# Regression for Categorical Data

**Course Content:**This courses explores the use of regression techniques for data that has nominal or ordinal outcome variables. The course includes a review of multiple regression and then focuses on three main categorical techniques: binary logistic, ordinal and multinomial regression.

**Course Objectives:**

- To refresh participants' knowledge of using multivariate statistical methods to analyse data.
- To introduce participants to the use of regression techniques for categorical outcome variables.
- To provide experience of analysing data using SPSS and interpreting the output.

**The topics covered during this course include:**

- Multiple regression
- Simple regression
- Multiple regression - the basics
- Maximising prediction
- Interpreting the co-efficients - Bs versus Betas
- Interacctions
- Nominal IVs
- Warnings
- Running Multiple Regression in SPSS
- Assumptions for testing hypotheses using regression
- Logistic Regression
- Introduction to logistic models
- Notation for logistic models
- Can we fit a linear model to binary variables?
- Model formation in binary logistic regression
- Assumptions of logistic regression
- Running logistic regression in SPSS
- Examples of logistic regression
- Interpreting the logit scale
- Interpreting the odds scale
- Interpreting categorical X variables
- Interpreting continuous X variables
- The proportional odds model
- Model diagnostics
- Measure of effect size
- Using logistic regression for prediction
- Interpreting reference categories and interaction effects
- Ordinal regression
- Introduction to ordinal regression
- Can we fit a linear model to ordinal variables?
- Assumptions of ordinal regression
- Odds of the cumulative probability
- The proportional odds model
- Ordinal intercepts or thresholds
- Ordinal regression model formulation
- Link functions
- Example from SCJS 2008/09
- Running ordinal regression in SPSS
- Interpreting the results
- Alternative models
- Multinomial logistic regression
- Introduction to multinomial logistic regression
- Can we fit a linear model to nominal variables?
- Model formation in mutinomial regression
- Assumptions of multinomial regression
- Running multinomial regression in SPSS
- Interpreting output from multinomial regression