Journal of Economic Psychology 46 (2015) 1225

Contents lists available at ScienceDirect

Journal of Economic Psychology

journalhomepage:www.elsevier.com/locate/joep

Lying and age: An experimental study

Daniela Gltzle-Rtzler
a
b

a,

, Philipp Lergetporer

University of Innsbruck, Austria

Ifo Institute at the University of Munich, Germany

article

info

Article history:
Received 29 December 2013
Received in revised form 1 September 2014
Accepted 6 November 2014
Available online 18 November 2014
JEL classification:
C91
D03

abstract
Lying and deceiving is present not only in adults but also among children and teenagers and represents an
economically and psychological relevant behavioral trait. It is therefore surprising that evidence from
economic experiments on deceptive behavior in children and teenagers is scarce. In this paper, we study how
age influences the propensity to tell black and white lies in a sample of 383 children and teenagers aged
10/11 and 15/ 16 years. Our results show that a non-negligible fraction of subjects in both age cohorts
exhibits lying-aversion and that the propensity to lie decreases significantly with age.
2014 Elsevier B.V. All rights reserved.

PsycINFO Classification:
2840
Keywords:
White lies
Lying
Deception
Experiment
Children
Teenagers

1. Introduction
Lying and cheating is not uncommon in the human species. Day-to-day evidence suggests that lies are frequently told spontaneously and
unintentionally (DePaulo, Kashy, Kirkendol, Wyer, & Epstein, 1996). Nevertheless, lies or cheating on oth-ers have significant economic
impacts. Examples include tax evasion, illegal download of music, riding the bus without a ticket, managers giving a biased feedback in order to
increase their employees effort, claiming sick benefits when healthy, accepting bribes on the job, not returning excessive change in a restaurant
(Azar, Yosef, & Bar-Eli, 2013; Bucciol, Landini, & Piovesan, 2013; Ederer & Fehr, 2007; Erat & Gneezy, 2012; Loewen et al., 2013), or
teenagers misreporting their age when purchasing alcohol and tobacco or trying to enter a night club.
Two opposing understandings of the justification of lying dominate its normative debate (see Gneezy, 2005, for a short account): While the
assumption of homo economicus dictates lying whenever it pays off in economic terms, Christian phi-losophers claim that every lie is a sin,
irrespective of its consequences and therefore prohibited. Gneezy (2005) rejects this
Corresponding author at: Department of Public Finance, University of Innsbruck, Universitaetsstrasse 15, A-6020 Innsbruck, Austria. Tel.: +43 512 507 7467.
E-mail address: daniela.ruetzler@uibk.ac.at (D. Gltzle-Rtzler).

http://dx.doi.org/10.1016/j.joep.2014.11.002 0167-

4870/ 2014 Elsevier B.V. All

rights reserved.

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

13

strict distinction by arguing that the frequency of lying depends on its costs and benefits for the sender respectively the receiver of a deceptive
message (see also Gibson, Tanner, & Wagner, 2012).
With respect to payoff consequences for the receiver, scholars distinguish between Black Lies and White Lies ( Erat & Gneezy, 2012). Black
Lies have been investigated extensively in numerous experimental studies (see, for example, Angelova & Regner, 2013; Danilov, Biemann,
Kring, & Sliwka, 2013; Dreber & Johannesson, 2008; Erat, 2013; Gneezy, 2005; Gneezy, Rockenbach, & Serra-Garcia, 2013; Hurkens & Kartik,
2009; Ismayilov & Potters, 2013; Lightle, 2013; Sutter, 2009; Snchez-Pags & Vorsatz, 2007). Those lies harm the receiver while being
profitable for the sender and rep-resent the classic form of deception. Many examples taken from real life correspond to this lying type such as
taxi driv-ers giving wrong information to a foreign passenger in order to overcharge or lies which are told for concealing a theft. White Lies, on
the contrary, are characterized by the fact that they benefit the receiver. Stating You look good today! could be a lie of this type. An
economically more important example is a manager who gives too positive a feedback to a badly performing employee. In such a situation, both
parties may profit from the lie: the employee receives intrinsic sat-isfaction and the manager prevents a decrease in productivity as negative
feedback may hamper the employees willing-ness to work. Doctors who dose their patients with placebos (see Erat & Gneezy, 2012) or parents
and teachers who exaggerate their praises towards a poorly performing child in order to avoid decreasing effort levels due to frustration fall into
this category as well. Experimental evidence on White Lies is still somewhat scarce. Important exceptions rep-resent the work by Erat and
Gneezy (2012), Cappelen, Srensen, and Tungodden (2013), Gino, Ayal, and Ariely (2013) and Lightle (2013).

So far, experimental studies have only considered adults attitudes towards different types of lies. In this paper we use Erat and Gneezys
(2012) design in order to investigate the lying behavior among 383 children and teenagers, being either 10/11 or 16/17 years old. Deceptive
behavior of children is economically relevant in several dimensions. An important exam-ple is the consumption of tobacco and alcohol.
Consuming these products before the legal age to do so (which is, for instance, sixteen in many EU-countries) is clearly related to deception as
misreporting age is an integral part of the acquisition pro-cess. Despite its dramatic health consequences, 20% of all 15-year-old Austrians were
1

smokers in 2006. As smoking induces annual net costs of 511.4 million for the Austrian economy, lying of teenagers as means to illegally
2

