Smoking Behavior and Friendship Formation: The Importance of Time Heterogeneity in Studying Social Network Dynamics

This study illustrates the importance of assessing and accounting for time heterogeneity in longitudinal social network analysis. We apply the time heterogeneity model selection procedure of [1] to a dataset collected on social tie formation for university freshman in the Netherlands by [2]. Within the context of analyzing selection effects for smoking homophily to understand the implications of tobacco policy at a university, we show that failing to account for time heterogeneity yields quite different results substantively from the model arrived at using [1]. While the results are limited by the small scope of the dataset, the paper motivates the testing of time heterogeneity within longitudinal studies of social network behavior and further study of tobacco policy within university settings.