The course will provide a preview of the econometrics questions that arise when analyzing models with multiple decision makers interacting simultaneously. This simultaneous causality creates challenges to standard approaches to identification and estimation. We first, in a simple bivariate binary game, underscore how observed data can be linked to response functions, and how assumptions about what agents know, their beliefs and how they play impact the kinds of learning that one can do with these data. Given the identification results, we will then provide approaches to statistical estimation of such models and the challenges that come up in building confidence regions. We consider complete and incomplete information games and also observational and experimental data. We will also provide empirical illustrations where practical and computational issues are studied.