acquiring these products has a major impact on the economy (Pock, Czypionka, Mllbacher, & Schnabl, 2008).
Apart from that, a well-established finding from psychological studies is that lying in childhood is positively related to a broad range of other
problematic behaviors such as disruptive behavior, aggression, conduct disorders, stealing, truancy or delinquency (Gervais, Tremblay,
Desmarais-Gervais, & Vitaro, 2000; Gervais, Tremblay, & Hroux, 1998; Rutter, Tizard, & Whitemore, 1970; Stewart & DeBlois, 1985;
Stouthamer-Loeber & Loeber, 1986). Furthermore, longitudinal studies reveal that lying at an early age serves as a good predictor for
delinquency or drug use in adolescence and adulthood (Farrington, Ttofi, & Coid, 2009; Mitchell & Rosa, 1981; Pulkkinen, 1983).
The evidence discussed so far emphasizes the importance to extend the economic research towards deceptive behavior in children and
teenagers. Nevertheless, lying in children has received little attention in the economic literature so far. A nota-ble exception is Bucciol and
Piovesan (2011) who examine the relationship between honesty, age and self-control in an incentivized decision problem. Their subjects (aged 5
15 years) were asked to toss a fair, two-colored coin (black and white) in private and report the outcome to the experimenter. Participants only
received a gift if they stated that the outcome was white. They find that, on average, more than 85% claimed to have tossed the prize winning
color (this fraction is signifi-cantly above 50% and lower than 100%). Therefore, quite a few, but not all children cheated. The tendency to cheat
is uniform by age, gender and the number of siblings.
A drawback of this approach, however, is that individual behavior is unobservable and thus the possibilities for investi-gating determinants of
the development of lying are limited. We overcome this constraint by using Erat and Gneezys (2009, 2012) design which yields observable
individual deceptive behavior, and thus allows us to disentangle determinants of lying in children and adolescents.
The rest of the paper is structured as follows: The experimental design is described in the next section and our results are presented in Section
3. Section 4 concludes the paper.
2. Experimental design
We use the same cheap-talk senderreceiver game as Erat and Gneezy (2009, 2012). At the beginning of this sequential two-player game,
3
only the sender is informed about the outcome of rolling a six-sided die. She is then asked to send one out of six possible messages to the
receiver. These messages are The outcome of the roll of the die is i where i e {1, 2, 3, 4, 5, 6}. The sender knows that the payment depends on
the decision of the receiver and that the message sent is the only information the receiver will get about the die roll. There are two alternative
payment options A and B, which constitute the payoffs for
1

The legal age for drinking and smoking in Austria is 16 years. Dr, Griebler, and Kremser (2010) claim that insufficient child safety locks on cigarette machines lead to easy
accessibility, which, in turn, promotes high consumption rates. Some studies report that 80% of adult smokers started before their eighteenth birthday (see Moolchan, Ernst, &
Henningfield, 2000).
2
See Currie et al. (2012) for similar facts on alcohol consumption.
3
Following Erat and Gneezy (2012), the senders did not roll a die themselves in order to being able to record whether the chosen message was a lie (see Appendix A for
details).

14

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

Table 1
Payoffs (in points) in the different treatments for sender (x) respectively receiver (y) and lying-type-categorization.
Treatment

Payment option

T[ 5, +10]
T[ 1, +10]
T[+1, 1]
T[+1, 5]
T[+1, +10]
T[+10, +10]
T[+1, 0]
T[+10, 0]

Lying types

A (xa, ya)

B (xb, yb)

(20, 20)
(20, 20)
(20, 20)
(20, 20)
(20, 20)
(20, 20)
(20, 20)
(20, 20)

(15, 30)
(19, 30)
(21, 19)
(21, 15)
(21, 30)
(30, 30)
(21, 20)
(30, 20)

Altruistic White Lies

Selfish Black Lies
Strong Pareto White Lies
Weak Pareto White Lies

the sender and the receiver. Payoffs for both players in each option are private information for the sender and she furthermore knows that the
receiver will remain ignorant about these payoffs. After reading the senders message, the receiver chooses j e {1, 2, 3, 4, 5, 6}. If j corresponds
to the actual outcome of the roll of the die, payment option A is implemented. If the receiver chooses any other number, payoffs for both players
are determined according to payment option B. Both sender and receiver are informed about this payment rule.
Inspired by Erat and Gneezy (2009, 2012) we use eight treatments which differ from one another with respect to the rel-ative payoff
consequences of lying for the sender and the receiver. Table 1 summarizes the treatments and classifies them into four different types according
to the players relative gains/losses induced by the lie.
If the receiver chooses the number which was actually rolled, payment option A is implemented. In this case, a payoff of 20 for both players
is realized in all treatments. In this design, messages that indicate the actual number are referred to as truth telling. If, in contrast, the receiver
chooses any other number, payoffs are determined by option B. The relative differ-ences in payments for sender and receiver in option A and
option B constitute the experimental treatments. Generally speak-ing, in treatment T[X, Y] the sender receives 20 + X in Option B, while the
receiver gets 20 + Y. Both X and Y, may be positive, negative, or zero. For example, in T[ 5, +10] option B yields a payoff of 15 for the sender
and 30 for the receiver. In this treat-ment, lying to the receiver about the actual outcome of the roll in order to implement payment option B
rather than A dimin-ishes (increases) the payoff of the sender (receiver) by 5 (10) points. According to these payoff-consequences we assign each
treatment to different lying types.
By telling an Altruistic White Lie the sender sacrifices some of her payoff in order to increase the payoff of the receiver (i.e., X < 0 and Y >
0). On the contrary, Selfish Black Lies enlarge the senders payoff at the expense of her opponent (i.e., X > 0 and Y < 0). Strong Pareto White
Lies are beneficial for both players (i.e., X > 0 and Y > 0) and Weak Pareto White Lies are pay-off neutral to the receiver and increase the
4

senders payoff (i.e., X > 0 and Y = 0).

