ArticlePDF Available

Team goal incentives and individual lying behavior

Authors:

Abstract and Figures

In this article we examine the influence of two goal compensation schemes on lying behavior. Based on the die rolling task of Fischbacher/Föllmi-Heusi (2013), we apply an individual goal incentive scheme and a team goal incentive scheme. In both settings individuals receive a fixed bonus when attaining the goal. We find that under team goal incentives subjects are less inclined to over-report production outputs beyond the amount which is on average necessary for goal attainment. Investigating subjects’ beliefs on their team mates’ behavior under team goal incentives reveals that subjects who either believe that lying is not profitable (i.e., the team goal cannot be reached with a lie) or not absolutely necessary (i.e., there is a good chance that the team goal can also be reached without lying) tend to be honest. We also find that subjects who believe that the team goal has already been reached by their team mates tend to over-report production outputs. Across treatments, women are found to be more honest than men. Subjects’ personality is not associated with reported production outputs. Our work contributes to previous research on how different compensation schemes affect unethical behavior in organizational settings.
Content may be subject to copyright.
WP 17/02
WORKING PAPER SERIES
Team Goal Incentives and Individual Lying Behavior
Julian Conrads, Mischa Ellenberger, Bernd Irlenbusch,
Elia Nora Ohms, Rainer Michael Rilke, and
Gari Walkowitz
March 2017
Economics Group
Team Goal Incentives and Individual
Lying Behavior
Julian Conrads
Mischa Ellenberger
Bernd Irlenbusch
Elia Nora Ohms
Gari Walkowitz
University of Cologne
Rainer Michael Rilke
WHU – Otto Beisheim School of Management
WP 17/02
March 2017
Working Paper 17/02
March 2017
ISSN 2511-1159
WHU - Otto Beisheim School of Management
Economics Group
Burgplatz 2
56179 Vallendar, Germany
Phone: +49 (261) 65 09 - 0
whu@whu.edu
Any opinions expressed here are those of the author(s) and not those of WHU. Re-
search published in this series may include views on policy, but the institute itself
takes no institutional policy positions.
WHU Working Papers often represent preliminary work and are circulated to encou-
rage discussion. Citation of such a paper should account for its provisional charac-
ter. A revised version may be available directly from the author.
Abstract
In this article we examine the inuence of two goal compensation schemes on ly-
ing behavior. Based on the die rolling task of Fischbacher/Föllmi-Heusi (2013), we
apply an individual goal incentive scheme and a team goal incentive scheme. In
both settings individuals receive a xed bonus when attaining the goal. We nd that
under team goal incentives subjects are less inclined to over-report production out-
puts beyond the amount which is on average necessary for goal attainment. Inves-
tigating subjects’ beliefs on their team mates’ behavior under team goal incentives
reveals that subjects who either believe that lying is not protable (i.e., the team
goal cannot be reached with a lie) or not absolutely necessary (i.e., there is a good
chance that the team goal can also be reached without lying) tend to be honest. We
also nd that subjects who believe that the team goal has already been reached by
their team mates tend to over-report production outputs. Across treatments, women
are found to be more honest than men. Subjects’ personality is not associated with
reported production outputs. Our work contributes to previous research on how dif-
ferent compensation schemes aect unethical behavior in organizational settings.
JEL-Classication:
C91, C92, M52
Keywords:
Compensation schemes, Lying, Teams, Goals, Individual dierences, Experiment
Corresponding author:
Gari Walkowitz, gari.walkowitz@uni-koeln.de
Funding: Financial support from the German Science Foundation through grant
‘TP3 Design of Incentive Schemes within Firms: Bonus Systems and Performance
Evaluations’ (sub-project of the DFG-Forschergruppe ‘Design and Behavior’ (FOR
1371)) and through the Leibnitz-Award to Axel Ockenfels is gratefully acknowledged.
Note: This paper has been published in Business Administration Review / Die Be-
triebswirtschaft, 76(1), 2016, 103-123, Special Issue on “Experimental Research
on Governance, Risk, Compliance, and Taxation”. The paper is also published here
with the permission of the publishing house Schaeer Poeschel.
WP 17/02
March 2017
1
Team Goal Incentives and Individual Lying Behavior
[Zielanreize in Teams und individuelles Lügenverhalten]
Julian Conrads, Mischa Ellenberger, Bernd Irlenbusch, Elia Nora Ohms,
Rainer Michael Rilke, Gari Walkowitz
Abstract:
In this article we examine the influence of two goal compensation schemes on lying
behavior. Based on the die rolling task of Fischbacher/Föllmi-Heusi (2013), we apply an
individual goal incentive scheme and a team goal incentive scheme. In both settings
individuals receive a fixed bonus when attaining the goal. We find that under team goal
incentives subjects are less inclined to over-report production outputs beyond the amount
which is on average necessary for goal attainment. Investigating subjects’ beliefs on their
team mates’ behavior under team goal incentives reveals that subjects who either believe that
lying is not profitable (i.e., the team goal cannot be reached with a lie) or not absolutely
necessary (i.e., there is a good chance that the team goal can also be reached without lying)
tend to be honest. We also find that subjects who believe that the team goal has already been
reached by their team mates tend to over-report production outputs. Across treatments,
women are found to be more honest than men. Subjects’ personality is not associated with
reported production outputs. Our work contributes to previous research on how different
compensation schemes affect unethical behavior in organizational settings.
JEL classification: C91, C92, M52
Keywords: Compensation schemes, Lying, Teams, Goals, Individual differences,
Experiment
All authors: University of Cologne, Department of Corporate Development and Business Ethics, Albertus-
Magnus-Platz, 50939 Cologne, Germany. Send correspondence to: gari.walkowitz@uni-koeln.de. Financial
support from the German Science Foundation through grant ‘TP3 Design of Incentive Schemes within Firms:
Bonus Systems and Performance Evaluations’ (sub-project of the DFG-Forschergruppe ‘Design and Behavior’
(FOR 1371)) and through the Leibnitz-Award to Axel Ockenfels is gratefully acknowledged. We want to thank
Hannes Rusch, the editors and two anonymous referees for helpful comments and suggestions.
2
1. Introduction
The application of work or performance goals can lead to significant increases in worker
output (see Goerg/Kube, 2012, who study in a randomized field experiment the link between
goals at work, adherent incentives, and worker performance). Indeed, many organizations
use goal settings for motivational purposes. For instance, in Germany, 87% of the companies
with more than 500 employees apply fixed target agreements.1 Thereby, goals are often
linked to organizational reward systems (Yearta/Maitlis/Briner, 1995). In practice, however,
we also observe several cases where goal incentive schemes are connected to the
misrepresentation of sales reports to ensure the obtainment of bonuses
(Degeorge/Patel/Zeckhauser, 1999; Jensen, 2001). In this context, Jensen (2003) points out
that the use of performance goals linked to monetary rewards may induce people to lie about
their actual performance, measured, for example, in profits, sales or production outputs. This
can result in detrimental repercussions for a company’s profitability because the perception
of the company’s economic situation and ethicality might be distorted if such behavior is
detected and made public.
Several recent experimental studies show how goals or targets tied to monetary rewards
can negatively impact ethical behavior. In a laboratory experiment,
Schweitzer/Ordóñez/Douma (2004) find that people, who get paid according to goals, lie
more about their performance than people who are asked to do their best and only receive a
lump-sum payment. In another experiment, Grover/Hui (2005) demonstrate that people tend
to lie more about their performance when achieving a specific performance level is linked to
obtaining a monetary bonus. Additionally, Cadsby/Song/Tapon (2010) show that a goal
compensation scheme produces significantly more dishonesty than a piece-rate or a
tournament compensation scheme (refer also to Ordóñez/Schweitzer/Galinsky/Bazerman,
2009, for a general discussion on the potential harm of goal setting).
The studies focus on the way in which individual goal incentive schemes affect unethical
behavior. However, in practice people do not work alone and team work is very popular.
Many companies use team incentives because it enhances workers’ performance and workers
associate non-pecuniary benefits with it (Hamilton/Nickerson/Owan, 2003). For instance,
70% of the Fortune 1000 companies in the United States use some form of team incentives
(Ledford/Lawler/Mohrman, 1995). In this context, Hoffman/Rogelberg (1998) identify
several categories of team incentive schemes and one of them entails team goal incentives.
Despite the popularity of goal and team incentives in practice, to the best of our
knowledge current research on lying behavior mainly focuses on the effects of individual
goal incentives but lacks insights on the impact of team goal incentive schemes. Therefore,
1 See „Forschungsbericht Arbeitsmarkt 442: Arbeitsqualität und wirtschaftlicher Erfolg: Längsschnittstudie in
deutschen Betrieben“ of the Federal Ministry of Labour and Social Affairs, released 30.6.2013.
3
the aim of this work is to experimentally investigate the influence of a team goal incentive
scheme on lying about individual performance. In this respect, we look at the impact of a
team goal compensation scheme on reporting individual production outputs compared to an
individual goal incentive scheme. In our experiment, we adapt the die rolling game of
Fischbacher/Föllmi-Heusi (2013), henceforth FFH. Subjects have to privately roll a six-sided
die in order to determine their individual production output which is anonymously reported
to the experimenter. To investigate the effects of team goal incentives, two treatments are
implemented: one team goal incentive scheme for teams consisting of six team members and
one individual goal incentive scheme. In the individual goal incentive scheme, subjects are
given a fixed goal they have to reach by rolling the die in order to obtain a monetary bonus.
