Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
We propose a novel empirical strategy to distinguish two sources of peer effects in risky choices: (i) assortative matching in the formation of the peer relationship, and (ii) social interaction effects arising in the choice situation. We combine network information on existing peer relationships with controlled randomised exposure to social interaction conditions. We apply this strategy in a field experiment on risky decision making among adolescents between ages 13 and 15. School classes were randomly allocated to two social interaction treatments. Students were allowed to discuss their choices with a natural peer – either a friend or a randomly selected classmate – before individually making choices in an incentivised lottery task. In the control group, adolescents made choices without being able to discuss them with a peer. In a dyadic analysis, we find that friends and classmates are matched on socio-demographic characteristics but are not assortatively matched on risk preferences. In contrast, social interaction strongly increases the similarity of teenagers’ risky choices. A large fraction of peers align their choices perfectly.