Since the risk of being busted is an integral part of lying, we also elicited risk preferences as a control in our regressions in an incentivized
investment game. Following Gneezy and Potters (1997) and Charness and Gneezy (2010) subjects are endowed with 10 tokens and asked how
many tokens they wanted to invest in a risky lottery. The lottery yields three or zero points for each token invested. That is, the tokens are either
tripled or lost. The chance of winning is 50%. Let k be the amount invested. The payoff for subject i in this game is

3k

if win

pi 10 k
0
if loss
After collecting all decision sheets, a coin was tossed in front of the class to determine the outcome of the lottery (see Appendix B for
instructions and decision sheet).
The experiment was run as pen and paper experiment in four secondary schools in the Federal State of Tyrol, Austria. The State Board of
Education in Tyrol and the schools headmasters approved this project. Furthermore, the parents of the par-ticipating children and teenagers were
informed about the experiment in a letter without revealing any content-related information. We analyze the decisions of 383 5th and 11th
graders (aged 10/11 and 15/16 years, respectively) from 20 dif-ferent classes in the role of the sender and 69 subjects from the same grades as
receivers. The total numbers of participants split up by gender, age and role are presented in Table 2.
In this experiment, subjects earned tokens which were exchanged for cash afterwards. In order to provide similar mon-etary incentives, stake
sizes were adjusted to average pocket money rates of the respective age group (see Fehr, Gltzle-Rtzler, & Sutter, 2013). For 5th and 11th
graders, one point was converted into 0.1 and 0.4 Euros, respectively. Subjects received their payments of both the senderreceiver game and the
risk task several days after the experiment in sealed enve-lopes marked with an anonymized ID.

It is important to note that this experimental design is intended to . . . reveal important insights into the interaction of incentives, lying aversion and preferences for payoff
distributions. (Eart and Gneezy, 2012, p. 723).

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

15

Table 2
Number of participants by age and gender.
Age

Senders
Girls

10/11 years (5th grade)

16/17 years (11th grade)
Total

110
86
196

Receivers
Boys

Total

Girls

Boys

Total

114
73
187

224
159
383

22
13
35

20
14
35

42
27
69

Each participant was asked to answer two control questions in order to check for comprehension of the game. Ten senders and one receiver who did not answer both questions
correctly were excluded from our analysis. Therefore, we analyzed the decisions of 383 (69) subjects in the role of the sender (receiver).

Fig. 1. Average lying frequencies by lying type and age group.

Each session of the experiment lasted about 100 min, which includes a post-experimental questionnaire on demographic data (on siblings and
5
gender) and the measure of risk attitudes at the very end of the session. Following a fixed script, the experiment was explained in great detail at
the beginning of each session. We phrased our instructions as simple as possible in order to assure comprehension, particularly for our younger
participants. Prior to the decision phase, participants had to answer control questions privately. Subjects who failed to answer these questions
6
correctly were dropped from the analysis (this was the case for six 5th-graders and five 11th-graders). See Appendix A for the instructions,
control questions and decision sheets. All questions from the participants were answered privately.
Erat and Gneezy (2012) present evidence from two experiments on deception. In their first experiment, each decision maker participated in
one treatment only (between-subject design) while each decision maker participated in several treat-ments in their second experiment (betweensubject design). They find the same behavioral patterns in both experiments. We use a within-subject design which allows us to analyze the
decisions of the same individuals in different treatments. Senders (receivers) had to choose a message (number) in each of the eight randomly
ordered treatments. After senders made their decisions, the answer sheets were collected and one treatment was randomly determined to be
payoff relevant for both players. At the beginning of the experiment, each sender was informed that she would be matched with a receiver from
the same grade in a different school. Apart from that, our participants did not receive any further information (e.g. demographic details) about
their opponents.
The receivers were not informed about which treatment was selected for payment by the sender and they made their choice for each treatment
in the upper part of the decision sheets. The lower part (including the payoff information) was cut off after the sender-sessions. Due to the
surplus of senders, each receiver was matched with five to eight senders and received the payment from the relevant treatment of each partner
7

(see Appendix A for details). The receivers followed the senders message in 35% of all cases which is significantly different from random
choices which would induce a follow-ing-rate of 1/6 (p < 0.001, Wilcoxon signed rank test). In what follows, we focus exclusively on the
decisions of senders.
5
6

Participants were informed about the risk task only after completing the questionnaire in order to avoid any influence of this task on the decisions in the senderreceiver game.

It is worth emphasizing that the ability of taking another persons perspective is a necessary prerequisite for participating in strategic interaction experiments. Psychological
studies on the development of Theory of Mind show that children are able to differentiate the others view from their own one already by the age of four to six years ( Astington,
1993; Selman, 1971). Thus, age differences in the ability to think strategically are unlikely to bias our results.
7
As in Erat and Gneezy (2012), we told each sender that she interacts with another partner which does not exclude the possibility that the receiver interacts with several
senders. It is very important to emphasize that we did not deceive our subjects at any point and that this matching protocol cannot influence our senders choices.

16

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

Fig. 2. Average lying frequencies by age and treatment. The arrows represent the sensitivity of lies to incentives. The numbers at the arrows are the p-values of two-sided
McNemars chi-squared tests. Lying frequencies of 10/11-year olds (16/17-year olds) are presented in the gray (white) boxes.

3. Experimental results
3.1. Aggregate treatment differences among both age groups
3.1.1. Main results
Across all treatments and both age cohorts we find that our subjects send a deceptive message in 62.4% of all cases. In the aggregate, 5thgraders lie significantly more often (64.4%) than 11th-graders (59.6%) (p < 0.05, Mann Whitney U-test). As shown in Fig. 1, we find a
remarkable degree of heterogeneity in lying frequencies across lying types: average lying rates are highest in the domain of (Weak and Strong)
Pareto White Lies (75% and 69%, respectively), followed by Selfish Black Lies (66%) and Altruistic White Lies with only 38%. While 5thgraders tell more Weak Pareto White Lies than 11th-graders do (79% vs. 70%; p < 0.01, Mann Whitney U-test), we find no statistically
significant age effects in the other lying types (p > 0.1, Mann Whitney U-tests). Thus, the overall age effect appears to be due to differences in
telling a Weak Pareto White Lie.
This aggregate view on the data might conceal important differences between the single treatments. Therefore we show the lying frequencies
in all eight treatments in Fig. 2. The case of truth telling is represented by the origin of the graph: If the receivers choice is the same as the actual
outcome, payoffs of 20 for both players are realized. Pareto White Lies (Altruistic White Lies) are in the upper right (left) quadrant and the lower
right quadrant comprises Selfish Black Lies.
Lying frequencies of 10/11-year olds (16/17-year olds) in each treatment are presented in the gray (white) boxes. The signif-icance levels of
McNemars chi-squared tests between treatments are indicated at the arrows. In general, our results are largely in line with Erat and Gneezys
(2009, 2012) findings for adults. The fact that 18% of the 5th graders and 17% of the 11th graders opted for truth-telling in T[10, 10] even if
lying would yield a Pareto-improvement is particularly noteworthy since it offers evidence for the existence of pure lie aversion already at a
young age. Furthermore, Fig. 2 indicates that an increase in senders payoffs from lying when one moves from left to right increases the
relative frequency of lying. Hence, the incentives for senders matter for the decision to lie. The picture from Fig. 2 is less clear cut when we look
at the effects of the receivers payoffs from lying when one moves from bottom to top. The regression analysis below will shed more light on
this aspect of the decision to lie.
3.1.2. Robustness check
Erat and Gneezy (2009) contrast treatment T[10, 10] (in which option A yields 30 ECUs for each player and option B yields 20 ECUs per
player) with a reversal of payoffs, T[ 10, 10], in which truth-telling yields an extra 10 ECUs. Since lying in T[ 10, 10] reduces payoffs for both
8

