0

I have a question on interpreting interaction effects: I've read a lot of posts with similar problems but not with the exact same one, so I hope someone can help. Huge thank you in advance!

In a first model, I have two independent variables A and B which have a statistically significant coefficient on my dependent variable.

In a second model, I add an interaction between A and B. The interaction does not attain a statistically significant coefficient, but now A and B alone are no longer statistically significant either.

How do I interpret this?

Does it mean that while the interaction did not reach significance, it nonetheless explained further variance, so that it was shown that the main effect are actually not significant?

Or should I not interpret the model as the interaction is insignificant and instead say that A does have a significant effect as displayed in the first model?

laramolitor
  • 1
  • 1
    Does this answer your question? Nonsignificant interaction still causes main effect to flip? With the interaction, the significance of the "main effect" for each predictor is for the difference of its value from 0 when the other interacting predictor has a value of 0. That value might make no sense whatsoever in practice. Also, every predictor your add to a model (like an interaction term) uses up another degree of freedom, making it potentially harder to find true associations. – EdM Jun 26 '22 at 17:35
  • 2
    Plotting your data might give you more insight than a theoretical explanation. – dipetkov Jun 26 '22 at 17:36
  • Please clarify your specific problem or provide additional details to highlight exactly what you need. As it's currently written, it's hard to tell exactly what you're asking. – Community Jun 26 '22 at 20:17

0 Answers0