After showing why ordinary regression analysis is not appropriate for investigating dichotomous or otherwise 'limited' dependent variables, this volume examines three techniques which are well suited for such data. It reviews the linear probability model and discusses alternative specifications of non-linear models.