looking for interaction with logistic regression with dichotomous DV

marginally significant regression, only interested in significant interaction


I'm looking at my data for a 2x2 design with a dichotomous DV. Specifically, I want to know whether there's an interaction--I don't care about main effects. I ran a logistic regression in Stata (logit) with both conditions and an interaction term as predictors. Overall, my likelihood-ratio chi-squared(3) = 6.69, p = 0.0825. Since this is just marginally significant, does the significance of interaction term count for anything or not? Do I have to throw my interaction baby out with my overall regression bathwater? The z of the interaction term is -1.99, p = 0.047.

I appreciate any guidance--thank you!!!!
Last edited:


Ambassador to the humans
If the only thing you were interested in was the interaction (and you're not just saying that after the fact) then it seems fine to me to just look at the interaction term.