Students Cheat More: Comparing the Dishonesty of a Student Sample and a Representative Sample in the Laboratory*

DOIhttp://doi.org/10.1111/sjoe.12326
Date01 January 2020
Published date01 January 2020
Scand. J. of Economics 122(1), 257–279, 2020
DOI: 10.1111/sjoe.12326
Students Cheat More: Comparing the
Dishonesty of a Student Sample and a
Representative Sample in the Laboratory*
Toke Reinholt Fosgaard
University of Copenhagen, 1958 Frederiksberg C, Denmark
tf@ifro.ku.dk
Abstract
Unethical behavior has been found in numerous experiments, yet mainly among university
students. The use of student participants is potentially problematic for generalizability and
the resulting policy recommendations. In this paper, I report on an experiment with potential
dishonesty.The experiment was completed by a representative non-student sample and a student
sample. The results show that cheating does exist, but also that students cheat systematically
more. This suggests that focusing on students as participants tends to overestimatethe magnitude
of cheating. I further find that age is an important explanation for this difference in dishonesty.
The older the participants are, the less they cheat.
Keywords: Age differences; cheating; experimental methods; non-students; students
JEL classification:C91; K40
I. Introduction
The study of dishonest behavior has been the focus of a thriving research
field for the last two or three decades. The fact that many people engage in
dishonesty is well documented in this literature, although typically people
do not cheat at the maximum possible level (Gneezy, 2005; Mazar et al.,
2008; Shu and Gino, 2012; Shu et al., 2012; Fischbacher and F¨ollmi-Heusi,
2013; Fosgaard et al., 2013; Pascual-Ezama etal., 2015; Abeler et al., 2016;
Jacobsen et al., 2017).
The main approach to studying the temptation of dishonesty has been
to use student samples. Using such samples is an obvious and convenient
choice for the documentation of dishonesty. The existence of dishonesty
can be proved in any sample, but students are conveniently at hand. Student
samples have been used not only to document dishonesty per se, but also
to explore a long array of mechanisms that are pivotal for the decision to
behave dishonestly or not. Dishonesty is, for instance, influenced by peer
*I am grateful for the useful comments provided by Alessandro Bucciol, Lars Gaarn Hansen,
Torben Heien Nielsen, and Marco Piovesan. Furthermore, the financial support of the Danish
Agency for Science, Technology and Innovation (10-093637) is deeplyappreciated.
C
The editors of The Scandinavian Journal of Economics 2018.
258 Students cheat more
effects (Fosgaard et al., 2013), the possibility of justifying one’s actions
(Shalvi et al., 2012), and being (primed to be) creative (Gino and Ariely,
2012), to mention a few examples.
Although this body of literature has already provided us with a massive
number of insights, an obvious question is how sensitive these findings
are to the fact that students are the primary source of participants (Falk
and Heckman, 2009). It has been argued that students are a very particular
type of sample and that extrapolation of student data to non-students might,
therefore, be problematic (Henrich et al., 2010). Such extrapolation becomes
particularly critical if student-based insights are generalized and translated
into recommendations about how to prevent dishonesty. Only if the evidence
collected from student populations is predictive for the population as a
whole can we rely on these insights when formulating advice to policy-
makers in governments and private enterprises.
A recent approach to addressing this concern has been to move away
from student participants to study dishonesty in field experiments with non-
students as participants (Shu et al., 2012; Azar et al., 2013; Abeler et al.,
2014; Cohn et al., 2014). With the exception of Abeler et al. (2014), this
stream of literature asserts that dishonesty also takes place in field settings.
Although these findings are exceptionally interesting, we still do not know,
strictly speaking, whether the findings from the main part of the literature
(i.e., the laboratory experiments) also apply to non-students. This is because
these initiatives to study the dishonesty of non-students have been in natural
field settings and have therefore changed two elements simultaneously: the
subject pool and the experiment context (field/laboratory). Until now, no
study has disentangled the effect of changing the sample from the effect of
being in the field or laboratory.
My approach in this paper is to make a contribution by comparing the
dishonesty of students and non-students when only one of the two changes
is made: I compare the behavior of students and non-students in the same
laboratory setting. By conducting the study in the laboratory, I ensure that
the choice environment remains equal to the main part of the existing body
of literature and that I merely compare the effect of the samples. This
comparison benefits from the fact that I can perform a selection analysis
on all invited non-students with the use of third-party register data.
I use a modified die rolling experiment (Fischbacher and F¨ollmi-Heusi,
2013). I find that dishonest reporting of the outcome of the die clearly
exists, and also that students are systematically more dishonest than non-
students. Interestingly, I am also able to relate this tension of behavior
to underlying determinants. Hence, I can explore the potential causes of
dishonesty. Through econometric assessments, it turns out that age is an
important explanation of the difference. The older the participants are, the
less they over-report. Thus, the high instance of cheating among students
C
The editors of The Scandinavian Journal of Economics 2018.

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT