0

I've got a conventional OLS-regression with some nested models, first specifying main effects.

In the first model, I get a positive b-coefficient on my main effect. However, when I add an interaction term with this variable, the interaction becomes significant, but the main effect switches direction (i.e., to negative) and is still significant.

I am a bit unclear as to how to interpret this. Could someone help me out?

Best, Tony

user17917723
  • 1
  • I think your question would be more clear if you could write the equations for the two models – fabiob Apr 06 '22 at 11:05
  • Note that the "main effect" of a predictor involved in the interaction is its association with outcome when its interacting predictor is at a value of 0. Please look at the many other questions on this site about "main effects" in the presence of interactions, and ask again with more details about your particular models if those don't provide the answers you need. – EdM Apr 06 '22 at 12:42

0 Answers0