Sociology 612-301   Categorical Data Analysis
Fall 2004

Paul D. Allison
276 McNeil
898-6717
allison@soc.upenn.edu
www.ssc.upenn.edu/~allison

Ofc. Hours: Tues., Wed. 1:30-2:30, or by appointment

Lectures: Tues. and Thurs. 9-10:30

Content. This course deals with techniques for analyzing multivariate data in which the dependent variable is a set of categories (a dichotomy or polytomy). Topics will include linear probability models, logit (logistic) regression models, probit models, logit analysis of contingency tables, cumulative logit and probit (for ordinal data), multinomial logit, conditional logit (discrete choice), unobserved heterogeneity, log-linear models, square tables, response-based sampling, and repeated measures.

Prerequisites. A course on linear regression analysis such as Soc. 536, Stat. 102, Stat. 112, or Econ. 6. Experience with applications of regression analysis will be helpful, but not essential. No knowledge of matrix algebra or calculus is needed.

Texts.
Required:

Paul D. Allison, Logistic Regression Using the SAS System: Theory and Application. Available at the U. Bookstore.
Lecture notes available on  Blackboard site. Recommended: J. Scott Long, Regression Models for Categorical and Limited Dependent Variables. Exam. One exam near the end of the semester. There will be two weeks advance notice.

Assignments. There will be approximately six assignments using SAS to analyze data furnished by the instructor.  Students will also be required to read several research articles and write 1-page critiques.

Paper. Students must turn in a research paper on or before December 14. The typical paper will report on analysis of some data set chosen in consultation with the instructor. Also acceptable is a mathematical analysis or computer simulation of some statistical procedure. The paper should consist of at least 10 pages of text, not including tables, notes and references. Collaborative efforts are welcome.

Grading. Final grades will depend about 1/3 on exam, 1/3 on paper, and 1/3 on homework assignments.