Institution: University of Hamburg
Lecturer: Shawna N. Smith, Ph.D. (University of Michigan, USA)
Date: June 22-25, 2015
Place: University of Hamburg
Language of instruction: English
Contents: Many variables of interest to social, political and behavioral scientists are non-continuous, either through nature or through measurement. Outcomes like vote choice, social class, condom use, and/or number of Facebook friends necessarily violate key assumptions of the simple linear regression framework and require other model estimation strategies. Although advances in software have made estimation of these models trivial, model non-linearities make post-estimation interpretation difficult and require investigators to make choices about which aspects of the data space best represent underlying social dynamics.