Regression

Regression

1. How do I predict an outcome?

   • Linear model (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
     • What are the model parameters?
       • b0: the outcome when all predictors are zero
       • bn: quantifies the relationship between the outcome and the nth predictor
     • What variables can I use?
       • Predictors: continuous or dichotomous category
       • Outcome: continuous, interval or ratio

2. Sources of bias

   • Outliers
   • Heteroscedasticity
   • Non-normality
     • Bootstrap
       • F-ratio: variance explained relative to error in the model
         • How do I measure fit?
           • R-square: the proportion of variance explained by the model
       • t-statistic: tests if a parameter is different to 0
       • 95% CIs
   • Independent residuals
   • Influential cases
   • Plot zpred vs zreisd