I have some electrophysiology data that I am analyzing with SPSS (I know R however as well if that makes things easier). Participants were recruited (healthy controls and patients) to undergo an electrophysiology experiment. Patients then participated in two different treatments, and underwent the same electrophysiology experiment 8 weeks later. Controls also did the same experiment 8 weeks later from their initial assessment.
I have 92 participants, with the between-group variables:
- Treatment Group: Controls, treatment 1, or treatment 2 (no randomization was done for patients)
- Response status (as in whether they responded to the treatment or not): Controls, responders, or non-responders
I also have 2 within-group variables:
- emotion of task during the experiment (happy or angry emotional stimuli)
- assessment time (baseline or week 8)
Finally, I have some covariates I would like to control for
- site of data collection (data was collected from 4 different locations and then pooled together).
I have many questions with regards to this dataset, but my primary question is this: were there any differences in electrophysiology data at baseline that differentiated eventual responders and non-responders, and if so, was this relationship modulated based on the treatment type.
My initial thought was to use a repeated measures ANOVA. Here, I added 'Response Status' and 'Treatment Group' as my between-subject measures, and my two within-subject variables of 'Time' and 'Emotion Valence'.
However, my question is, with the described dataset, how can I go about "controlling" for the three described covariates? Would this have to be done via linear mixed modelling, or would an ANCOVA be okay?