Logistic regression

Logistic regression

1. How do I predict a categorical outcome?

   • Linear model (logistic regression)
     • More than one predictor
       • How do I build the model?
         • Forced entry: predictors entered simultaneously
         • Stepwise: predictors entered based on statistics. Yuk!
         • Hierarchical: predictors entered in steps based on theory
     • How do I measure fit?
       • R-square: the proportion of variance explained by the model
       • Chi-square: change in the model fit
     • What variables can I use?
       • Predictors: continous or categorical
       • Outcome: categorical
     • What are the model parameters?
       • b0: the outcome when all predictors are zero
         • 95% CIs
       • bn: quantifies the realtionship between the outcome and nth predictor

2. Problems

   • Wald statistic: tests if a parameter is different to 0
   • Linearity of the logit
   • Complete seperation
   • Influential cases
   • Outliers
   • Independent residuals
   • Multicollinearity