Statistics 501/Psychology 612: 

Introduction to Nonparametrics

and Log-Linear Models

Paul R. Rosenbaum

Professor, Statistics Department, Wharton School

   

Description

 

Statistics 501/Psychology 612 is a second course in statistics for PhD students in the social, biological and business sciences.  It covers nonparametric statistical methods for data that are not Normally distributed and methods for discrete data, such as log-linear and logit models.  Students should have taken an undergraduate course in statistics prior to Statistics 501.  Statistics 500 is suggested but not required as background.

 

Course Materials

 

The texts are Nonparametric Statistical Methods by M. Hollander and D. Wolfe, second edition, John Wiley and Sons, and S. Fienberg, The Analysis of Cross-Classified Categorical Data.  There is also a bulkpack consisting of photocopies of computerized data analysis which is available from Wharton Reprographics.  Students make extensive use of the computer to analyze data.  There is a midterm on nonparametrics, a final on discrete data, and 5-page report on an analysis of data you select.  In addition, there are ungraded homeworks.

 

Topics

 

Nonparametrics:  Nonparametric tests, estimates, and confidence intervals, nonparametric analysis of variance, and correlation.

 

Discrete data:  Log-linear, logit and ordinal logit models.