the sender and the receiver by 10 ECUs, such behavior might indicate the prevalence of cognitive limitations. Therefore, it is a worthwhile
exercise to test for age differences in lying rates in T[ 10, 10] which would hint to cognitive differences which were not captured by our control
questions. To this end, we ran treatments T[ 10, 10] and T[10, 10] with a different set of 65 5th- and 61 11th graders as a robustness check (4
5th-graders and 2 11th-graders did not understand the task and were thus excluded from the analysis). Apart from the different treatments, we
employed exactly the same protocol as in the original experiment. The main finding of this robustness check is that 20% of the 5th-graders and
2

17% of the 11th-graders chose to lie in T[ 10, 10] (p = 0.70, chi -test). The fact that lying rates in the reverse treatment do

We thank an anonymous referee for emphasizing this important point.

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

17

Table 3
Probit regression on the decision to lie over all subjects (clustered on individual level).
Explanatory variables

Dependent variable decision to lie

Female
Grade 11
Risk

(3.1) All subjects

(3.2) All subjects

(3.3) All subjects

(3.4) All subjects

0.024
(0.023)
**
0.047
(0.023)
0.0002
(0.004)

0.021
(0.023)
**
0.045
(0.023)
8.21e 05
(0.004)
0.014
(0.010)

0.023
(0.025)
**
0.048
(0.024)
7.48e 05
(0.004)
0.016
(0.011)
***
0.027
(0.002)
***
0.010
(0.002)

3064

3064

3064

0.023
(0.025)
**
0.049
(0.024)
7.10e 05
(0.004)
0.016
(0.011)
***
0.010
(0.003)
**
0.006
(0.003)
***
0.024
(0.004)
3064

# Siblings
Own benefit
Other benefit
Disad. inequality

Observations

Robust standard errors in parentheses; standard errors are clustered over treatments for each of the 383 subjects. p < 0.1.
a

Risk is measured as the number of tokens invested in a risky lottery.

b This variable takes the value of Other benefit-Own benefit if Other benefit > Own benefit and 0 else.

* p < 0.05.

p < 0.01.

Table 4
Clustered probit regression on the decision to lie by lying type (clustered on individual level).
Explanatory variables

Dependent variable decision to lie

(4.1) Altruistic White Lies

Female
Grade 11
Risk

Own benefit

0.026
(0.043)
0.052
(0.042)
0.001
(0.007)
0.001
(0.007)

Other benefit
# Siblings
Observations

0.019
(0.020)
766

(4.2) Selfish Black Lies

0.001
(0.040)
0.061
(0.041)
5.35e 05
(0.007)

8.69e 05
(0.007)
**
0.041
(0.018)
766

(4.3) Strong Pareto White Lies

(4.4) Weak Pareto White Lies

0.021
(0.038)
0.031
(0.037)
0.003
(0.007)
***
0.030
(0.004)

0.040
(0.037)
***
0.094
(0.037)
0.001
(0.006)
*
0.005
(0.003)

0.015
(0.017)
766

0.021
(0.016)
766

Robust standard errors in parentheses; standard errors are clustered over treatments for each of the 383 subjects.
Risk is measured as the number of tokens invested in a risky lottery.
a
b

Sample restricted to treatments T[ 5, +10] and T[ 1, +10]; dependent variable equals 1 if the sender lies, 0 else.
Sample restricted to treatments T[+1, 1] and T[+1, 5]; dependent variable equals 1 if the sender lies, 0 else.
Sample restricted to treatments T[+1, +10] and T[+10, +10]; dependent variable equals 1 if the sender lies, 0 else.
Sample restricted to treatments T[+1, 0] and T[+10, 0]; dependent variable equals 1 if the sender lies, 0 else.
e
* p < 0.1.
* p < 0.05.
*
p < 0.01.
c

not differ with age indicates that subjects in both age cohorts exhibit the same degree of (ir)rational behavior and it is therefore unlikely that
cognitive differences drive the age effect reported above. Furthermore, the fact that the lying rates of 76% in T[10, 10] are similar to those found
in the original experiment points towards the robustness of our results.
3.2. Determinants of lying on the individual level
3.2.1. Aggregate results
The results of clustered probit regressions over all eight lying treatments are presented in Table 3. The dependent variable is coded 1 if the
sender lied and 0 otherwise. In Regression (3.1) we include the dummies Female and Grade 11 as well as our measure for risk-taking.

18

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

While there is no significant effect of gender and risk tolerance on the overall propensity to deceive, lying frequencies decrease significantly
with age. This finding resembles the age effect in our non-parametric analysis. Regression (3.2) shows that the number of siblings does not
influence the decision to lie. We consider the benefits induced by lying for sender and receiver (measured in deviations from the truth telling
payoffs) as additional explanatory variables in specification (3.3) in order to quantify the effect of incentives. As expected, own gains have a
significant and positive impact. Receivers benefits, in contrast, influence the decision to lie in the opposite direction. Regression (3.4) shows that
this effect is driven by equality concerns: When controlling for disadvantageous inequality, we find that subjects lie more in order to increase
their own or the partners payoff. If a lie yields a relative payoff disadvantage for the sender, however, she lies less.