In the team goal incentive scheme, subjects are given a fixed goal, which they have to reach
together in order to obtain the monetary bonus. The sum of the team members’ reported
production outputs is compared to the provided team goal. In case the goal is reached, the
team bonus is obtained and distributed equally among the team members.
As our main result, under the team goal incentive scheme we find less over-reporting
beyond the average amount needed for goal attainment. Investigating subjects’ beliefs on
their team mates’ reports under team goal incentives reveals that subjects who either believe
that lying is not profitable (i.e., the team goal cannot be reached by lying) or not absolutely
necessary (i.e., there is a good chance that the team goal can as well be reached by reporting
truthfully) tend to be honest. We also find that subjects who believe that the team goal has
already been reached by their team mates tend to over-report production outputs. The
remainder of this paper is arranged as follows. In the next section 2, we present our
experimental design and procedure. To enhance the understanding of subjects' behavior in
our setup, in section 3 we provide a line of theoretical arguments on expected behaviors in
our treatments, present previous empirical evidence and formulate two hypotheses. In section
4, we present our results. In the fifth and last section, we discuss our findings, reflect upon
practical implications and conclude.
2. Experimental Design and Procedure
In our experiment we apply a short one shot decision-making task after a different
independent experiment (for this procedure, see also FFH). At the stage of our task, subjects
do not receive any feedback on their earnings in the preceding experiment. Two treatments
are designed to test our hypotheses formulated in the next section. The first treatment is
implemented as an individual goal setting task, in the following referred to as the Individual
treatment. The second treatment is a team goal setting with teams consisting of six subjects
who are randomly assigned to the teams at the beginning of the task. In the following, we
will refer to this treatment as the Team treatment.
4
Following the procedure of FFH, in the two treatments the instructions explain that
subjects are rewarded for filling in a short questionnaire for a statistical survey and that
payoffs depend on points pi randomly determined through rolling a fair six-sided die. With a
slight adaption compared to FFH, we use points that are understood as random production
output to investigate in which way different goal compensation schemes affect lying.
Subjects are told that the diced number di determines the points pi of subject i. In further
detail, pi=di if di ϵ {1, 2, 3, 4, 5} while no points are obtained if di=6, i.e., pi=0. Due to the
goal modification, subjects in the Individual treatment are told that they would get an
individual payoff πi of 5€ if they reach 3 points or more, i.e., that πi=5€ if pi≥3. Otherwise,
the payoff πi equaled 0€. Hence, subjects have a fair chance of 50% to reach the goal and
receive 5€. In the Team treatment, a subject i is anonymously and randomly assigned to five
other subjects j, k, l, m, and n. The members of one team roll the die individually without any
interaction. The instructions inform the subjects that the team as a whole would earn
6x5€=30€ if the team members together reached 18 points or more, i.e., that π=30€ if
pi+pj+pk+pl+pm+pn≥18. When reaching the team goal, the 30€ are split equally among all
team members. Thus, asubject i earns an individual payoff πi=5€ if pi+pj+pk+pl+pm+pn≥18.
Otherwise the individual payoff πi equaled 0€. The induced uncertainty on goal attainment
(and attached reward payment) is typical for work practice when compensation depends on a
subgroup’s performance and not solely on the individual’s performance
(Gill/Prowse/Vlassopoulos, 2012).
To examine the impact of different goal incentive schemes on lying and to make
incentive schemes comparable across treatments, we hold the potential reward for each
subject with 5€ constant over both treatments. Moreover, for the Team treatment we want to
ensure that 3 points, as in the Individual treatment, remains the individual reference point,
i.e., if each member of a team on average reported 3 points (or more), the team goal was
reached. Hence, for the Team treatment, we multiply the goal of 3 points from the Individual
treatment with the number of team members leading to a goal of 3x6=18 points.
The instructions tell the subjects to roll the die and to report the rolled number on the
instruction sheet.2 Additionally, the instructions make clear that the subjects could roll the
die several times to make sure it is fair but to only report the first number rolled. Due to the
nature of the experiment’s procedure, the experimenters are neither able to reconstruct
whether the subjects report their first diced number nor whether they honestly report an
actually rolled number. Firstly, this is because the subjects privately sat in separate cubicles,
2 The original instruction sheets for the Individual and the Team treatment were in German. They are available
upon the authors request.
5
secondly, because the experimenters leave the experiment room after handing out the
instructions and dices, and thirdly, because the subjects are allowed to roll the die several
times. Through this procedure it is easy for the subjects to lie about their production output
without risking detection. In line with FFH, in our experiment lying is understood as
“reporting a different number than the one actually rolled on the first roll”, which we assume
to be usually higher. As the true outcome of the individual die rolls is unknown, aggregated
reported production outputs are compared to the distribution that can be expected from
rolling the die truthfully. Differences in lying behavior across treatments are generally
measured by the deviation of the frequencies of the distinct reported production outputs from
the frequencies that can be expected from truthfully reporting and, more specifically, by the
frequency subjects report a production output ≥3 in either treatment. In the Individual
treatment a reported production output of 3 or more guarantees the reward of 5€.
Analogously, in the Team treatment, team members need to – on average – report at least a
production output of 3 points to get the reward.
After reporting their rolled number the subjects are asked to fold the instruction sheet and
wait until the experimenters’ return to the laboratory. The folded instruction sheets are then
collected and the questionnaire is handed out. In both treatments, the questionnaire included
questions about subjects’ age, gender and personality. Personality is measured using a 10-
item version of the Big Five Inventory introduced by Rammstedt/John (2007). Additionally,
the subjects of our Team treatment are asked about their beliefs regarding the reported
production output of each of their team mates. We also ask subjects whether they had
participated in a similar experiment before to control for experience effects. After all
subjects had filled in the questionnaire, the experimenters collected it. The sessions ended
with paying out the money that had been earned by the subjects. The payment consisted of
the possible 5€ and of the amount of money subjects had earned in the respective preceding
experiment including a general participation fee of 2.50€.
The experimental sessions of the Individual treatment were conducted in the experimental
economics laboratories of the University of Cologne and Bonn University from August to
September 2010. The sessions for the Team treatment were conducted in the experimental
economics laboratory at the University of Cologne from May to June 2013. In total, 181
subjects took part in our experimental sessions. The mean age of the subjects was 24.28
years. Our sample included 48.31% female students.
3. Theoretical Considerations and Hypotheses
Before turning to the results, we briefly discuss potential effects of different factors on
subjects’ strategy choice in the Individual and in the Team treatment. Providing a line of
theoretical arguments on expected behaviors and presenting prior empirical evidence, we
6
want to enhance the understanding of subjects' behavior in the setup at hand. We will derive
two behavioral hypotheses which will be tested by our experiment.
From a purely self-interested, rational economic perspective, people solve the question of
lying or not by trading off the potential benefit of lying against the potential cost of being
detected and punished (e.g., Becker, 1968). Based on the fact that detection was ruled out in
our experiment, a selfish subject without lying aversion, i.e., who has no cost for lying, will
always report a production output of ≥3 in the Individual treatment. The subject should be
indifferent regarding reporting 3, 4, or 5 as these production outputs all yield the same
reward. In the Team treatment, it is dominant for a subject without lying costs to report a
production output of 5 because – no matter what the other team members report – this
production output yields the highest probability for obtaining the reward. Based on these
considerations, we should observe no differences across the Individual and the Team
treatment regarding the number of reported production outputs ≥3.
Previous work has shown that lying involves ethical deliberations and a psychological
cost for the liar. The growing literature on lying aversion (e.g., Gneezy, 2005,
Mazar/Amir/Ariely, 2008, Kartik, 2009, Sutter, 2009, Erat/Gneezy, 2011,
Shalvi/Handgraaf/De Dreu, 2011, FFH) suggests that people apparently face a conflict in
situations like our die rolling task: on the one hand they want to pursue their monetary
interest but on the other hand they want to avoid the cost of lying and (appear to) behave
ethically. To handle this dilemma, people often look for ways to behave untruthfully enough
to profit from their unethical actions but still truthful enough to not need to revise their
positive self-image (e.g., Mazar et al., 2008, Gino/Ayal/Ariely, 2013). In line with this
argument, recent studies show that people tend to avoid lying to the maximum extent giving
up the opportunity to reap the greatest possible amount of money even if there appears to be
no risk of getting caught (Shalvi et al., 2011, Conrads/Irlenbusch/Rilke/Walkowitz, 2013,
FFH, Conrads/Irlenbusch/Rilke/Schielke/Walkowitz, 2014).
If subjects are lying averse (as the above literature suggests) different considerations have
to be made. Subjects might only lie when there is an incentive to do so, i.e., when the
expected benefit of the lie outweighs its costs. In the Individual treatment, the probability of
getting the reward with an honest report is 50% (by truthfully reporting 3, 4, or 5 points).
