Topic 2 - Forms of Multiple Regression

Standard Multiple Regression

regression where predictors are added into the model simultaneously

when to use a standard multiple regression

where theory suggests a combination of predictors together explain variance in an outcome variable

Ideal predictor characteristics in a standard regression

have low correlation with one another, have moderate/high correlation with the outcome, and have equal importance in explaining the outcome

R squared

The percentage of variance overall in the outcome accounted for by all of the predictors

multicollinearity

occurs where more than one predictor variable included in the model are linearly related

measures of multicollinearity

tolerance and variance inflation factor

how to interpret tolerance

ranges from 0-1, the closer to 0, the higher the multicollineaity

how to intepret VIF

the reciprocal of tolerance, the larger the VIF, the higher the multicollinearity

Hierarchical Regression Analysis

regression where variance in the outcome is attributed to the predictors based on the order they are entered into the analysis

when to use hierarchical regression

when theory suggests after accounting for the variance in predictors known to influence the outcome, further predictors also make a significant contribution

R^2 change

how much additional variance is accounted for by each step of the hierarchical regression

significance of Fchange

whether the added predictors in a hierarchical regression add a significant amount of unique variance to the model

Statistical / Stepwise Regression

A data-driven form of hierarchical regression where the predictor with the largest correlation is entered first