multiple regression

• regression assumptions

  • variable type:
  • outcome variable must be continuous
  • predictors can be continuous
• regression assumptions

  • non-zero variance
  • predictors must have zero variance
• regression

  • linear
• trucky assumptions

  no multicollinearity

  • predictors must not be highly correlated
  • tested using collinearity diagnostic
• assumptions of regression

  • for each value of the predictors, the variance of the error term should be constant
• assumptions of mr

  independent errors

  • for any pair of observations, the error terms should be uncorrelated
• standard multiple aggression

  • all predictors enteresr at once
• stepwise multiple regression

  • mathematical criteria for order in which predictors are entered - predictors are selected using semi-partial correlations with the outcome
• standardised residuals

  • 3 or more - outlier

  cook

  • anything greater than 1, cause for concern
• hierarchical regression

  • experimenter decides the order in which variables are entered into the model
  • known predictors are entered first