Hence, with 50% probability it is necessary but also profitable for subjects to lie and report a
production output ≥3. In the Team treatment, the distribution of truthfully reported
production outputs has a mean of 6x2.5=15 points. There, the probability of getting a team
total of 18 points (the team goal threshold) if all team members report truthfully drops down
7
to 27.9% (see Table A1 in the Appendix).3 Assuming that all other team members report
truthfully, subjects in the Team treatment should consider lying to be necessary (because the
sum of all die rolls is expected to be smaller than 18 points) and profitable (because the
subject can move the sum to 18 or above by misreporting his/her own roll) in about 21.8% of
all cases. As depicted above, in 27.9% of all cases the team wins honestly, while in 50.3% of
all cases, a single subject’s lying would be insufficient to increase the total amount of
reported production outputs for the team to win the reward because the other team members
reach in total at most only 12 points (see Table A2 in the Appendix). Thus, if subjects were
perfectly informed and lied only when it was necessary and profitable, we could expect to
see less than half of the lying frequency in the Team as compared to the Individual treatment.
The situation changes when subjects believe that their fellow team members do not
necessarily report truthfully. If there is some (small) expectation that the other team
members lie, e.g., by increasing their reports by one point if they roll a production output <3,
a subject’s incentive to lie increases because the probability that he or she can influence the
total outcome increases to 30.5%. Yet, if the expectation becomes that most or all team
mates lie, lying-averse subjects’ incentives to lie decrease again because it will be
unnecessary for them to lie if they expect others to over-report their die throws which in turn
most likely guarantees goal attainment. For example, if subjects believe that all other team
mates over-report by one point (if they do not get 5 points), the probability for necessary and
profitable lying decreases to 21.8%. Likewise, if subjects believe that all others lie to the
maximum extent, i.e., that all team mates report 5 points, subjects do not need to lie because
the team goal will be reached no matter what they report on.
Beyond these very particular examples, the incentive to lie - in the vast majority of
possible distributions of beliefs - is smaller than 50% in the Team treatment.4 In general, in
3 The probabilities applied here were derived from simulations with n=100.000 random draws. Therefore, they
might be slightly imprecise in a few cases.
4 There are 65=7776 distinct five tuples of the other five team members’ reported production outputs. In the Team
treatment a subject has the highest probability that lying is necessary and profitable if he or she believes that the
other team members’ reported production outputs add up to 13 points (since the subject then has to lie in 5/6 of
the cases). If the other five group members reported honestly the production outputs add up to exactly 13 points
in 10.03% of the five-tuples. However, a sum of exactly 13 reported points can also be reached if the sum of true
production outputs is smaller than 13 and lying is involved (assuming that subjects do not under-report). The
proportion of five tuples that result in a true total production output of 13 and smaller amounts to 60.03%. In
these cases a reported production output of exactly 13 can in principle arise – may it be by truthful reporting or by
a combination of lying and truthful reporting of by all five team members lying. Thus, the probability that lying is
necessary and profitable can never be higher than 60.03% * 5/6 = 50.03%. This probability is already very close
to 50%.
8
both treatments the incentives to lie may depend on the cost of lying. Assuming that subjects
have constant cost of lying, i.e., whenever they lie they incur a constant cost irrespective how
large their lie is, and these costs do exceed 1.99, we can show that there exists a Nash
equilibrium in the Team setting where all team members report truthfully. In the Individual
Treatment, a subject would lie only if the cost of lying is smaller than 5 (if they are equal to
5, a subject is indifferent between lying and not lying). In this case the benefit from lying
outweighs the cost for lying. Thus, if the cost for lying is between 0 and 5, a subject lies in
50 % of all cases, i.e., when it does not honestly reach the goal (assuming that the cost for
lying is constant, i.e., independent from how much a subject lies). Hence, in the Team
treatment, there is a broad range of lying costs - between 2 and 5 (as compared to 0 and 5 in
the Individual treatment) - where all team members report truthfully. If the cost of lying is
lower than 1.99, the individual incentive to lie gradually increases in the Team setting, up to
a threshold amount where all subjects lie (see Tables A1 to A3 in the Appendix for a detailed
illustration).
Taken together, the expected return of lying tends to be lower in Team than in the
Individual treatment. Based on these considerations, we expect to detect less lying in the
Team as compared to the Individual treatment. Consequently, we formulate:
Hypothesis 1. Lying decreases under team goal incentives compared to an
individual goal incentive scheme.
Lying can also be considered as an act that (also) benefits others if the liar empathizes with
the beneficiaries or if some sort of connection exists between them
(Loewenstein/Thompson/Bazerman, 1989, Gino/Pierce, 2009, 2010, Gino/Ayal/Ariely,
2009, Erat/Gneezy, 2013, Gino et al. 2013). Wiltermuth (2011) asserts that people may be
more inclined to lie when others benefit from it, even if no connections between the liar and
the beneficiaries exist. Likewise, in their recent study Gino et al. (2013) show that caring for
others’ outcomes encourages people to act dishonestly even if the beneficiary is unknown to
the liar. People can also more easily justify immoral actions to themselves if other people
benefit from it, too. Once others benefit as well, a subject’s motive for lying becomes
ambiguous (Wiltermuth, 2011) and the liar appears to be better able to preserve a positive
self-image (Gino et al., 2013). This effect is shown to mitigate the extent to which people
perceive their dishonesty to be immoral (and therefore might reduce their cost of lying) since
if only the liars themselves benefit from their dishonesty it clearly appears to be self-serving.
The above literature points out that the justification motive together with the care motive
apparently have the greatest affect in fostering individuals’ inclination to lie. It also shows
that people tend to lie more as the number of beneficiaries from the lie increases.
9
In a team setting, subjects also have the opportunity to hide their dishonesty within the
team as recently highlighted by Conrads et al. (2013). With this opportunity to diffuse one’s
own responsibility for a lie, a team goal setting potentially disguises an individual’s
contribution and leads to a reduced risk of being identified as a liar and therefore held
accountable (see also Bandura/Underwood/Fromson, 1975,
Bandura/Barbaranelli/Caprara/Pastorelli, 1996).
Taken together, the above evidence suggests that people under a team goal incentive
scheme can be expected to lie more than in an individual goal setting. In the Team treatment
there are other subjects who can potentially benefit from a lie which helps to preserve a
positive self-image and, likewise, other people are let down if the team reward is potentially
not obtained due to having refrained from lying. Moreover, in the Team treatment people
may be more inclined to lie because it is more difficult to identify them as liars within a
group of people. Consequently, we formulate our second hypothesis:
Hypothesis 2. Lying increases under team goal incentives compared to an individual
goal incentive scheme.
To asses which of the two hypotheses can be supported we will now turn to our data.
4. Results
Figure 1 and Table 1 display the results for the distributions of the reported numbers,
converted into production outputs with 6 equaling 0 in the two treatments.5 For better
readability, from now on we will refer to reported production outputs 0 to 5. Experienced
subjects were excluded from the analyses.
4.1. Reported production outputs
To start with, we compare average reported production outputs across our two treatments.
We find no evidence that average reported production outputs significantly differ across the
Individual (3.55) and the Team (3.39) treatment (p=.508, Fisher-Pitman permutation test for
two independent samples6). Similarly, the distributions of reported production outputs do not
significantly differ across treatments (p=.476, Kolmogorov-Smirnov test). In both treatments
reported production outputs are significantly different from the average outcome of 2.5
points one can expect if subjects report truthfully (both p<.01, binomial test), which indicates
that in both treatments subjects over-reported their rolled numbers. If we assume that the
difference between the relative frequency that can be expected from a truthfully conducted
5 A part of our data was collected for a master thesis project. They are presented in Ellenberger/Ohms (2013).
6 In the following denoted as FPPT. All statistical tests are carried out two-sided if not denoted otherwise.
10
die roll (16.67%) and the observed relative frequency of distinct reported production outputs
displays subjects’ dishonesty, we can estimate for each treatment the approximate amount of
subjects who probably lied. In the Individual treatment, 32.46% of the subjects deviate from
the benchmark distribution whereas in the Team treatment they amount to 20.15%.
Figure 1. Relative frequencies of reported production outputs in the Individual and in the Team treatment.
The solid line marks the frequency of reported production outputs (16.67%) according to a uniform distribution.
The dashed line marks the induced goal threshold from an (average) individual perspective.
A set of binomial tests reveals that in both treatments the distribution of reported
production outputs significantly differs from a uniform distribution and that there are
differences across treatments in this regard. In the Individual treatment the production
outputs 4 and 5 are reported significantly more frequently than 16.7% while, on the other
hand, the production outputs 0, 1, and 2 are reported significantly less frequently than
expectable. Interestingly, the relative frequency of a reported production output of 3 is not
found to be statistically different from 16.7% in both treatments even though this reported
production output would (on average) already ensure goal attainment. With 35.09% the
reported production output of 4 is reported most frequently in the Individual treatment while
with 3.51% the reported production output of 2 is reported least frequently.
In the Team treatment the reported production outputs 4 and 5 are also reported
significantly more frequently than the expected 16.7% while the reported production output
of 0 is reported significantly less frequently than expected. However, and partly contrary to
the Individual treatment, the relative frequencies of the reported production outputs 1, 2, and
3 are not found to be statistically different from 16.7%.
020 40
0 1 2 3 4 5 0 1 2 3 4 5
Individual Team
Frequency in %
Reported Production Output
11
Table 1. Descriptive statistics on reported production outputs
Treatment Obs. AV pi≥3 in % Production output pi (relative frequency in %)
0 1 2 3 4 5
Individual 114 3.55 82.46 6.14– – – 7.89– – – 3.51– – – 18.42 35.09+++ 28.95+++
Team 67 3.39 70.15 5.97– – – 10.45 13.43 10.45 28.36++ 31.34+++
AV is the average reported production output. Plus and minus signs display the significance of a one-
sided binomial test indicating that the observed frequency is smaller (larger) than 16.67% ((+)=10%-
level, – –(+ +)=5%-level, – – –(+ + +)=1%-level).
