VoxEU Column Frontiers of economic research

Individual versus group decisions: Female versus male honesty

Many nations and corporations strive to raise female membership in decision-making bodies. This column discusses new experimental evidence suggesting that there is more lying (and more extreme lying) in male groups and mixed-gender groups than in female groups. Moreover, group decision-making exacerbates men’s tendency to lie while the opposite is true for women. This suggests that the gender composition of decision-making bodies is important when the goal is to limit the scope of unethical behaviour.

A vast literature has investigated gender differences in a wide variety of economic contexts, both with respect to behaviour and outcomes (Niederle 2014, Croson and Gneezy 2009). Recently, a number of authors have raised the question of whether females behave differently than males in the context of unethical behaviour such as lying. Indeed, some (but not all) studies suggest that males are somewhat more prone to unethical behaviour than females (see e.g. Erat and Gneezy 2012, Childs 2012). As with the bulk of the economic literature, these studies look at individual decision-making. However, in many real world settings, a group of individuals must reach a decision jointly (e.g. in work teams, supervisory boards, or political committees). In fact, there is growing evidence from contexts other than lying that groups often reach decisions in markedly different ways than individuals (Charness and Sutter 2012).

In a recent experiment conducted at the University of Regensburg in June 2014, we investigate how the gender composition of groups affects the degree of unethical behaviour, particularly the extent of lying (Muehlheusser et al. 2014). We build on the simple and widely used die-rolling paradigm introduced by Fischbacher and Föllmi-Heusi (2013). In this setup, individuals are asked to report the private realisation of a die roll, where the respective report determines the subject’s payoff. In particular, the payment that the subject receives corresponds to the declared number of the die roll, except for a declared outcome of six, which leads to a payment of zero. Evidence for lying on the aggregate level is obtained by comparing the actual payoff distribution with the uniform distribution, which would result under-truth-telling. This constitutes the benchmark treatment I.

In a second treatment G, decisions are made jointly in randomly formed groups of two. Hence, groups can be classified as either ‘female’ (only females), ‘male’ (only males), or ‘mixed’ (one female, one male). As a given group received one die and one answer sheet only, they needed to coordinate on both who rolls the die and on which realisation to declare. Group members were aware that each of them would receive the declared payoff.

We find that, when neglecting gender, the behaviour of groups is not different from that of individuals. On average, both groups and individuals report an outcome level well above the no-lying benchmark of 2.5 (individuals: 3.48, groups: 3.47). This masks substantial gender differences, in particular with respect to decision-making in groups.

  • Comparing the average payoffs of individual males and females in Treatment I (3.58 and 3.40, respectively) reveals a slight tendency of males to lie more than females (although this is not statistically significant).
  • This slight tendency is amplified in groups (treatment G), where the average payoff of male groups and mixed groups increases (to 4.00 and 3.71 respectively), while that of female groups decreases (to 2.74).

Gender differences also emerge with respect to the extent of lying, measured in terms of the relative frequencies with which realisations of rolls of 4 and 5 are declared:

  • Comparing the behaviour of male individuals and male groups, the frequencies with which either 4 or 5 are reported are almost identical (0.66 and 0.69, respectively).
  • While similar fractions of individual males report 4 or 5 (respectively 0.34 and 0.32), male groups report 5 substantially more often than 4 (0.54 compared to 0.15).
  • A different picture emerges when comparing female individuals and female groups: female groups are less likely to report 5 than female individuals.
  • Mixed groups behave more like male groups rather than female groups.

To summarise, from the viewpoint of organisational design, when attempting to limit the scope of unethical behaviour, the gender composition of relevant decision-making bodies (such as supervisory boards or political committees) seems to be an important factor.


 Charness, G, and M Sutter (2012): “Groups Make Better Self-Interested Decisions,” Journal of Economic Perspectives, 26(3), 157–176.

Childs, J (2012): “Gender Differences in Lying,” Economics Letters, 114(2), 147–149.

Croson, R, and U Gneezy (2009): “Gender Differences in Preferences,” Journal of Economic Literature, 47(2), 448–474.

Erat, S, and U Gneezy (2012): “White Lies,” Management Science, 58 (4), 723–733.

Fischbacher, U, and F Föllmi-Heusi (2013): “Lies in Disguise: An Experimental Study on Cheating,” Journal of the European Economic Association, 11(3), 525–547.

Muehlheusser, G, Roider, A and N Wallmeier (2014): “Gender Differences in Honesty: Groups Versus Individuals,” CEPR Discussion Paper No 10317.

Niederle, M (2014): “Gender,” NBER Working Paper No 20788.

3,045 Reads