Chapter 8: Models for Polytomous Responses
SummaryThis chapter generalizes logistic regression models for a binary response to handle a multi-category (polytomous) response. Different models are available depending on whether the response categories are nominal or ordinal. Visualization methods for such models are mostly straightforward extensions of those used for binary responses.
- 8.1. Ordinal response
- 8.2. Nested dichotomies
- 8.3. Generalized logit model
- 8.4. Chapter summary
- 8.5. Lab exercises
Selected figuresview R code
Figure 8.2Latent variable representation of the proportional odds model for m = 4 response categories and a single quantitative predictor, x.
Figure 8.6Effect plot for the effects of Treatment, Sex, and Age in the Arthritis data.
Figure 8.10Fitted probabilities from the models for nested dichotomies fit to the data on women’s labor force participation.
Figure 8.12Fitted probabilities from the generalized logit model fit to the data on women’s labor force participation.