Looking at the distributions of reported production output across treatments conveys that in
the Individual treatment subjects tend to under-report production outputs <3 implying that
they over-report production outputs ≥3. The same tendency holds for the Team treatment.
Though, in the Team treatment subjects under-report more frequently compared to the
Individual treatment. The production output of 2 (i.e., the production output closest to the
average amount needed for goal attainment) is reported significantly more often in the Team
treatment (p=.017, Fisher's exact test). The depicted distributions disclose a further
interesting detail: in both treatments we find that subjects do not over-report a production
output of 3 which represents the threshold to be (on average) reached in order to attain the
goal (the frequencies of this production output do not significantly differ across treatments,
p=.152, Chi2 test). Contrarily, in both treatments, the production output of 4 is significantly
more often reported than expected according to a uniform distribution of reported production
outputs. In the Individual treatment it even represents the mode within the distribution of
reported production outputs and is weakly significantly more often reported than in the Team
treatment (p=.089, Chi2 test).
In the following, we will account for the specificity of the induced goal threshold
considering a production output of 3 as a reference point because it represents the (average)
individual contribution necessary for goal attainment. Therefore, we will compare the
cumulative proportion of subjects reporting a production output of 3 or higher across
treatments. A Chi2-test reveals that the proportion of subjects reporting a production output
of 3 or higher is significantly lower in the Team treatment (70.15%) compared to the
Individual treatment (82.46%) (p=.059, Chi2-test).
12
Table 2. Individual differences explaining reported production output ≥3
Reported production (1) (2) (3)
output ≥3
Team treatment -.123* -.118* -.146**
(.067) (.068) (.068)
Female -.123* -.160**
(.064) (.064)
Age .002 -.002
(.010) (.010)
Openness -.002
(.032)
Conscientiousness .059
(.032)
Extraversion .022
(.032)
Neuroticism -.006
(.035)
Constant .933*** .952*** .524**
(.138) (.863) (1.214)
Observations 181 176 174
R2 .005 .020 .023
Note. The table depicts marginal effects from a probit regression predicting reported production output ≥3.
Robust standard errors in parentheses. The number of observations slightly differs across models due to missing
values. *** p<.01, ** p<.05, * p<.1
To assess the robustness of this finding, in the next step, we will control for the individual
difference variables collected after the die roll task. Relating gender, age and Big Five
personality traits with reported production outputs might unveil some further interesting
insights about potential determinants of lying behavior under goal incentives. In Table 2 we
run a series of probit regression models in order to predict reported production outputs ≥3 (a
reported production output <3 is coded 0, a reported production output ≥3 is coded 1) by
stepwise including a dummy variable for the treatment (the Individual treatment is coded 0
and the Team treatment is coded 1), Female (male subjects are coded 0 and female subjects
are coded 1), Age, and the elicited Big Five personality factors as explanatory variables. To
assess the influence of the Big Five personality factors, we included four of the five
personality factors in model (3). Scale reliability is acceptable for Extraversion (Cronbach’s
α=.768), Conscientiousness (.493), Neuroticism (.601), and Openness (.645). For
Agreeableness scale reliability is unacceptably low (.194). Therefore, we exclude this factor
from our analysis.7 Models (1) – (3) show that our finding on the influence of team goal
7 Scale reliability is measured by calculating Cronbach’s Alpha. It is a measure for the overall consistency of a
measure (here the distinct personality factor scales) and expresses how a set of test items (the list of questions for
each personality factor) can be considered measuring a single latent construct (the personality factor) (see
Schnell/Hill/Esser, 2005). According to George/Mallery (2012) a Cronbach’s Alpha ≤.5 is inacceptable.
Rammstedt/John (2007) already note a loss in validity and reliability in their Agreeableness scale with two items
compared to larger measures of the Big Five personality factors.
13
incentives on lying behavior is quite robust, i.e., the likelihood of reporting production
outputs ≥3 decreases under the team goal incentive scheme. We also observe that women
report significantly lower production outputs than men. This effect is also robust when
controlling for individual personality factor scores (models (2) and (3)). There is no
interaction effect between the treatment and subjects’ sex. Model (3) shows that none of the
four personality factors is significantly associated with reporting a production output ≥3.
Taken together, the above figures indicate support for Hypothesis 1.
Observation 1. Under team goal incentives subjects are less likely to report
production outputs larger than the average amount necessary for goal attainment than
under individual incentives.
4.2. Beliefs in the Team treatment
In a second step, we will shed some light on the motives behind subjects’ behavior in the
Team treatment by assessing their beliefs regarding their team mates’ reported production
outputs. As argued above, subjects’ beliefs might play a crucial role in determining subjects’
reports because they might influence their calculation on the necessity and the profitability of
lying. To elicit beliefs, subjects in the Team treatment were asked after they have reported
their production output to state what they believe which number each of the other five team
mates has reported as a result of the die roll.8 The belief elicitation was not incentivized. To
carry out our analyses, we again convert subjects’ stated beliefs into production outputs.
On average, subjects believe their team mates to report a production output of 3.65 which
is statistically weakly higher than the average reported production output (p=.09, Fisher-
Pitman permutation test for paired replicates). There is also a significant positive correlation
between subjects’ average belief on their team mates’ reports and their own reported
production output (ρ=.539, p=.000, Spearman rank correlation).
To analyze internally homogenous subgroups with regard to our theoretical
considerations we cluster subjects according to their accumulated belief on their team mates’
reported production output (i.e., the sum of all beliefs regarding their team mates).9 As we
have reasoned, if subjects have lying costs, they will only lie when it is profitable (i.e., if
they believe that lying contributes to the team’s goal attainment) and necessary (i.e., if the
team has not yet attained the goal according to their belief). In Table 3 we have depicted the
distribution of subjects’ beliefs and their reported production outputs.
8 Please refer to the Appendix for the questionnaire on belief elicitation.
9 Due to the limited number of observations for each subgroup, we mostly do not apply statistical tests. Hence, we
believe that a descriptive analysis is still insightful with regard to our hypotheses.
14
Table 3. Subjects’ beliefs and reported production outputs
Reported production output pi (frequency)
Sum of Belief Frequency (%) IL AV 0 1 2 3 4 5
10 2 (3.03) 0 .5 1 1
11 1 (1.52) 0 4 1
12 4 (6.06) 0 1.25 2 1 1
13 1 (1.52) .83 4 1
14 3 (4.55) .67 1.67 2 1
15 7 (10.61) .50 3.14 1 2 3 1
16 4 (6.06) .33 3 1 1 1 1
17 4 (6.06) .17 2.75 1 1 1 1
18 9 (13.64) 0 3.33 3 2 2 2
19 3 (4.55) 0 3.33 1 1 1
20 4 (6.06) 0 4 1 2 1
21 13 (19.70) 0 4.08 2 1 4 6
22 3 (4.55) 0 5 3
23 2 (3.03) 0 3 1 1
24 1 (1.52) 0 4 1
25 5 (7.58) 0 4.8 1 4
Note. Sum of beliefs denotes the sum of beliefs on team mates’ reported production outputs. AV
denotes the average reported production output. IL denotes the incentive to lie which is given by (1 – the
probability of truthful goal attainment).
We find that 10.45% of the subjects believe that their team mates have reached a total
production output of <13, i.e., they believe that their team mates do not lie for their own or
their team mates’ advantage. In this range lying will not be profitable because, based on a
subject’s beliefs, the team goal threshold cannot be reached even with the highest possible
production output of 5. Subjects in this range on average reported a production output of
1.43. We find no evidence that reported production outputs of this subgroup significantly
differ from the expected output of an honest die roll (p=.453, binomial test). According to
our deliberations, if subjects believe their team mates to report in sum at least 13 points
(which is above the expected sum of production outputs yielded by truthful reports) they
might consider lying to be profitable (because they can altogether reach the team goal) and
necessary (if the team goal has not been reached yet). Based on our data, this subgroup
consists of 28.36% of all subjects. However, this group on average believes that the sum of
the other team members’ reports is 15.37 points. Hence, lying is not quite necessary because
the goal threshold can be almost reached by honesty. This subgroup of subjects reports an
average production output of 2.84 points. Again, we find no evidence that reported
production outputs of this subgroup significantly differ from the expected output of an
15
honest die roll (p=.648, binomial test). Taken together, the behavior of these two subgroups
delivers some support for Hypothesis 1. Contrarily, 61.19% of our subjects believe that the
team goal has already been reached by the reported production outputs of the other team
members. Yet, they report on average 3.98 points, which is statistically significantly larger
than the amount that can be expected from a truthfully conducted die roll (p=.001, binomial
test). This finding regarding a third subgroup of subjects indicates some support for
Hypothesis 2, i.e., these subjects might have lied in favor of their team members no matter
what they expect the other team members to do or because others are expected to lie as well
(care and justification motive). If subjects believe that others also lie for them their choice
might have as well been guided by reciprocity considerations. Subjects in the later subgroup
might have also shifted the responsibility for their lies relying on the fact that the final team
output is a composite of all team members’ individually reported production outputs, i.e.,
their share is indistinguishable from the other reports, especially when they believe that the
others report high outputs as well. A third explanation addresses the possibility that
dishonest subjects have ex-post adapted their beliefs to their reports in order to merely signal
a care or justification motive. Alternatively, these subjects might have no lying costs and
expect other subjects to also not have lying costs reflected by the expectation and the actual
report of high production outputs.
