Home / Program

Categorical Data Analysis – 4 Week Course

Instructor: Daniel A. Powers (email: dpowers@mail.la.utexas.edu)

Course Description

This course provides an overview of the major statistical methods and models for categorical dependent variables. We will cover regression-like models for binomial and binary outcomes, contingency tables, individual-level count data, ordered and nominal polytomous response variables, as well as extensions to these models.

Prerequisites

Students should have had a previous course in linear regression and should have some experience using a statistical software package. Examples will be illustrated using Stata, but SAS, SPSS, R can produce similar results.

Course Syllabus

  1. [Week 1]  Introduction

(Readings: Powers & Xie, Ch. 1-2):

    1. why categorical data analysis?
    2. regression-like models and categorical dependent variables
    3. transformational vs. latent variable approaches to CDA
    4. review of linear regression
    5. extending linear regression to categorical responses using GLS
      1. estimation and interpretation of parameters
    1. programming examples
  1. [Week 2] Binary Response Models

(Readings: Powers & Xie, Ch. 3, 5, Handout)

    1. binomial and binary data
    2. logit model
      1. motivation: transformational approach
      2. interpretation: odds, odds ratios, and relative risks.
    1. probit model
      1. motivation: latent variable approach
      2. interpretation: probits, marginal effects
    1. estimation of binary response models
      1. the generalized linear model (GLM)
      2. programming examples
    1. model comparisons and model fit
      1. deviance and chi-square statistics
    1. alternative probability model
      1. complementary log-log
    1. extending binary response models (time permitting)
      1. multilevel models
      2. longitudinal models
      3. Rasch models
    1. miscellaneous topics
      1. censored regression models
      2. selection models
      3. bivariate outcomes
      4. endogenous switching regressions models
  1. [Week 3] Models for Count Data

(Readings: Powers & Xie, Ch. 4, 6,  Handout)

    1. frequency data and count data
    2. frequency data
      1. dissecting a contingency table
      2. odds ratios revisited: measuring association
      3. loglinear models for contingency tables
      4. interpreting parameters from loglinear models
    1. deviance, chi-square, and model fit
    2. programming
    3. models for ordinal data
      1. modeling various forms of association
      2. scaled association models
      3. estimation and interpretation of association parameters
      4. programming
    1. count data
      1. Poisson regression models for counts
      2. estimation, interpretation, and prediction
      3. extensions: zero-inflated Poisson regression, negative binomial regression, extensions
    1. extending models for count data (time permitting)
      1. loglinear models for event-history analysis (proportional hazards models)
      2. multilevel models for hierarchical or repeated measures data
      3. incorporating frailty in demographic models for rates
      4. loglinear models for table standardization
      5. miscellaneous topics
  1. [Week 4]  Models for Ordered and Nominal Response Variables

(Readings: Powers & Xie, Ch. 7, 8, Handout)

    1. ordered responses
      1. types of  logits
        1. cumulative logits and baseline logits
      2. ordered logit and ordered probit models
        1. specifications and parameterizations of ordered response models
      3. proportional odds (PO) assumption
        1. relaxing PO
        2. testing PO
        3. compromises via intermediate models
      4. estimation, interpretation, and prediction from ordered response models
      5. programming examples
    1. nominal responses
      1. types of logits
        1. baseline logits
      2. multinomial logit model
        1. estimation and interpretation of parameters
        2. estimation: as a loglinear model
        3. estimation: using standard programs
      3. conditional logit model
        1. discrete choice models
        2. random utility (latent variable) approach
        3. specification and interpretation of parameters
      4. general multinomial model
      5. specification issues
        1. the IIA assumption
      6. programming examples
    1. sequential responses
        • types of logits:
        1. sequential logit
        2. continuation ratio logit models
        • continuation ratio logit model
        1. specification, estimation and interpretation of parameters
        • programming examples