project so
successful.
My
social psychology laboratory professor, Dr. Shelley E. Taylor, inspired
this research. I want to
thank Dr. Taylor for all her time and effort spent on reviewing this paper
and for contributing her ideas.
Correspondence concerning this article should be addressed to Jennifer
Huen, Department of Psychology, UCLA, Los Angeles, California 90024. E-mail: jenhuen@wildmail.com
Appendix
The four scenarios and questions appeared in the following manner
(the conditions are indicated in parentheses and were not included as part
of the text). Every
participant was exposed to one of each condition in a counterbalanced
order.
WALLET SCENARIO
You are at your friend's apartment building and you are on your way
out. You see a wallet lying
on the floor outside someone's front door.
You pick it up and see $100
(high benefit); $10 (low
benefit) inside the wallet. You notice there is a man about two doors down, painting the walls in
the hallway (high risk); the
hallway is empty (low risk).
How likely do you think (you, close friend, a stranger) would be to take the cash?
1 2 3 4 5
least likely
most likely
MIDTERM SCENARIO
You are taking a midterm for one of your required classes. You
are doing badly in this class and have a chance of failing (high
incentive); You are doing okay in
this class but if you do well on this exam, you are most likely to get the
final grade that you want (low incentive). You have a clear view of
another person's exam and you know that this person had set the curve on
the last exam. There are several proctors walking around, keeping an eye on the
students (high risk); there is
one proctor reading a book at the front of the room (low risk).
How likely do you think (you, a close friend, a stranger) would be to look at this person’s
exam?
CAR SCENARIO
You are
at a busy (high risk); empty (low risk) parking lot and you accidentally hit another parked
car as you back out. You
notice that you left a minor scratch
(low incentive to leave a note for the owner);
large dent (high incentive to leave a note for the owner).
How likely do you think (you,
a close friend, a stranger) would be to leave a note with your
name and contact information for the owner of this car?
RELATIONSHIP SCENARIO
You have been in a loving relationship for about 6 months, but
a person whom you were interested in for a long time finally calls you for
a date (high benefit); and an
attractive person that you recently met at a friend’s party calls you
for a date (low benefit). The person you are currently dating and the
person who asked you out have mutual friends (high risk); do not have mutual friends (low risk).
How likely do you think (you,
a close friend, a stranger) would be to go out with this person?
Table 1
Mean Ratings for each Target in Midterm Scenario (N=32)
___________________________________________________________________
Target Person
Self
Friend
Stranger
___________________________________________________________________
Risk Level
High
2.31
2.19a
3.25b
Low
2.65
3.00a
3.94b
____________________________________________________________________
a Significant at
level of p < .10
b Significant at level of p < .05
Table 2
Mean Ratings for each Target in Car Scenario (N=32)
___________________________________________________________________
Target Person
Self
Friend
Stranger
___________________________________________________________________
Risk Level
High
2.13a
2.25
1.94
Low
3.19a
2.83
2.38
____________________________________________________________________
a Significant at
level of p < .05
Table 3
Frequencies
for filler item, “Do you know of anyone who has encountered a similar
situation?” (N=32)
_____________________________________________________________________
Response
Yes
No
Total
_____________________________________________________________________
Scenario
Wallet
18
14
32
Midterm
21
11
32
Car
24
8
32
Relationship
26
6
32
Copyright 2003 by the Undergraduate Psychology Journal.
(Vol. 1, No.2.)
< < back
|