Finally, based on subjects’ beliefs we can also calculate their actual incentive to lie which
is given by the probability of not reaching the goal truthfully. If the sum of a subject’s beliefs
is smaller than 13, subjects have no incentive to lie because lying is not profitable, i.e., even
the highest possible production output will not be sufficient to attain the goal. Accordingly,
if the sum of beliefs is larger than 17, subjects have no incentive to lie because it is not
necessary, i.e., the goal is supposed to be attained by the expected production outputs of the
other team members. If the sum of beliefs is larger than 12 but smaller than 18, subjects have
an incentive to lie because lying can be profitable and necessary (see Table 3). The average
incentive to lie in the Team treatment, given subjects’ beliefs, is about 12.64% (assuming
that each production output has an equal probability of 1/6). This number is significantly
smaller than the incentive to lie in the Individual treatment which equals 50% for every
subject. As we have shown, if subjects believe that the sum of reported production outputs is
smaller than 17 the prevalence of lying must be very moderate. However, contrary to our
considerations, subjects holding an aggregate belief larger than 17 tend to lie although they
should (at least) be indifferent regarding the choice whether to lie or not.
Observation 2. Subjects’ beliefs are positively correlated with subjects’ reported
production outputs. For subjects who believe that over-reporting is not profitable
because the goal threshold cannot be reached and for subjects for whom, based on
16
their beliefs, over-reporting would be profitable and necessary, we find no evidence
for over-reporting. Subjects who believe that their team mates have already reached
the goal threshold - who represent the majority in our sample - tend to over-report.
5. Discussion and Conclusion
In this article we have examined the influence of two compensation schemes on individual
lying behavior: an individual goal incentive scheme and a team goal incentive scheme. In
both schemes subjects received a bonus when an externally fixed goal was attained. In
accordance with our first hypothesis, we find that under the team goal incentive scheme
subjects are less inclined to report production outputs beyond the reference production
output (3 points) which is on average necessary for goal attainment. We show that this result
is partly driven by those subjects who either believe that lying is not profitable (i.e., the goal
cannot be reached with a lie) or not absolutely necessary (i.e., there is a good chance that the
goal can as well be reached by reporting honestly). Based on their beliefs, these subjects in
the Team treatment have a lower incentive to lie as compared to the subjects in the
Individual treatment. In line with this notion, we do not find evidence that these subjects
over-reported their production outputs. Hence, contrary to our theoretical considerations, we
find that subjects who believe that the team goal has already been reached by their team
mates significantly over-report their production outputs. Across treatment, women are found
to be more honest than men. None of the personality factors was significantly associated
with reported production outputs beyond the (reference) goal threshold.
Overall, our results provide further evidence on the influence of goals on performance
(e.g., Goerg/Kube, 2012) and lying behavior (e.g., Cadsby et al., 2010). Regarding previous
evidence on the prevalence of lying behavior in teams – which conveys that people tend to
lie more in teams due to the possibility to split the benefits of lying with other persons (e.g.,
Wiltermuth, 2011) or because they can diffuse their responsibility for the lie (Conrads et al.,
2013) – we contribute in different ways. Our analysis on subjects’ beliefs in the Team
treatment has shown that subjects’ reports depend on their beliefs regarding the reports of
their team mates. In line with our first hypothesis, we find most subjects with a low incentive
to lie to evidently abstain from lying. However, the subgroup of subjects who believe that the
goal has already been attained by the reported production outputs of their team mates tends
to significantly over-report.
Taken together, as hypothesized, subjects lie less in the Team treatment because they
have a lower incentive to lie as compared to the Individual treatment. However, some
subjects might have lied in the Team treatment because others benefit from their lies or they
can shift the responsibility for their lies. As subjects in the Team treatment in general tend to
17
report a production output beyond the individual reference production output less frequently,
the former seems to over-compensate the later effect.
Our finding on the treatment difference in reporting a production output of 2 – the
production output which is closest to the (reference) output needed for goal attainment – is
related to previous results from Cadsby et al. (2010) and Schweitzer et al. (2004). They find
that people under an individual goal compensation scheme are more inclined to lie about
their performance when being close to reaching the goal. In line with these studies, our data
suggest that subjects in the Individual treatment may feel particularly more encouraged to lie
about their die roll outcome (because lying costs induced by shifting the reported production
output by one point might not be perceived high as compared to the lost benefit from not
lying) if their actual production output is close to the goal threshold.
Moreover, the depicted distribution of reported production outputs in the Individual
treatment adds to an observation made by FFH. They convey that many subjects do not lie to
the full extend, i.e., they report an outcome of 4 instead of 5 in their individual piece-rate
framework. FFH argue that subjects are aware that honesty might be a favorable trait and if a
4 is assessed differently than a 5 in respect to honesty, it might be reasonable not to lie to the
full extent and to try to disguise the lie and appear honest (see also Conrads et al., 2013 for a
similar reasoning). Hence, subjects in our Individual treatment might think that reporting a
production output of 4 – which implies the same monetary benefit as a 3 or a 5 – is assessed
differently than exactly hitting the goal (or clearly reaching it with 5 points) which may look
suspicious. Therefore, they might find it reasonable to disguise their lie by reporting an
output of 4.
Regarding the question whether women are more honest than men when payoffs are at
stake the literature is split. Some studies provide support for this notion (e.g.,
Ross/Robertson, 2000, Dreber/Johannesson, 2008, Pruckner/Sausgruber, 2008,
Ellingsen/Johannesson/Lilja/Zetterqvist, 2009). Yet, there are also studies which endorse that
women are bigger liars (e.g., Tyler/Feldman, 2004, Tyler/Feldman/Reichert, 2006) or that
there are no differences in lying behavior among sexes (Lewis, 1993,
DePaulo/Kashy/Kirkendol/Wyer/Epstein, 1996, Rowatt/Cunninghan/Druen, 1998, Cadsby et
al., 2010, Belot/Schröder, 2013). Our study backs the first stream of literature.
Regarding the disassociation of personality traits with lying behavior, our evidence is
contrary to previous evidence on lying in teams (e.g., Conrads et al., 2013). As an
explanation for this difference might serve the multiplicity of motives which can potentially
influence subjects’ behavior and beliefs in our team setting (see
Lönnqvist/Verkasalo/Wichardt/Walkowitz, 2013, for a discussion on the association between
personality factors and behavior dependent on the complexity of potential motives for a
behavior).
18
From an applied perspective, our findings suggest that although goal settings appear to be
an effective means for motivating agents in order to improve their performance, some
caution is required in situations of asymmetric information regarding the observability of
agents’ actual performance, e.g., concerning employees’ presence at work, or their actual
responsibility for desired outcomes. In this respect, according to our study, individual goals
might distort people’s ethicality in such settings more as compared to team goal settings and
might represent a better alternative, especially for male employees. Referring to the
underlying motives for reporting production outputs under a team goal setting, our
suggestion is twofold: Firstly, to attenuate the diffusion of responsibility motive,
organizations should enhance working environments which foster transparency and shape
beliefs on actual ethical behavior of others. In other words, if agents believe others to behave
honestly, they may also tend to do so. Secondly, based on our theoretical considerations, we
argue that if goals are set realistically (e.g., by involving agents’ opinion, see also
Goerg/Kube, 2012, on the effectiveness of self-chosen goals) agents might have a positive
believe on the contribution of their team mates for reaching the common goal and therefore
refrain from unethical action.
Finally, there are some potential limitations to our work that we want to address. Firstly,
our experiment is a laboratory experiment entailing a non-real effort task. This restricts
generalizability to organizations. Moreover, we only compared two treatments including one
team setting and, in the end, we cannot know whether our subjects actually lied. Future
research could examine our research question in more natural non-laboratory settings, e.g., in
field experiments, altering the number of subjects building a team and tracking actual
individual behavior. Secondly, subjects in our team treatment were not able to interact with
each other which may not be representative of typical team work situations within
organizations. It would therefore be interesting to see how unethical behavior is linked to
team goals and bonuses when interaction and communication between the team members are
enhanced. Thirdly, we are not able to derive causality with our experimental design
concerning subjects’ (non-incentivized) beliefs. The influence of beliefs can, at this point,
only be assumed but seems very plausible (see, e.g., Gächter/Renner, 2010, for a discussion
on this issue). In addition, we elicited beliefs after the die rolling task which might have
distorted them (e.g., those who reported high production outputs might have ex-post
rationalized this choice by stating that they expect others to do the same). Yet, we decided to
elicit beliefs after the main task to not influence subjects’ decision in the die rolling task.
Despite these potential limitations, to the best of our knowledge, our study is the first that
provides controlled evidence on the influence of a team goal incentive scheme on lying
behavior. Hence, fruitful future work could shed more light on the question whether team
19
incentives are effective compared to individual incentive schemes in reference to
performance and ethical considerations.
