Logit and Probit Models
Prof. Marco Steenbergen, University of Bern
21 July - 1 August (two week course / 35 hrs)
Marco Steenbergen is Professor of Political Methodology at the University of Zurich, Switzerland. He earned his MA at the University of Amsterdam and his PhD at Stony Brook University, New York. He previously taught at Carnegie Mellon University, the University of North Carolina, Chapel Hill, and the University of Bern. His publications have appeared in the American Political Science Review, the American Journal of Political Science, Political Analysis, and numerous other journals and edited volumes. He is co-editor of European Integration and Political Conflict and co-author of Deliberative Politics in Action (both Cambridge University Press) and The Ambivalent Partisan (Oxford University Press). His major research interests are electoral behaviour, public opinion, deliberative politics, and quantitative research methods with a thematic focus on the United States and Western Europe.
- This course covers statistical models for the analysis of categorical dependent variables. This includes binary logit and probit models, ordinal logit and probit models, as well as conditional and multinomial logit models.
- Categorical dependent variables are extremely common in the social sciences and, consequently, models for analyzing these variables are now standard research tools. By the end of this course, students will be able to read articles that employ logit and probit models. In addition, they will be able to use these models in their own research. To make this possible, the course balances statistical theory with practical skills. The practical skills include hands-on analysis of categorical data in Stata, as well as extensive training in the interpretation of logit and probit results.
- The course assumes familiarity with the following subjects.
- (1) Mathematics: algebra, exponents and logarithms, functions, and basic calculus (students should understand the meaning of derivatives and their usage in finding minimums and maximums of functions).
(2) Probability theory: the concept of probability, random variables, probability calculus, probability density functions, cumulative distributions, and expectations.
(3) Regression analysis: multiple regression model, ordinary least squares (OLS), inference about the regression model, and interpretation of regression co-efficients.
(4) Stata: opening Stata data files, saving Stata data files, creating log files, and generating/recoding variables.
- Students who are uncertain about their mathematical background are encouraged to consult the following:
- Kleppner, Daniel, and Norman Ramsey. 1985. Quick Calculus: A Self-Teaching Guide. New York: Wiley. 2nd edition.
- Students who are uncertain about their background in probability theory and/or regression analysis are encouraged to consult any introductory econometrics text. An excellent text is:
- Kmenta, Jan. 1997. Elements of Econometrics. Ann Arbor, MI: University of Michigan Press. 2nd edition.
Representative Backround Reading
- Long, J. Scott, and Jeremy Freese. 2006. Regression Models for Categorical Dependent Variables Using Stata. College Station, TX: Stata Press. 2nd edition.
- Steenbergen, Marco R. 2010. Categorical Dependent Variables. Berne: University of Berne.
- The following text is recommended for students who have not had much prior exposure to Stata: Acock, Alan C. 2010. A Gentle Introduction to Stata. College Station, TX: Stata Press. 3rd edition