POL S 510 A: Maximum Likelihood Methods for the Social Sciences

Most social science data do not fit the assumptions of the linear regression model taught in introductory statistics courses. For example, social science data often consist of discrete categorizations or counts of events, rather than continuous outcomes. Observations may be correlated across periods, as in time series, or clustered into correlated groups, violating the linear regression assumption of independence. In this course, students will learn how use the method of maximum likelihood to derive statistical models that suit the particular behavior of their social science data and how to clearly communicate the substantive import of their findings to a broad audience. In the process, students will gain familiarity with basic statistical programming in R, a free and increasingly popular language. Topics of special interest to students will be covered as time permits.