Help with linear mixed models

I'm new to mixed modelling/borderline understanding what to do with it. I wanted to ask for help with understanding what kind of analysis I should do with my data. In short the design is this:

50 participants take part in the study
Each participant rates 20 faces on how angry they think the face is on scale 1-10
Each participant gets a depression score from a questionnaire

Now, I want to do analysis to look if there is any association between how angry participants think the faces are, and their depression scores. I probably could add the ratings of each face to get a mean rating for each participant and then run a linear regression, with depression score as an outcome variable. I don't think this would be appropriate though. Instead I would like to include every face rating, from every participant in the model, and see how it related to depression. But I don't understand how I can do this with mixed models - I have 20 anger scores and only one depression score for each participant. Does anyone have advice on the best way to analyse this?



Active Member
I imagine that the perception of angriness depends on how depressed you are. So angriness is the dependent variable DV and depression is the independent var IV.
So it seems to me you could use regression of average angriness (Y) on depression (X). Why don't you like it?
You say you would like to include every face rating in the model. Does that mean that you would like to have 20 separate regressions? and 20 p values? and an adjusted critical p value of 0.05/20 for significance (Bonferroni)?
My suggestion - consider a GLM General Linear Model with depression as the sole IV and the faces as DVs. A sort of mancova without groups.