Spring 2014 Research Methods Short Course
EVENT COUNT ANALYSIS FOR
SOCIAL SCIENCE DATA
Instructor: Professor Tse-Min Lin
Professor of Political Science
University of Texas, Austin
Friday February 14th, 10 A.M. – 4 P.M.
Saturday February 15th, 10 A.M. – 4 P.M.
Classroom: Green 3.206
Course Description: In many social science studies, the dependent variable is an event count, e.g., outbreaks of war in a given year, annual deaths from domestic conflict in a given country, and the number of items a respondent answered correctly in a survey on political knowledge. As an event count, the dependent variable is a non-negative integer which does not follow the normal distribution as required of the classical linear regression model. The estimation of event count models thus requires special techniques. It typically involves choosing an appropriate distribution for the dependent variable as the data generation process and estimating the model with the maximum likelihood method. This course will introduce these special regression models and discuss their assumptions and limitations. It will also include a comparative analysis of event count models and alternative models based on item response theory (IRT).
Tse-min Lin is Associate Professor of Government at the University of Texas at Austin. His teaching and research interests cover methodology, formal theory, and American and comparative political behavior. He has published in American Journal of Political Science, American Political Science Review, Democratization, Issues & Studies, Journal of Democracy, Journal of Politics, Political Analysis, Political Research Quarterly, Public Choice, Social Science History, Taiwan Journal of Democracy, World Politics, as well as in edited volumes. At UT Austin, he teaches several graduate courses, including Advanced Statistical Analysis, Mathematical Methods for Political Analysis, and Time Series Analysis.