3-way rmANOVA with covariate or ANCOVA?
Hi all,
my goal is to investigate the role of different sleep measures on skin conductance responses (DV) in a fear conditioning paradigm. IVs are Conditioned Stimulus (CS, 3 levels), Context (3 levels) and trial (2 levels). (That is, every CS was presented twice in each context.) To achieve this, I tried two things:
First, a 3-way rmANOVA with one of the sleep variables as covariate. My problem here is that I can't create plots that include the covariate. Can I? 🤔
Second, I ran an ANCOVA with CS, Context and Trial as fixed factors and the sleep variable as covariate. My (potential?) problems here are: I had to sort my data in long format to do the analysis in JASP and I'm not sure if this is actually a problem since, as far as I understood, JASP now "thinks" that every row is a new participant. Furthermore, JASP doesn't offer a possibilty to run a 3-way rmANCOVA, does it? Is that even a thing?
Is there an actual difference between a 3way rmANOVA with a covariate an an ANCOVA ? If yes, what is the more accurate analysis for my purposes? Sorry about the rather general questions but I couldn't really find any help online so far.
Many thanks in advance, highly appreciate your help!
Comments
Let me get this straight. So you have a 3x3x2 RM ANOVA:
"my goal is to investigate the role of different sleep measures on skin conductance responses (DV) in a fear conditioning paradigm. IVs are Conditioned Stimulus (CS, 3 levels), Context (3 levels) and trial (2 levels). (That is, every CS was presented twice in each context.)"
But then you also have a "sleep variable", as a covariate. Is that correct? If so, you can just run the RM ANOVA with the sleep variable as a covariate...
EJ
Yes, this is what I ended up doing, thank you for this. My problem now is that there is no way to assess the effects of the covariate more closely, e.g. with post hoc testing. As far as I understand, I can only test the independent variables with the covariate being included in the analysis but "invisible". Is that correct, or is there any way to take a closer look at the effects of the covariate?
Hi @Dexterama,
If you add a continuous covariate in the RM ANOVA, it will appear in the between subjects effects table, in addition to all possible interactions being added (since any interaction effect will affect the interpretation of the main effects, it's always good to check this and why this is the default behavior of jasp here). The evidence for the main effect of your covariate can then be assessed using the F/p-value (although ideally, you'd do a Bayesian analysis ;-)) or its effect sizes eta squared.
Does that answer your question?
Cheers
Johnny
Yess, that helped, I have to admit that I completely ignored that table - thank you!