0

My question is directly linked to the popular lmer cheat sheet. There was one situation that I didn't see mentioned in that cheat sheet and I wanted to know if it was a valid scenario or something that has no interpretation? I am still a neophyte to the complex world of mixed modelling so forgive me if the formulation does not make sense.

I wanted to know if:

V1 ~ V2 + (V3|V4)

Was correctly stated model. Particularly I'm interested in how I interpret the random effect associated with V3. In the usual case the model would have been specified in the form V1 ~ V2 + V3 + (V3|V4). And in this case the random effect on V3 is measuring how much it deviates from the global effect of V3 which would be the fixed slope effect. I was wondering if in the scenario I stated above would the interpretation of V3 be similar but this time there is no "global effect" estimated in the model and only the variation from V3?

  • 2
    Yes, this model has a population-level V3 slope of zero and the subject-level slopes are centered around that. Usually, this is not considered sensible. You should always include a corresponding fixed effect if you include a random effect. – Roland Aug 22 '23 at 05:51
  • Fantastic. Thanks for the reply Roland. – D.C. the III Aug 23 '23 at 02:54

0 Answers0