3

I have having trouble understanding the difference between a suppressor variable and multicollinearity in multiple regression. If a suppressor variable is one that is not correlated to the outcome, but is correlated to another predictor, isn't this multicollinearity? Or does it depend "how strong" the correlation between predictors is? (i.e., VIF is <10).

Does anyone have a complete resource on linear and multiple regression, including outputs? I need to refresh my knowledge on this because the more I read, the more I feel I was lied to in university that these models are "simple".

DataNoob7
  • 71
  • 6
  • 1
    [Suppression](https://stats.stackexchange.com/q/73869) and [multicollinearity](https://stats.stackexchange.com/q/70899/3277) are different phenomena. – ttnphns May 12 '18 at 15:32
  • This thread goes into a lot of depth in terms of explaining suppression: https://stats.stackexchange.com/questions/73869/suppression-effect-in-regression-definition-and-visual-explanation-depiction. – Isabella Ghement May 12 '18 at 16:21
  • Thanks for the resources. I have read those threads many times, and I still don't understand the difference, other than that suppressor variables are low correlated explanatory variables. Granted, the algebra and geometry are a bit over my head (for now). – DataNoob7 May 12 '18 at 16:33

0 Answers0