One could deduce a general increase in lie aversion as subjects get older from the results in Table 3. However, Fig. 2 indi-cates that this
conclusion is misleading as lying frequencies of 11th graders are lower in some, but not all treatments. Hence, we subsequently discuss the
effects of our explanatory variables in each lying type separately.

3.2.2. Analysis of each lying type

Table 4 displays separate probit regressions on the propensity to lie for each lying type. In each of the four regressions the sample is restricted
to the treatments on the specific lying type. The dependent variable equals 1 if the sender decides to tell a lie and 0 if she tells the truth. We do
not find any gender, age, risk-taking or incentive-effects for Altruistic White Lies or Selfish Black Lies (see Regressions (4.1) and (4.2)).
However, specification (4.2) reveals that subjects with more siblings are more reluctant to tell Selfish Black Lies. The analysis of Strong
Pareto White Lies is shown in Regression (4.3). Our senders tend to lie more in this lying type if ten rather than one extra points are at stake for
them while keeping the receivers gain constant at ten. Specification (4.4) reveals that the aforementioned negative age effect on lying is driven
by Weak Pareto White Lies. Furthermore, a weakly significant and positive effect of own benefits on lying frequencies is detected for this lying
type.
4. Conclusion
In this paper we study the effects of age, incentives and socio-demographic variables on the lying behavior of children and teenagers. Besides
investigating the origins of deceptive behavior in adults, analyzing lying in children and adolescents is important for at least two reasons. First,
smoking and alcohol consumption of minors is clearly associated with lying in the acquisition process and induces significant health-related
costs (see Currie et al., 2012; Dr et al., 2010; Moolchan et al., 2000; Pock et al., 2008). Second, numerous psychological studies show that lying
in children is correlated with various kinds of delinquent behavior and serves as a good predictor for such problematic attitudes in adulthood (see
Farrington et al., 2009; Gervais et al., 1998, 2000; Mitchell & Rosa, 1981; Pulkkinen, 1983; Rutter et al., 1970; Stewart & DeBlois, 1985;
Stouthamer-Loeber, 1986; Stouthamer-Loeber & Loeber, 1986; Warr, 2007). While many studies investigate lying of adults in the domain of
Selfish Black Lies (that is, lies which increase the payoff of the sender on the receivers expense), behavior in other economically relevant lying
types have been examined only recently for the first time in Erat and Gneezys (2009, 2012) papers. In order to study age differences in different
dimensions of lying, we employ their cheap-talk senderreceiver design in an experiment with 383 children and teenagers from two age groups,
10/11-year olds and 16/17-year olds, respectively.

Our results show that lying frequencies decrease significantly with age and that a considerable fraction of subjects in both age cohorts
exhibits lying-aversion. It is reassuring that these findings are consistent with psychological studies: Busseys (1992) results show that children
aged from 4 to 11 years already morally disapprove false state-ments and that they associate negative internal values (e.g. guilt) to lies. Lee and
Ross (1997) find that the willingness of evaluating a statement as a lie is increasing between age 12 and 19. Since recent research in experimental
econom-ics has highlighted the fact that distributional preferences change between the age cohorts under consideration, one could argue that the
negative age effect on lying is entirely driven by differences in social preferences. Since several well-established findings on age dynamics in
social preferences are inconsistent with this objection (for instance, more pronounced efficiency concerns as reported in Alms, Cappelen,
Srensen, and Tungodden (2010) and Fehr et al. (2013), would predict that older subjects lie more in our experiment), it is unlikely that our
results are entirely driven by social preferences.

10

In our regression analysis, we additionally controlled for gender, risk tolerance and the number of siblings. We were not surprised to find no
11
effect of gender on lying since results from psychological as well as economic literature are ambiguous.

Note that the effects of own and other benefits within each type can only be measured if there is variation in the variables. In Altruistic White Lies, for example, other benefits
cannot be analyzed because the receivers payoff does not change across treatments.
10 While not very likely, we note that age differences in lying could possibly also be driven by age differences in the utility from the incentives provided. As mentioned in
Section 2, we tried to minimize this possibility by adjusting point values to average pocket money rates.
11 Dreber and Johannesson (2008) find that men are more likely to tell lies to secure own benefits. Erat and Gneezy (2012) replicate this result and show that females tell more
Altruistic White Lies. Other economic studies do not detect any gender effects in the decision to lie (Aoki, Akai, & Onoshiro, 2010; Childs, 2011; Gibson et al., 2012). In their
psychological study Jensen, Arnett, Feldman, and Cauffman (2004) find that young men are more likely to lie than women while DePaulo et al. (1996)

opposite.

report the

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

19

It is interesting that risk tolerance does not correlate with lying since the risk of being caught is inherent to each attempt to deceive. Finally, the
number of siblings is negatively related to the propensity to tell a Selfish Black Lie. In accordance with other studies (see e.g., Fehr, Bernhard, &
Rockenbach, 2008; Sutter et al., 2010) this finding underpins the importance of family struc-tures for the development of economic decision
making.
We see several interesting avenues for future research. First, it would be interesting to investigate whether and how the knowledge about the
interaction partners demographic details (e.g. gender) influences the decision to lie respectively to fol-low a message. Second, another
straightforward extension of our study would be to examine how lying develops in children from younger age cohorts. Third, it would be
worthwhile to study behavioral spillovers from peers, parents and siblings on an individuals decision to lie. Finally, long-term intervention
studies which aim to decrease deceptive behavior in children would be very fruitful.

