VoxEU Column Frontiers of economic research

Self-control and peer groups: An empirical analysis

Peers and role models play a key role in the choices young people make. This column discusses research suggesting that students who are part of a social circle have more self-control than those who are alone, and the larger this social circle, the greater the self-control. However, having peers who are too similar can be detrimental to self-control.

Peers and role models play a key role in young people’s choices, in particular those related to risky behaviours such as drinking, smoking (Baldwin 2008), drug use, sexual activity (Fernandez-Villaverde et al. 2012), and habits such as procrastination. In fact, people with self-control or addiction problems often seek relief in self-help groups such as Alcoholic Anonymous, Narcotics Anonymous, and Gamblers Anonymous. In these groups, people with similar issues share experiences and develop coping skills. These organisations are intended to build emotional support and a sense of belonging. They also aim to provide meaningful role models and other benefits that professionals may not be able to offer. 

Interestingly, members of self-help groups do not abide by contracts, binding implicit agreements, or any other standard commitment device. The programmes of these organisations explicitly state that their aim is not to control the behaviour of their members, but to foster informational interactions among members. For instance, in “What does Alcoholics Anonymous (A.A.) do?”, the organisation states that “A.A. members share their experience with anyone seeking help with a drinking problem.” Battaglini et al. (2005) develop a theory to explain how and when observing the behavior of others can be beneficial for overcoming self-control problems (as with support groups). They also explore how and when groups can be detrimental, as is often the case among schoolmates or neighbourhood youths.

A theory of social influences on self-control

In general, economists treat the existence of peer influences as a starting point: instead of explaining the origins of the peer effects, researchers focus on studying their implications for individual choices.1 An exception is the paper by Battaglini et al. (2005), who propose a theory of social influences on self-control. This paper addresses the following question: In the presence of self-control problems, is it beneficial to be exposed to the choices made by another player? According to their theory, people care about other’s actions because of their informative content. In other words, peers learn how to overcome self-control issues by observing each other’s behavior.

The advantage of Battaglini et al.’s (2005) approach is that it provides testable predictions on the nature of the peer effects. Two theoretical predictions emerge from the analysis. First, observing how another individual deals with impulses and temptation can be beneficial or detrimental. Second, when peers are heterogeneous in their expected levels of self-control, they generally value the ‘quality’ of their peers non-monotonically. The ideal friend is someone with a slightly worse self-control problem than one’s own: this makes one’s own successes more encouraging and one’s own failures less discouraging. Intuitively, there is little benefit from having a friend who is too self-controlled or a friend who is too impulsive. This fact is in contrast with the standard assumption that peer effects are monotonic and, indeed, linear in other individuals’ behaviour (i.e. the more, the better).

A recent literature empirically studies the link between self-control and peer effects following Battaglini et al.’s (2005) theoretical work. Early contributions in this literature have focused on the possibility of contagion of self-control problems through peer effects (Christakis and Fowler 2007, 2008, Cohen-Cole and Fletcher 2008). A different approach is used by Buechel et al. (2014), who rely on laboratory experiments to explicitly study Battaglini et al.’s (2005) predictions.

Similar is good, not too similar is optimal

In this column, we present the results of a recent paper where we use data from the National Longitudinal Survey of Adolescent to Adult Health (Add Health) to test the prediction of this theory (Battaglini et al. 2017). This panel is an ideal data set to test this theory since it provides information on self-control both for the individuals and for their friends.2 Our analysis exploits the longitudinal nature of the Add Health panel, which follows six cohorts of individuals and their friends from adolescence to adulthood (from Grades 7-12 through ages 24-30) from the years 1994-1995.  

The main challenge encountered in empirical studies on peer effects is the endogeneity stemming from the fact that individuals choose their peers as well as their actions.3 Our paper investigates whether a student’s ability to deal with her own impulses changes when the student’s social circle is affected by the loss of an older friend who terminates the school programme. Graduation/termination of the school programme, therefore, provides an exogenous shock to the social circle of the other members in lower grades who remain enrolled.

