0
$\begingroup$

I'm reading an article where the author has interacted two variables -- X and Y. The model contains variable X, variable Y, and the interaction of variables X and Y. However, I know from existing work that variable X is a predictor of variable Y. Is this problematic? If so, how?

$\endgroup$
  • $\begingroup$ It is very possible for two variables to be interrelated, while also modifying each other's effects with regard to the outcome. $\endgroup$ – gammer Feb 5 '17 at 1:29
  • $\begingroup$ Can you link to the article? $\endgroup$ – Maddenker Feb 5 '17 at 5:53
0
$\begingroup$

If two of the independent variables are strongly correlated, this can lead to a (multi)collinearity problem. This can potentially make one or several of the variables appear non-significant, although they would be significant if the regression contained only either one of them. To check for collinearity, the VIF (variation inflation factors) is a good test - many empirical papers (depending on the field, of course) report the VIFs of their regressions. Regarding the interaction term, it will also be correlated with $X$ and $Y$, but I don't think this makes it any more problematic than the correlation between $X$ and $Y$ themselves.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.