Acknowledgements
We would like to thank Thomas Plankensteiner and Christina Wallas from the State Board of Education in Tyrol and the headmasters of the
participating schools (Christoph Berger, Gerlinde Christandl, Max Gnigler, Hermann Lergetporer and Gerhard Sailer) for making this study
possible. We have received very helpful comments from Matthias Sutter, Silvia Angerer, two anonymous referees and audiences at the IMEBE in
Castellon and the eeecon-workshop at the University of Innsbruck. Financial support through the Austrian Federal Ministry of Science program
Sparkling Science (Grant SPA/02-99 Project: Gender and competition) and the University of Innsbruck is gratefully acknowledged.

Appendix A. Experimental instructions, control questions and decision sheets of the senderreceiver game
A.1. Experimental instructions for the senderreceiver-game SENDER (translated from German)
Note: Italic font is used for the instructions to the experimenter.
Welcome to our game and thank you for participating. The game will take about two school hours (100 min) in total and you can earn money
in the game. You are asked to make choices during the game and the amount of money you earn is influ-enced by your own decisions. For this
reason it is very important that you properly understand the rules of the game. Please raise your hand if you have any questions. We will then
come to your desk in order to answer your questions privately. It is very important that you do not talk to your desk neighbor or any other
participant during the whole game. If you do not adhere to this rule, 1, will be deduced for each warning. You will receive the money of this
game in a sealed envelope labeled with your anonymized ID-code in the last week before the summer break. All decisions in this game are, of
course, anonymous.
You play this game together with another randomly selected partner from another school who attends the same grade as you. There are two
different roles in this game. The role of the sender (this is your role) and the role of the receiver (this is the role of your partner). Each of you
receives eight decision sheets which look similar to this one (show the example decision sheet on the projector). The sender (you) must send a
message to the receiver (your partner) about the rolled number. You find the rolled number on the lower part of your decision sheet. It is the
framed number on the very left.
The rolled number on this example sheet is 4 (refer to the example decision sheet). The sender can send one of the following six messages to
the receiver:
The outcome of the roll of the die is 1.
The outcome of the roll of the die is 2.
The outcome of the roll of the die is 3.
The outcome of the roll of the die is 4.
The outcome of the roll of the die is 5.
The outcome of the roll of the die is 6.
Please tick the respective box in the upper part of your decision sheet depending on which message you choose. The mes-sage which you
send to your partner is the only information he or she receives, i.e. he/she does not know the actual number rolled or the possible payment
options. We will cut off the lower part of the decision sheet. Your partner only receives the upper part of the sheet. Thus, the receiver will never
know whether the message you sent him/her corresponds to the actual outcome of the die roll. This also means that we do not inform the receiver
about your payoff. He/she just receives the amount of money his/her decision yields.
Your partner then must choose a number from one to six. The choice of your partner determines your and your partners payoff.

20

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

If your partner chooses the number which was actually rolled, both of you receive the payments from the LEFT side of the lower part of
your decision sheet. In the example (refer to the example decision sheet) this would be the number 4. If your partner chooses number 4
in this example, then you and your partner will receive 10 points each.
If your partner chooses one of the other five numbers, both of you receive the payments from the RIGHT side of the lower part of your
decision sheet. In the example (refer to the example decision sheet) this would be the numbers 13 as well as 5 and 6. If your partner
chooses one of these five numbers in this example, then you and your partner will receive 20 points each.

One point is worth 10/40 cents (adapt this information to age cohort of participants).
We ask you to complete 8 such decision sheets. Please examine each sheet carefully as they differ from one another with respect to the
payment options for LEFT and RIGHT and the actually rolled number (see boldly framed box). Notice that only one of the eight decision sheets
will be relevant for payment. For determining the relevant sheet, one pupil from your class will draw a card from these eight cards (show cards)
from this bag (show bag). Depending on which number was drawn, you and your partner will be paid the respective decision. Please note that the
receiver has to take a decision on each sheet and will not be told which sheet was drawn for payment.

A.2. Experimental instructions for the senderreceiver-game RECEIVER (translated from German)
Welcome to our game and thank you for participating. The game will take about two school hours (100 min) in total and you can earn money
in the game. You are asked to make choices during the game and the amount of money you earn is influ-enced by your own decisions. For this
reason it is very important that you properly understand the rules of the game. Please raise your hand if you have any questions. We will then
come to your desk in order to answer your questions in private. It is very important that you do not talk to your desk neighbor or any other
participant during the whole game. If you do not adhere to this rule, 1, will be deduced for each warning. You will receive the money of this
game in a sealed envelope labeled with your anonymized ID-code in the last week before the summer break. All decisions in this game are, of
course, anonymous.
You play this game together with another randomly selected partner from another school who attends the same grade as you. There are two
different roles in this game. The role of the receiver (this is your role) and the role of the sender (this is the role of your partner). Your partner
received eight decision sheets which looked similar to this one (show the example decision sheet on the projector). The sender (your partner) had
to send a message to the receiver (you) about the rolled number. It is the framed number on the very left in the lower part of the decision sheet.
The rolled number on this example sheet is 4 (refer to the example decision sheet). The sender sent you one of the following six messages:

The outcome of the roll of the die is 1.

The outcome of the roll of the die is 2.
The outcome of the roll of the die is 3.
The outcome of the roll of the die is 4.
The outcome of the roll of the die is 5.
The outcome of the roll of the die is 6.
Your partner ticked the respective box in the upper part of the decision sheet depending on which message he/she chose. The message which
your partner has sent to you is the only information you receive about the actually rolled number. We have cut off the lower part of the decision
sheet. You will never know whether the message corresponds to the actual out-come of the die roll. The sender knows exactly which information
you did and did not receive.
Now you must choose a number from one to six and tick the respective box. Your choice determines your and your partners payoff.

If you choose the number which was actually rolled, both of you receive the payments from the LEFT side of the lower part of your
decision sheet (we have cut away this part). In the example (refer to the example decision sheet) this would be the number 4. If you
choose the number 4 in this example, then you and your partner will receive 10 points each.

If you choose one of the other five numbers, both of you receive the payments from
the RIGHT side of the lower part of your decision sheet. In the example (refer to the
example decision sheet) this would be the numbers 13 as well as 5 and 6. If you

choose one of these five numbers in this example, then you and your partner will
receive 20 points each. Note that the payoffs in this example are just for illustration
purposes. For making your decisions, you wont see any number of points associated
with a particular decision!

