PSYC 3130 Lecture Notes - Lecture 5: Multicollinearity
Document Summary
Any time one says it depends , he or she is referring to a moderator. Mediator: the root by which two variables are linked together. Requirements: normal distribution, no high correlation between predictor variables, random observations. If standard residual is below -2 or above +2, there is error. One variable cannot be on a different scale. Ex, height variable is on a different scale than temperature (temperature is usually the same). Interaction variable is created by multiplying standardized variables by each other. Trichotomy: used to tease apart significant interactions.