References
Bandura, Albert/Barbaranelli, Claudio/Caprara, Gian V./Pastorelli, Concetta (1996):
Mechanisms of moral disengagement in the exercise of moral agency. In: Journal of
Personality and Social Psychology, 71(2), 364–374.
Bandura, Albert/Underwood, Bill/Fromson, Michael E. (1975): Disinhibition of aggression
through diffusion of responsibility and dehumanization of victims. In: Journal of
Research in Personality, 9, 253–269.
Becker, Gary S. (1968): Crime and punishment: An economic approach. In: Journal of
Political Economy, 76(2), 169–217.
Belot, Michéle/Schröder, Marina (2013): Sloppy work, lies and theft: A novel experimental
design to study counterproductive behaviour. In: Journal of Economic Behavior &
Organization, 93, 233-238.
Cadsby, C. Bram/Song, Fei/Tapon, Francis (2010): Are you paying your employees to
cheat? An experimental investigation. In: The B.E. Journal of Economic Analysis &
Policy, 10(1), 1–30.
Conrads, Julian/Irlenbusch, Bernd/Rilke, Rainer M./Walkowitz, Gari (2013): Lying and team
incentives. In: Journal of Economic Psychology, 34, 1–7.
Conrads, Julian/Irlenbusch, Bernd/Rilke/Rainer M./Schielke, Anne/Walkowitz, Gari (2014):
Honesty in Tournaments. In: Economics Letters, 123, 90–93.
Degeorge, François/Patel, Jayendu/Zeckhauser, Richard (1999): Earnings management to
exceed thresholds. In: Journal of Business, 72(1), S. 1–33.
DePaulo, Bella M./Kashy, Deborah A./Kirkendol, Susan E./Wyer, Melissa M./Epstein,
Jennifer A. (1996): Lying in everyday life. In: Journal of Personality and Social
Psychology, 70(5), 979–995.
Dreber, Anna/Johannesson, Magnus (2008): Gender differences in deception. In: Economics
Letters, 99(1), 197–199.
Ellenberger, Mischa/Ohms, Elia Nora (2013): The impact of team goals and bonuses on
lying – An experimental study, Master thesis, Seminar of Corporate Development and
Business Ethics, University of Cologne.
20
Ellingsen, Tore/Johannesson, Magnus/Lilja, Jannie/Zetterqvist, Henrik (2009): Trust and
truth. In: The Economic Journal, 119(534), 252–276.
Erat, Sanjiv/Gneezy, Uri (2012): White lies. In: Management Science, 58(4): 723–733.
Fischbacher, Urs/Föllmi-Heusi, Franziska (2013): Lies in disguise – An experimental study
on cheating. Journal of the European Economic Association, 11(3), 525–547.
Gächter, Simon/Renner, Elke (2010): The effects of (incentivized) belief elicitation in public
goods experiments. In: Experimental Economics, 13,3 (2010): 364–377.
George, Darren/Mallery, Paul (2012): IBM SPSS statistics 19 step by step: A simple guide
and reference (12.ed.). London: Pearson.
Gill, David/Prowse, Victoria L./Vlassopoulos, Michael (2012): Cheating in the workplace:
An experimental study of the impact of bonuses and productivity. IZA Discussion Paper
(Vol. 6725), 1–20.
Gino, Francesca/Ayal, Shahar/Ariely, Dan (2009): Contagion and differentiation in unethical
behavior: The effect of one bad apple on the barrel. In: Psychological Science, 20(3),
393–398.
Gino, Francesca/Ayal, Shahar/Ariely, Dan (2013): Self-serving altruism? The lure of
unethical actions that benefit others. In: Journal of Economic Behavior & Organization,
93, 285–292.
Gino, Francesca/Pierce, Lamar (2009): Dishonesty in the name of equity. In: Psychological
Science, 20(9), 1153–1160.
Gino, Francesca/Pierce, Lamar (2010): Lying to level the playing field: Why people may
dishonestly help or hurt others to create equity. In: Journal of Business Ethics, 95(1),
89–103.
Gneezy, Uri (2005). Deception: The role of consequences. In: American Economic Review,
95(1): 384–394.
Goerg, Sebastian J./Kube, Sebastian (2012): Goals (th) at work: Goals, monetary incentives,
and workers' performance. Preprints of the Max Planck Institute for Research on
Collective Goods.
Grover, Steven L./Hui, Chun (2005): How job pressures and extrinsic rewards affect lying.
In: International Journal of Conflict Management, 16(3), 287–300.
Hamilton/Barton H./Nickerson, Jack A./Owan, Hideo (2003): Team incentives and worker
heterogeneity: An empirical analysis of the impact of teams on productivity and
participation. In: Journal of Political Economy, 111(3), 465–497.
21
Hoffman, Jody R./Rogelberg, Steven G. (1998): A guide to team incentive systems.
In: Team Performance Management, 4(1), 22–32.
Jensen, Michael C. (2001): Corporate Budgeting is broken – Let’s fix it. In: Harvard
Business Review, 79, 94–101.
Jensen, Michael C. (2003): Paying people to lie: The truth about the Budgeting Process. In:
European Financial Management, 9(3), 379–406.
Kartik, Navin (2009): Strategic communication with lying costs. In: The Review of Economic
Studies, 76(4): 1359–1395.
Ledford, Gerald E./Lawler, Edward E./Mohrman, Susan A. (1995): Reward innovations in
fortune 1000 companies. In: Compensation & Benefits Review, 27(4), 76–80.
Lewis, Michael (1993): The development of deception. In Michael Lewis & Carolyn Saarni
(Eds.), Lying and deception in everyday life, 90–105, New York: The Guilford Press.
Lönnqvist, Jan-Erik/Verkasalo, Markku/Wichardt, Philipp C/Walkowitz, Gari (2013):
Personal values and prosocial behaviour in strategic interactions: Distinguishing value
expressive from valueambivalent behaviours. In: European Journal of Social
Psychology, 43(6), 554-569.
Loewenstein, George F./Thompson, Leigh/Bazerman, Max H. (1989): Social utility and
decision making in interpersonal contexts. In: Journal of Personality & Social
Psychology, 57(3), 426–441.
Mazar, Nina/Amir, On/Ariely, Dan (2008): The dishonesty of honest people: A theory of
self-concept maintenance. In: Journal of Marketing Research, 45(6), 633–644.
Ordóñez, Lisa D./Schweitzer, Maurice E./Galinsky, Adam D./Bazerman, Max H. (2009):
Goals gone wild: The systematic side effects of overprescribing goal setting. In: The
Academy of Management Perspectives, 23(1), 6–16.
Pruckner, Gerald J./Sausgruber, Rupert (2008): Honesty on the streets – A natural field
experiment on newspaper purchasing. Working Paper (Vol. 24). University of
Innsbruck.
Rammstedt, Beatrice/John, Oliver P. (2007): Measuring personality in one minute or less: A
10-item short version of the Big Five Inventory in English and German. In: Journal of
Research in Personality, 41, 203–212.
Ross, William T./Robertson, Diana C. (2000): Lying: The impact of decision context. In:
Business Ethics Quarterly, 10(2), 409–440.
Rowatt, Wade C./Cunninghan, Michael R./Druen, Perri B. (1998): Deception to get a date.
In: Personality and Social Psychology Bulletin, 24(11), 1228–1242.
22
Schnell, Rainer/Hill, Paul B./Esser, Elke (2005): Methoden der empirischen Sozialforschung.
München: Oldenbourg.
Schweitzer, Maurice E./Ordóñez, Lisa/Douma, Bambi (2004): Goal setting as a motivator of
unethical behavior. In: Academy of Management Journal, 47(3), 422–432.
Shalvi, Shaul/Handgraaf, Michel J. J./De Dreu/Carsten K. W. (2011): Ethical manoeuvring:
Why people avoid both major and minor lies. In: British Journal of Management, 22,
16–27.
Sutter, Matthias (2009): Deception through telling the truth?! Experimental evidence from
individuals and teams. In: The Economic Journal, 119(534): 47–60.
Tyler, James M./Feldman, Robert S. (2004): Truth, lies, and self-presentation: How gender
and anticipated future interaction relate to deceptive behavior. In: Journal of Applied
Social Psychology, 34(12), 2602–2615.
Tyler, James M./Feldman, Robert S./Reichert, Andreas (2006): The price of deceptive
behavior: Disliking and lying to people who lie to us. In: Journal of Experimental
Social Psychology, 42(1), 69–77.
Wiltermuth, Scott C. (2011): Cheating more when the spoils are split. In: Organizational
Behavior and Human Decision Processes, 115(2), 157–168.
Yearta, Shawn K./Maitlis, Sally/Briner, Rob B. (1995): An exploratory study of goal setting
in theory and practice: A motivational technique that works? In: Journal of
Occupational and Organizational Psychology, 68, 237–252.
23
Appendix
Table A1 shows results from simulations with n=100.000 random draws of 6 team members’
die rolls. The first column indicates each distinct case. The second column displays all
possible sums of production outputs which can be achieved in a team. The third column
shows the probability for each distinct case if team members are honest.
