I'm planning on using a (linear mixed model) LMM for my analyses but am not 100% sure if it is actually suitable.

This is what my data are like:

  • I have one variable measured 4 times for each participant (let's call it dependent variable (D)). It is a subjective measure of my participants rating something.
  • I have a variable indicating what participants saw before rating the dependent variable (let's call it predictor 1 (P1)). It has 4 levels (i.e. the four different things participants saw before rating D).
  • I have another variable measured once (let's call it predictor 2 (P2)). It is a continuous personality measure from 0-100.

I want to investigate if D is different depending on the level of P1. Furthermore, I want to investigate if the magnitude of the effect of P1 on D is influenced by the magnitude of P2.

Here is what I planned to do:

Run an LMM with D as dependent variable and P1 and P2 as predictors specified as fixed effects (also I plan to add random effects for my participants because there is reason to believe that the effect might not be the same for everybody). Then I will have a look if P1 is a significant predictor and if there is a significant interaction between P1xP2 to answer my question.

Is that possible or am I thinking in the wrong direction?


1 Answer 1


The approach you described seems to be a reasonable way to investigate your research question(s). In particular, since you have repeated measures within subject, a mixed model with random intercepts for subject may account for the non-independence of observations within subjects, while the estimates of the fixced effects including the interaction will answer your research questions.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.