According to our results, a decrease in the number of friends is associated with a decrease in the ability to resist temptations. To be more precise, adolescents who lose 50% or more of their friends are expected to experience, on average, a decrease in the level of self-control of 11% relative to the sample mean. Moreover, we find that those with the least self-control are those who lose the most friends. The study also concludes that students with less self-control than their peers are less likely to control their impulses when faced with an exogenous increase in the self-control of peers, as predicted by Battaglini et al.’s (2005) theoretical model. Finally, our results confirm that the interactions with peers that have similar impulses is important to increase individual self-control. However, the relationship between individual and peers’ (average) self-control is non-monotonic. The estimates reveal that for adolescents with a percentage of similar peers below 82%, a 10% increase in the similarity of peers translates, on average, into a 0.9% increase in self-control. For adolescents with a percentage of similar peers above 82%, a 10% increase in the similarity of peers translates instead into a 2.9% reduction in self-control. In other words: similar is good, but not too similar is optimal.


We find evidence consistent with the two key predictions of the theory by Battaglini et al. (2005) regarding the relationship between an agent's expected self-control and the size and composition of his or her social circles.

  • Students embedded in social circles have more self-control than those who are alone, and their self-control is increasing in the size of their social group.
  • Students’ self-control is, however, a non-monotonic hump-shaped function of the average self-control of their friends.

These findings suggest that peer groups may play an important role in affecting self-control problems; but the effect of the ‘quality of peers’ (in terms of their self-control) is non-monotonic.

Naturally, this paper is only a first step in characterising the relationship between individual and peers’ abilities to control their own impulses. Nevertheless, our analysis uncovers non-linearities in peer influences that have thus far been neglected by the large literature on peer effects.


Baldwin, R (2008), “Social interactions and smoking”,, 31 May. 

Battaglini, M, R Bénabou and J Tirole (2005): “Self-Control in Peer Groups,” Journal of Economic Theory, 123 (2), 105-134.

Battaglini, M, C Díaz and E Patacchini (2017): “Self-Control and Peer Groups: An Empirical Analyisis,” Journal of Economic Behavior and Organization, 134, 240-254, February.

Buechel, B, L Mechtenberg and J and Petersen (2014): “Peer Effects and Students’ Self- Control,” MPRA Paper 53658, University Library of Munich, Germany.

Christakis, N A and J H Fowler (2007): “The Spread of Obesity in a Large Social Network over 32 Years,” The New England Journal of Medicine, 357, 370-379.

Christakis, N A and J H Fowler (2008): “The Collective Dynamics of Smoking in a Large Social Network,” The New England Journal of Medicine, 358, 2249-2258.

Cohen-Cole, E and J M Fletcher (2008): “Is Obesity Contagious? Social Networks vs. Environmental Factors in the Obesity Epidemic,” Journal of Health Economics, 27 (5), 1382-1387, September.

Fernández-Villaverde, J, N Guner and J Greenwood (2012), “Shame, peer effects, and sexual behaviour”,, 19 January. 

Jackson, M O (2014): “Networks and the Identification of Economic Behaviors,” SSRN Working Paper 2404632.

Jackson, M O, B W Rogers and Y Zenou (2014): “The Impact of Social Networks on Economic Behavior,” SSRN Working Paper 2467812.

Jackson, M O and Y Zenou (2015): “Games on Networks,” In: P. Young and S. Zamir (Eds.), Handbook of Game Theory, Vol. 4, Amsterdam: Elsevier Publisher, 91-157.

Nagin, D S and G Pogarsky (2004): “Time and Punishment: Delayed Consequences and Criminal Behavior,” Journal of Quantitative Criminology, 20 (4), 295-317.

Perrone, D, C J Sullivan, T C  Pratt and S Margaryan (2004): “Parental Efficacy, Self-Control, and Delinquency: a Test of a General Theory of Crime on a Nationally Representative Sample of Youth,” International Journal of Offender Therapy and Comparative Criminology, 48 (3), 298-312.


[1] See Jackson and Zenou (2015) for a recent survey.

[2] Add Health’s Wave I and Wave II in-home questionnaires contain the standard question that is used to measure self-control or willpower (Nagin and Pogarsky 2004, Fletcher et al. 2009): “When making decisions, you usually go with your ‘gut feeling’ without thinking too much about the consequences of each alternative?”.

[3] For surveys of the recent empirical literature that incorporates network analyses into studies of peer effects and behavior, see Jackson (2014) and Jackson et al. (2014).

1,575 Reads