Bayesian RM ANOVA with continuous covariate - posterior summary for interaction
I am fairly new to Bayesian analyses and have had more luck understanding what I'm doing with JASP than with R, but I am not sure there is a solution to my question in JASP. I have a repeated-measures design (three trial types) with age as a covariate in a sample of children, with reaction time as the dependent measure. The model comparison in the Bayesian repeated measures ANOVA indicates the best-fitting model includes main effects of age and trial type plus the interaction of age and trial type. In the Model Averaged Posterior Summary table, the entries for the interaction are all NaN, I assume because the interaction is with a continuous covariate.
If I try to add the interaction under Single Model Inference I get "Error in single model inference: Error in tcrossprod(datOnHot, posterior): non-conformable arguments"
I've tried searching online for a way to calculate the estimates in this case and have not had any luck. Is this something that can be done? When looking at the data it seems the interaction reflects a steeper slope for one trial type than the others.
Thanks in advance for any thoughts and suggestions!
Comments
Interesting question, I'll forward this to our expert. One would hope that a little more information would indeed be available.
Cheers,
E.J.
I know this is a relatively old post, but I wanted to share that this is no longer a problem in the latest version of JASP! So thanks for whatever happened behind the scenes to fix it :)
:-)