Setting different priors in a repeated measures ANCOVA
Hello everyone,
I am running a Bayesian repeated measures ANCOVA (2x3x3 with age as a covariate) using the default priors in JASP. However, I wanted to test the results under different priors, and this is where I get confused. Setting priors for the fixed effects seems a bit straightforward. I have found several examples and references and re-tested everything under narrow (0.2) and a wider prior (0.8). Yet, I struggle to find good references and examples of how to set priors for the covariate and random effects. Is there any ratio between the priors I must follow (fixed vs random vs cov). Can I just put any narrow and wide prior?
In sum, do you have any suggestions on how to test different priors in such cases (i.e., rANCOVA), or maybe there is a recent paper I missed that you could point me to?
Thank you in advance,
Mila
Comments
Hi Mila,
I think this topic is understudied. Pragmatically, I might suggest either doubling or halving the default setting...
Cheers,
E.J.