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

21

One point is worth 10/40 cents (adapt this information to age cohort of participants).
We ask you to complete 8 such decision sheets per sender, and that you complete the sheets from as many senders as you will get during the
experiment. Notice that, for each sender, only one of the eight decision sheets is relevant for payment. We already determined the relevant
decision sheet for each of your partners at random, but we do not tell you which decision sheet counts. You will receive the money from each of
your partners relevant decision sheet. You receive your money during the next week in a sealed envelope. You will only learn about your own
earnings but not about the payoff of your partners (senders).

A.3. Control questions: senderreceiver game (translated from German)

22

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

23

A.4. Sample decision sheet: senderreceiver game (translated from German)

Appendix B. Instructions and decision sheets of the risk task

B.1. Experimental instructions for the risk task (translated from German)
Now we would like to play a second game with you in which you also can earn money.
Each of you receives ten tokens. Each token is worth 10/40 cents (adapt this information to age cohort of participants). This means that all
your tokens together are worth 1 /4 . You can decide how many of these tokens you want to invest in a lottery. You can win extra money in this
lottery, but you can also lose the money you invest. At the end of the game we there-fore conduct a gamble after each of you decided how many
tokens to invest.
For this purpose I will toss a coin (show coin). The coin has a blue dot on the one side and red dot on the other. If the red side is tossed you
win and if the blue side is tossed you lose. If you win, you receive two extra token for each token you invested in the lottery. That is, you keep all
your invested tokens and win two more tokens for each token invested if red is tossed. If blue is tossed, however, you lose your invested tokens.
Each token which you do not invest will be paid out for sure.

24

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

You can easily compute your total earnings in this game. If red is
tossed (you win): 10 + invested tokens 2.
If blue is tossed (you win): 10
invested tokens.
Lets go through an example. If you invest 1 token in the lottery and keep the other 9 tokens, how many tokens will you earn if blue is
tossed? . . . In this example you lost your invested tokens. However, you did not invest 9 tokens and thus you earn 10 1 = 9 tokens in total. And
what happens if you win? . . . Then you get back your invested token, that is, you have 10 tokens again (the 9 tokens which you did not invest
plus the one you get back from the lottery) and you get 2 extra tokens for the invested token. You earn 10 + 1 2 tokens in total. This was only an
example.
Lets go through another example. If you do not invest any tokens in the lottery, how many tokens will you earn if red is tossed? 10 (10 + 0
2). And how many tokens will you earn if blue is tossed? 10 (10 0).
Lets go through a final example. Assume that you invest 3 tokens in the lottery. How many tokens will you earn if blue is tossed? 10 3 = 7.
And how many tokens will you earn if red is tossed? 10 + 3 2 = 16.
We now hand out a sheet which helps you to calculate how many tokens you earn if you invest from 0 to 10 tokens in the lottery and if blue
respectively red is tossed.
We now hand out the decision sheets. In the middle of the sheet there are 11 boxes. Below each box you see the number of tokens which you
can choose to invest in the lottery (that is 0, 1, 2, 3, . . ., 10). Please tick the box which represents the number of tokens you want to invest in the
lottery. Please only tick one box. Please raise your hand if you ticked the box so that we can collect the decision sheet. No other child will know
your decision. It is a secret.
B.2. Sample decision sheet: risk task (translated from German)

References
Alms, I., Cappelen, Srensen, E. ., & Tungodden, B. (2010). Fairness and the development of inequality acceptance. Science, 328, 11761178.
Angelova, V., & Regner, T. (2013). Do voluntary payments to advisors improve the quality of financial advice? An experimental deception game. Journal of Economic Behavior
and Organization, 93, 205218.
Aoki, K., Akai, K., & Onoshiro, K. (2010). Deception and confession: Experimental evidence from a deception game in Japan. Osaka University Discussion Paper No. 786.
Astington, J. W. (1993). The childs discovery of the mind. Cambridge: Harvard University Press.
Azar, O., Yosef, S., & Bar-Eli, M. (2013). Do customers return excessive change in a restaurant? A field experiment on dishonesty. Journal of Economic Behavior and
Organization, 93, 219226.
Bucciol, A., Landini, F., & Piovesan, M. (2013). Unethical behavior in the field: Demographic characteristics and beliefs of the cheater. Journal of Economic Behavior and
Organization, 93, 248257.
Bucciol, A., & Piovesan, M. (2011). Luck or cheating? A field experiment on honesty with children. Journal of Economic Psychology, 32(1), 7378. Bussey, K. (1992).
Lying and truthfulness: Childrens definition, standards and evaluative reactions Child Development, 63(1), 129137. Cappelen, A., Srensen, E. ., & Tungodden, B.
(2013). When do we lie? Journal of Economic Behavior and Organization, 93, 258265.
Charness, G., & Gneezy, U. (2010). Portfolio choice and risk attitudes: An experiment. Economic Inquiry, 48(1), 133146. Childs, J.
(2011). Gender differences in lying. Economics Letters, 114(2), 147149.
Currie, C., Zanotti, C., Morgan, A., Currie, D., De Looze, M., Roberts, C., Samdal, O., Smith, O.R.F., & Barnekow, V. (2012). Social determinants of health and well-being among
young people. Health behaviour in school-aged children (HBSC) study: International report from the 2009/2010 survey. Copenhagen, WHO Regional Office for Europe
(Health Policy for Children and Adolescents, No. 6).
Danilov, A., Biemann, T., Kring, T., & Sliwka, D. (2013). The dark side of team incentives: Experimental evidence on advice quality from financial service professionals. Journal
of Economic Behavior and Organization, 93, 266272.
DePaulo, B. M., Kashy, D. A., Kirkendol, S. E., Wyer, M. M., & Epstein, J. A. (1996). Lying in everyday life. Journal of Personality and Social Psychology, 70(5), 979995.
Dreber, A., & Johannesson, M. (2008). Gender differences in deception. Economics Letters, 99(1), 197199.
Dr, W., Griebler, R., & Kremser, W. (2010). HBSC Factsheet Nr. 1/10: Rauchen, Alkohol, Cannabis. Update 2010. Wien: Bundesministerium fr Gesundheit. Ederer, F., & Fehr,
E. (2007). Deception and incentives: How dishonesty undermines effort provision. IZA Discussion Paper 3200.
Erat, S. (2013). Avoiding lying: The case of delegated deception. Journal of Economic Behavior and Organization, 93, 273278. Erat, S., &
Gneezy, U. (2009). White lies. Working Paper.
Erat, S., & Gneezy, U. (2012). White lies. Management Science, 58(4), 723733 .