Table A1. Possible sums and probabilities of all team members’ productions outputs under honesty
Case Sum Probability
1 0 0.00001
2 1 0.00014
3 2 0.00035
4 3 0.00117
5 4 0.00272
6 5 0.00543
7 6 0.00984
8 7 0.01614
9 8 0.02482
10 9 0.03540
11 10 0.04753
12 11 0.06104
13 12 0.07548
14 13 0.08452
15 14 0.09186
16 15 0.09115
17 16 0.09105
18 17 0.08233
19 18 0.07419
20 19 0.06131
21 20 0.04764
22 21 0.03560
23 22 0.02458
24 23 0.01651
25 24 0.00956
26 25 0.00514
27 26 0.00265
28 27 0.00121
29 28 0.00047
30 29 0.00013
31 30 0.00003
The probability that the team goal is reached honestly (27.9%) is given by the accumulated
probability of cases 19 to 31 (gray area).
24
Table A2 shows results from simulations with n=100.000 random draws of 5 team members’
die rolls. The first column indicates each distinct case. The second column displays all
possible sums which can be achieved. The third column provides the probability for each
distinct case if the 5 team members are honest. The gray area shows cases where lying is
necessary (the team goal has not yet been reached) and profitable (the team goal can be
reached by lying) for the sixth subject. The fourth column indicates the incentive to lie for
the sixth subject. In cases 14 to 18 the incentive to lie is given by the probability of achieving
a distinct sum (13, 14, 15, 16, 17) multiplied by the number of instances where the sixth
subject has to lie (5/6, 4/6, 3/6, 2/6, 1/6) in these cases in order to reach the goal.
Table A2. Incentives to lie dependent on the other team members’ honestly reported die rolls
Case Sum Probability Incentive to lie
1 0 0.00012 0
2 1 0.00056 0
3 2 0.00194 0
4 3 0.00442 0
5 4 0.00937 0
6 5 0.01562 0
7 6 0.02677 0
8 7 0.03914 0
9 8 0.05391 0
10 9 0.07042 0
11 10 0.08387 0
12 11 0.09561 0
13 12 0.10144 0
14 13 0.09856 *5/6=0.08213
15 14 0.09520 *4/6=0.06347
16 15 0.08303 *3/6=0.04152
17 16 0.06860 *2/6=0.02287
18 17 0.05375 *1/6=0.00896
19 18 0.03910 0
20 19 0.02599 0
21 20 0.01608 0
22 21 0.00962 0
23 22 0.00422 0
24 23 0.00186 0
25 24 0.00065 0
26 25 0.00015 0
Total: 0.21894
As shown in Table A1, the team wins honestly with 27.9%. With a probability of 50.3% it is
not profitable to lie for the sixth subject. This number is given by the accumulated
probability of cases 1 to 13. The sixth subject has an incentive to lie of 21.8% - there lying is
necessary and profitable. This number is given by the accumulated probability of cases 14 to
18.
25
Table A3 shows results from simulations with n=100.000 random draws of 5 team members’
die rolls. The first column displays the actual outcome of the sixth team member’s die roll.
The second column displays the sixth team member’s strategy (report honestly, lying)
depending on the actual outcome of her die roll. The third to sixth column depict the sixth
team member’s expected payoff depending on whether she has reported honestly or not and
on her cost of lying (5, 2.00, 1.99, or 0). The gray area indicates cases where the expected
payoff from lying is larger than the expected payoff from reporting honestly.
Table A3. Expected payoffs dependent on actual outcome of die roll, honesty and cost of lying
Cost of lying
Actual outcome
of die roll
Report
(honestly/lying) 5 2.00 1.99 0
0 0 0.48835 0.48835 0.48835 0.48835
5 -2.51595 0.48405 0.49405 2.48405
1 1 0.75710 0.75710 0.75710 0.75710
5 -2.51595 0.48405 0.49405 2.48405
2 2 1.1001 1.1001 1.1001 1.1001
5 -2.51595 0.48405 0.49405 2.48405
3 3 1.51525 1.51525 1.51525 1.51525
5 -2.51595 0.48405 0.49405 2.48405
4 4 1.99125 1.99125 1.99125 1.99125
5 -2.51595 0.48405 0.49405 2.48405
Expected payoffs are calculated in the following:
a) In case of honesty: Multiplying the probability of reaching the goal honestly and the
resulting payoff (i.e., 5). For example, in case the sixth subject has actually rolled an
outcome of 0 and reports it honestly, the team wins honestly with 9.8% (this is the
accumulated probability that the other five team members have already reached a
sum of 18; this probability is given by the sum of the respective probabilities of
cases 19 to 26 in Table A2). The expected payoff in this case is 9.8%*5=0.48835.
b) In case of lying: Multiplying the probability of reaching the goal by lying and the
resulting payoff (i.e., 5). For example, in case the sixth subject has actually rolled an
outcome of 0 but reports 5, the team wins with 49.7% (this is the accumulated
probability that the team members reached a sum of 18; this probability is given by
the sum of the respective probabilities of cases 14 to 26 in Table A2). The expected
payoff in this case is (49.7%*5)−0(2)=2.48405(0.48405) if costs of lying are equal
to 0(2).
The table shows that all team members always report truthfully if they have lying costs
larger than 1.99. In this range, the expected payoff from lying does not outweigh the cost
associated with it, no matter what the outcome of the die roll actually was (assuming
constant costs for lying).
... Another possibility for the lack of a significant effect of group size on collaborative dishonesty lies in the limited heterogeneity of group sizes across studies. With the exception of Conrads et al. (2017), who studied six-person groups, all other reports included in the meta-analysis studied groups of two or three people. Accordingly, if group-size effects emerge only when comparing smaller with relatively larger groups, our ability to detect such effects here is restricted. ...
Article
Although dishonesty is often a social phenomenon, it is primarily studied in individual settings. However, people frequently collaborate and engage in mutual dishonest acts. We report the first meta-analysis on collaborative dishonesty, analyzing 87,771 decisions (21 behavioral tasks; k = 123; nparticipants = 10,923). We provide an overview of all tasks used to measure collaborative dishonesty, and inform theory by conducting moderation analyses. Results reveal that collaborative dishonesty is higher (a) when financial incentives are high, (b) in lab than field studies, (c) when third parties experience no negative consequences, (d) in the absence of experimental deception, and (e) when groups consist of more males and (f) younger individuals. Further, in repeated interactions, group members' behavior is correlated-participants lie more when their partners lie-and lying increases as the task progresses. These findings are in line with the justified ethicality theoretical perspective, suggesting prosocial concerns increase collaborative dishonesty, whereas honest-image concerns attenuate it. We discuss how findings inform theory, setting an agenda for future research on the collaborative roots of dishonesty. (PsycInfo Database Record (c) 2022 APA, all rights reserved).
... As for the frequency of deceptive behavior, it varies in different deceptive circumstances. In competitive situations such as academic tests (Negre, Forgas, & Trobat, 2015;Niiya, Ballantyne, North, & Crocker, 2008;Sideridis, Tsaousis, & al Harbi, 2016), online dating (Guadagno, Okdie, & Kruse, 2012), and money allocation (Conrads et al., 2017;Dreber & Johannesson, 2008;Muehlheusser, Roider, & Wallmeier, 2015), males cheat more to surpass others and maintain their self-esteem (Gneezy, Niederle, & Rustichini, 2003;Niiya et al., 2008). In noncompetitive situations such as conversation, females lie more to foster intimacy, especially when they know they would meet their partner again in the future Tyler & Feldman, 2004). ...
Article
Full-text available
This study investigated the gender differences in deceptive strategies and their neural basis in the perspective of two-person neuroscience. Both male and female dyads were asked to perform a spontaneous sender-receiver deception task, while their neural activities in the prefrontal cortex (PFC) and right temporal parietal junction (rTPJ) were recorded simultaneously using functional near-infrared spectroscopy (fNIRS)-based hyperscanning. Results showed that female dyads showed higher simple deception rate than male dyads did, while male dyads displayed higher sophisticated deception rate than female dyads did. These results demonstrated that different deceptive strategies were used across genders. The fNIRS data showed that the interpersonal neural synchronization (INS) in PFC was significantly enhanced only in female dyads when performed the spontaneous sender-receiver deception task, while INS in rTPJ was increased only in male dyads during performance of task. Moreover, such INS was correlated with the success rate of deception (in both dyads???). Granger causality analysis showed that no significant directionality between time series of PFC (or rTPJ) in each dyad, which could indicated that sender and receiver played equally role during deception task. Finally, enhanced INS in PFC in female dyads mediated the contribution of eye contact to the success rate of deception. All findings in this study suggest that differential patterns of interpersonal neural synchronization in the brain are associated with gender difference of deceptive strategies. We provide the first inter-brain evidence for gender difference of deception, and make a deeper understanding of spontaneous face-to-face deception.
... Unlike die-roll tasks, such tasks enforce a dichotomous decision between reporting a win and a loss. For example, participants could win a flat bonus if they reported having rolled an even number (Ploner & Regner, 2013); if the sum of all pips reached a threshold (Conrads et al., 2016;Dubois, Rucker, & Galinsky, 2015); if the upwardfacing side yielded a greater number than the downward-facing side (or vice versa; Peer et al., 2017); or if the die showed a specific target number, such as a 4 (Hilbig & Hessler, 2013;. Alternatively, a wheel of fortune determined a winner (Lohse et al., 2018). ...
