I am pleased to announced that Rationality and Society just published a paper I co-authored with Andrea Knecht, titled “Alcohol use among adolescents as a coordination problem in a dynamic network.” In this somewhat unorthodox paper, we use a really cool longitudinal dataset on adolescent networks (collected by Andrea) to test predictions by a simulation model from one of my earlier papers.
This is the abstract:
Whereas most research on substance (ab)use by adolescents studies only the effects of personal networks of adolescents, we propose a theoretical approach that allows for predictions on the effects of the macro-level social network structure on usage rates. We model alcohol use as risk-dominant but inefficient behaviour in a coordination problem, given that adolescents face incentives to align their behaviour with their friends. We propose a game-theoretical model in which actors choose behaviour in a repeated coordination game in an endogenous network. Predictions on levels of alcohol use depending on initial network structure are based on computer simulations of this model. We test the predictions using longitudinal data on alcohol use and friendship choices in school classes in Dutch high schools. We replicate the predicted ‘catalysing’ effect of initial network density on the development of alcohol use but the predicted opposing effect of centralization could not be confirmed.