D. Gltzle-Rtzler, P. Lergetporer / Journal of Economic Psychology 46 (2015) 1225

25

Farrington, D. P., Ttofi, M. M., & Coid, J. W. (2009). Development of adolescence-limited, late-onset, and persistent offenders from age 8 to age 48. Aggressive Behavior, 35(2),
150163.
Fehr, E., Bernhard, H., & Rockenbach, B. (2008). Egalitarianism in young children. Nature, 454, 10791083.
Fehr, E., Gltzle-Rtzler, D., & Sutter, M. (2013). The development of egalitarianism, altruism, spite and parochialism in childhood and adolescence. European Economic Review,
64, 369383.
Gervais, J., Tremblay, R. E., Desmarais-Gervais, L., & Vitaro, F. (2000). Childrens persistent lying, gender differences, and disruptive behaviours: A longitudinal perspective.
International Journal of Behavioral Development, 24(2), 213221.
Gervais, J., Tremblay, R. E., & Hroux, D. (1998). Boys lying and social adjustment in pre-adolescence: Teachers, peers and self-reports. Criminal Behaviour and Mental
Health, 8(2), 127138.
Gibson, R., Tanner, C., & Wagner, A. F. (2012). Preferences for truthfulness: Heterogeneity among and within individuals. American Economic Review, 103(1), 532548.
Gino, F., Ayal, S., & Ariely, D. (2013). Self-serving altruism? The lure of unethical actions that benefit others. Journal of Economic Behavior and Organization, 93, 285292.
Gneezy, U. (2005). Deception: The role of consequences. American Economic Review, 95(1), 384394.
Gneezy, U., & Potters, J. (1997). An experiment on risk taking and evaluation periods. Quarterly Journal of Economics, 112(2), 631645. Gneezy, U., Rockenbach,
B., & Serra-Garcia, M. (2013). Measuring lying aversion. Journal of Economic Behavior and Organization, 93, 293300. Hurkens, S., & Kartik, N. (2009). Would I
lie to you? On social preferences and lying aversion. Experimental Economics, 12(2), 180192.
Ismayilov, H., & Potters, J. (2013). Disclosing advisors interests neither hurts nor helps. Journal of Economic Behavior and Organization, 93, 314320. Jensen, L. A., Arnett, J. J.,
Feldman, S. S., & Cauffman, E. (2004). The right to do wrong: Lying to parents among adolescents and emerging adults. Journal of
Youth and Adolescence, 33(2), 101112.
Lee, K., & Ross, H. J. (1997). The concept of lying in adolescents young adults: Testing Sweetsers folkloristic model. Merrill-Palmer Quarterly, 43(2), 255270. Lightle, J. P.
(2013). Harmful lie aversion and lie discovery in noisy expert advice games. Journal of Economic Behavior and Organization, 93, 347362. Loewen, P. J., Dawes, C. T., Mazar,
N., Johannesson, M., Koellinger, P., & Magnusson, P. K. E. (2013). The heritability of moral standards of everyday
dishonesty. Journal of Economic Behavior and Organization, 93, 363366.
Mitchell, S., & Rosa, P. (1981). Boyhood behavior problems as precursors to criminality: A 15-year follow-up study. Journal of Child Psychology and Psychiatry and Allied
Disciplines, 22(1), 1933.
Moolchan, E. T., Ernst, M., & Henningfield, J. E. (2000). A review of tobacco smoking in adolescents: Treatment implications. Journal of the American Academy of Child and
Adolescent Psychiatry, 39(6), 682693.
Pock, M., Czypionka, T., Mllbacher, S., & Schnabl, A. (2008): Volkswirtschaftliche Effekte des Rauchens Eine konomische Analyse fr sterreich. IHS research report.
Pulkkinen, L. (1983). Search for alternatives to aggression in Finland. In A. P. Goldstein & M. Segall (Eds.), Aggression in global perspective (pp. 104144). New York:
Pergamon Press.
Rutter, M., Tizard, J., & Whitemore, K. (1970). Education, health and behavior. London: Longman.
Snchez-Pags, S., & Vorsatz, M. (2007). An experimental study of truth-telling in a senderreceiver game. Games and Economic Behavior, 61(1), 86112. Selman, R. L. (1971).
Taking anothers perspective: Role-taking development in early childhood. Child Development, 42, 17211734.
Stewart, M. A., & DeBlois, C. S. (1985). Diagnostic criteria for aggressive conduct disorder. Psychopathology, 18(1), 1117. StouthamerLoeber, M. (1986). Lying as problem behavior in children: A review. Clinical Psychology Review, 6, 267289 . Stouthamer-Loeber, M., &
Loeber, R. (1986). Boys who lie. Journal of Abnormal Child Psychology, 14(4), 551564.
Sutter, M. (2009). Deception through telling the truth?! Experimental evidence from individuals and teams. Economic Journal, 119(534), 4760.
Sutter, M., Feri, F., Kocher, M.G., Martinsson, P., Nordblom, K., & Rtzler, D. (2010). Social preferences in childhood and adolescence: A large-scale experiment. IZA Discussion
Paper 5016.
Warr, M. (2007). The tangled web: Delinquency, deception and parental attachment. Journal of Youth and Adolescence, 36(5), 607622.