1

this is a topic that has been discussed in other posts, but there was never a direct answer to my question. I have a dataset whereby my dependent variable is predicted by three categorical factors of two levels each. I am interested in a specific contrast, whereby for one of the levels of factor2, I want to see if there is a difference between the levels of factor1. I am not interested in what happens for the other level of factor2, that is, I don't have a specific hypothesis about the presence or not of an interaction. I just want to see if my planned comparison gives a significant difference. Now, due to the design, I am running a mixed-effects model to model the data. Would it be correct for me to look at the planned comparison independently of what I obtain for the interaction term between factor 1 and factor2? As I am not interested in it? I would not run the planned comparison separately, as I wanted to factor in the variance explained by the other predictors of my dependent variable to get my estimates for the planned comparison. Does this make sense?

Many thanks for any help you might provide!

SinC
  • 21
  • 1
    If you are not interested in what happens at the other level of factor 2, why include those observations in the data set at all? – Peter Flom Nov 08 '23 at 10:59
  • 1
    Brief answer: yes, you can investigate your planned comparison in the absence of a significant interaction. See this answer and this answer for relevant info. – Sointu Nov 08 '23 at 11:00
  • Yes, good point. It is because of an exploratory analysis I wanted to run - I did not have a specific hypothesis about what would happen for the other level of factor 2. – SinC Nov 08 '23 at 11:01

0 Answers0