So I have two interaction terms (ordinal x ordinal--treating each as continuous) I'm trying to test in a logistic regression model. Interaction 1 is significant--but not when Interaction 2 is present. Likewise, interaction 2 is significant--but not when interaction 1 is included. I thus went ahead and tried a 3-way interaction, but it wasn't significant either. Not sure how to interpret this (do I have to choose one 2-way interaction?), so would appreciate some input. See below for screenshots of what I'm referring to.

enter image description here

  • $\begingroup$ Not a duplicate. That question doesn't ask about interaction terms. $\endgroup$ Commented Aug 12, 2017 at 1:20
  • 2
    $\begingroup$ This is a duplicate. It doesn't matter whether the variables in question are interactions or not. The reason is the same. $\endgroup$ Commented Aug 12, 2017 at 1:28


Browse other questions tagged or ask your own question.