Article
Over the past decade, a large and growing body of experimental research has analyzed dishonest behavior. Yet the findings as to when people engage in (dis)honest behavior are to some extent unclear and even contradictory. A systematic analysis of the factors associated with dishonest behavior thus seems desirable. This meta-analysis reviews four of the most widely used experimental paradigms: sender–receiver games, die-roll tasks, coin-flip tasks, and matrix tasks. We integrate data from 565 experiments (totaling N = 44,050 choices) to address many of the ongoing debates on who behaves dishonestly and under what circumstances. Our findings show that dishonest behavior depends on both situational factors, such as reward magnitude and externalities, and personal factors, such as the participant’s gender and age. Further, laboratory studies are associated with more dishonesty than field studies, and the use of deception in experiments is associated with less dishonesty. To some extent, the different experimental paradigms come to different conclusions. For example, a comparable percentage of people lie in die-roll and matrix tasks, but in die-roll tasks liars lie to a considerably greater degree. We also find substantial evidence for publication bias in almost all measures of dishonest behavior. Future research on dishonesty would benefit from more representative participant pools and from clarifying why the different experimental paradigms yield different conclusions.
... Unlike die-roll tasks, such tasks enforced a dichotomous decision between reporting a win and a loss. For example, participants could win a flat bonus if they reported having rolled an even number (Ploner & Regner, 2013); if the sum of all pips reached a threshold (Conrads et al., 2016;Dubois et al., 2015); or if the die showed a specific target number, such as a 4 (Hilbig & Hessler, 2013;. The standardized report M r and the rate of liars M liars were always adapted to the specific experimental design. ...
Thesis
Full-text available
When and why do people engage in (un)ethical behavior? This dissertation summarizes general theories and synthesizes experimental findings on (non)cooperation, (un)fairness, and (dis)honesty. To this end, Chapter 1 introduces experimental games as a rigorous tool for studying (un)ethical behavior. Chapter 2 demonstrates that small changes in the framing of context (e.g., referring to a social dilemma as a competition vs. a team endeavor) can have long-lasting effects on the participants’ propensity to cooperate. Context framing also shapes beliefs about the cooperative behavior of interaction partners and donations in non-strategic allocation decisions. Taken together, the results suggest that social norm theories provide a plausible explanation for cooperation, including its sensitivity to context framing. Chapter 3 investigates why experimental games regularly find that economics students behave more selfishly than their peers. The concept of social norms is thereby extended to include the enforcement of compliance per sanctions. The results indicate that economics students and students of other majors are about equally concerned with fairness and they have similar notions of fairness in the situation. However, economics students make lower allocations, expect others to make lower allocations, and are less willing to sanction allocations seen as unfair. Skepticism mediated their lower allocations, suggesting that economics students behave more selfishly because they expect others not to comply with a shared fairness norm. Chapter 4 shows that intrinsic sanctions (e.g., shame and guilt) can be sufficient for ethical behavior to emerge. The chapter provides answers to many of the ongoing debates on who behaves dishonestly and under what circumstances. The findings suggest that dishonest behavior depends on situational factors (e.g., reward magnitude and externalities), personal factors (e.g., gender and age) as well as on the experimental paradigm itself.
Article
Full-text available
Executive Overview: Goal setting is one of the most replicated and influential paradigms in the management literature. Hundreds of studies conducted in numerous countries and contexts have consistently demonstrated that setting specific, challenging goals can powerfully drive behavior and boost performance. Advocates of goal setting have had a substantial impact on research, management education, and management practice. In this article, we argue that the beneficial effects of goal setting have been overstated and that systematic harm caused by goal setting has been largely ignored. We identify specific side effects associated with goal setting, including a narrow focus that neglects nongoal areas, distorted risk preferences, a rise in unethical behavior, inhibited learning, corrosion of organizational culture, and reduced intrinsic motivation. Rather than dispensing goal setting as a benign, over-the-counter treatment for motivation, managers and scholars need to conceptualize goal setting as a prescription-strength medication that requires careful dosing, consideration of harmful side effects, and close supervision. We offer a warning label to accompany the practice of setting goals.
Article
Full-text available
Michael Lewis shares his views on the studies that have been conducted to investigate the origins of lying and deception in everyday life. Michael Lewis, a psychologist at Rutgers Medical School, investigates the development of lying as a double-edged social skill in young children. He states that some of these studies have found that particularly children are prone to habits of telling lies about completion of certain tasks assigned to them, despite not performing the given tasks.
Article
The present study tested derivations from social learning theory on the disinhibition of aggression through processes that weaken self-deterring consequences to injurious conduct. Subjects were provided with opportunities to behave punitively under diffused or personalized responsibility toward groups that were characterized in either humanized, neutral, or dehumanized terms. Both dehumanization and lessened personal responsibility enhanced aggressiveness, with dehumanization serving as the more potent disinhibitor. Escalation of aggression under conditions of dehumanization was especially marked when punitiveness was dysfunctional in effecting desired changes. The uniformly low level of aggression directed toward humanized groups, regardless of variations in responsibility and instrumentality of the conduct, attested to the power of humanization to counteract punitiveness. Results of supplementary measures are consistent with the postulated relationship between self-disinhibiting processes and punitiveness. Dehumanization fostered self-absolving justifications that were in turn associated with increased punitiveness. Findings on the internal concomitants of behavior performed under different levels of responsibility suggest that reducing personal responsibility heightens aggressiveness more through social than personal sources of disinhibition.
Article
We propose a novel experimental design to study counterproductive behaviour in a principal agent setting. The design allows us to study and derive clean measures of different forms of counterproductive behaviour in a controlled but non-obtrusive manner. We ask participants to complete a specific task (identify euro coins) and to report their output. Participants can engage in various forms of counterproductive behaviour, none of them being offered to them explicitly. They can make mistakes in the identification task, lie in their report or even steal coins. We present an application of the design to study the effects of different pay schemes (competition, fixed pay and piece rate) on counterproductive behaviour. On average counterproductive behaviour amounts to 10 percent of the average productivity, almost all arising through mistakes and overreporting of output. We find essentially no evidence of theft. Moreover, we find that both productive and counterproductive behaviour are significantly higher under competition than under the two other pay schemes.
Article
This study tests the usefulness of a person-situation interactionist framework in examining the willingness of a salesperson to lie to get an order. Using a survey of 389 salespersons, our results demonstrate that organizational relationships influence willingness to lie. Specifically, salespersons are less willing to lie to their own company than to their customer, than to a channel partner, and finally, than to a competitor firm. Furthermore, respondents from firms with a clear and positive ethical climate are less willing to lie. Finally, our study finds that interactions between personality factors, such as high Machiavellianism and high self-monitoring, and situational factors have an impact on willingness to lie. Our results suggest that firms can take steps to influence employee ethical behavior.
Article
Relying on the framework provided by Schwartz's theory of personal values, we investigated whether values can help explain prosocial behaviour. We first distinguished value‐expressive behaviours from value‐ambivalent behaviours. The former are compatible with primarily one value or with congruent values, the latter with mutually conflicting values. In Study 1, an analysis over all 41 (39 unpublished) samples in which we measured personal values and prosocial behaviour in monetarily incentivized strategic interactions (N = 1289; data collected between 2007 and 2010 in China, Finland, Germany, Israel, and the West Bank) supported our idea that personal values, universalism in particular, predict value‐expressive (prisoner's dilemma cooperation and trust game back‐transfers) but not value‐ambivalent behaviours (trust game transfers and ultimatum game proposals and responses). Study 2 (N = 56) focused on dictator game behaviours, which we expected and found to be strongly value‐expressive. The findings contribute to the ongoing discussion on whether and under which circumstances values shape behaviour. Copyright © 2013 John Wiley & Sons, Ltd.
Article
In this paper we distinguish between two types of white lies: those that help others at the expense of the person telling the lie, which we term altruistic white lies, and those that help both others and the liar, which we term Pareto white lies. We find that a large fraction of participants are reluctant to tell even a Pareto white lie, demonstrating a pure lie aversion independent of any social preferences for outcomes. In contrast, a nonnegligible fraction of participants are willing to tell an altruistic white lie that hurts them a bit but significantly helps others. Comparing white lies to those where lying increases the liar's payoff at the expense of another reveals important insights into the interaction of incentives, lying aversion, and preferences for payoff distributions. Finally, in line with previous findings, women are less likely to lie when it is costly to the other side. Interestingly though, we find that women are more likely to tell an altruistic lie. This paper was accepted by Teck Ho, decision analysis.
Article
In three experiments, we propose and find that individuals cheat more when others can benefit from their cheating and when the number of beneficiaries of wrongdoing increases. Our results indicate that people use moral flexibility to justify their self-interested actions when such actions benefit others in addition to the self. Namely, our findings suggest that when people's dishonesty would benefit others, they are more likely to view dishonesty as morally acceptable and thus feel less guilty about benefiting from cheating. We discuss the implications of these results for collaborations in the social realm.
Article
We investigate the influence of two popular compensation schemes on subjects’ inclination to lie by adapting an experimental setup of Fischbacher and Heusi (2008). Lying turns out to be more pronounced under team incentives than under individual piece-rates, which highlights a fairly neglected feature of compensation schemes. Moreover, when disentangling different motives of the more pronounced unethical conduct under team incentives, we find that subjects tend to lie more under team incentives because they can diffuse their responsibility, i.e., their deceptive acts cannot unambiguously be attributed to them individually. Our findings are robust even when controlling for individual difference variables. In both compensation schemes subjects who are younger, male, high on Extraversion, and high on Neuroticism tend to lie more.