[gov3009-l] Applied Statistics 4/11

Hi everyone! This week at the Applied Statistics Workshop we will be welcoming *Michael Windzio*, Professor of Sociology at the University of Bremen. He will be presenting work entitled *Does schoolwork cooperation improve pupils’ grades and well-being in school? Results from social network and propensity score analysis*. Please find the abstract below and on the Applied Stats website here <https://projects.iq.harvard.edu/applied.stats.workshop-gov3009>. As usual, we will meet at noon in CGIS Knafel Room 354 and lunch will be provided. See you all there! -- Dana Higgins *Title:* *Does schoolwork cooperation improve pupils’ grades and well-being in school? Results from social network and propensity score analysis * *Abstract:* Using panel data of school-class networks and outcomes of 11-13-year-old students, effects of collaboration in schoolwork networks on grades and school-related well-being will be investigated. The analysis might suffer from endogeneity-bias because pupils actively select their peers also with regard to their school-performance. This selectivity will be demonstrated by using p* models for ties in schoolwork-networks at t1 based data of 1,289 pupils in 76 classrooms. Predictions from this model will be used to generate propensity scores. Stochastic actor-based models (SOAM) for the co-evolution of networks and behavior/attitudes (N=244, k= 10) result in a systematic loss of data, whereas propensity score matching appropriately limits the data to the area of common support. However, violation of the SUTVA requires that indicators of network embeddedness are controlled, which can be done in a propensity score weighting regression. Overall, results of SOAMs and propensity score matching suggest that schoolwork networks do not have significantly positive effects, neither on grades nor on well-being.