I’m complete new to bayesian analysis and brms and am now taking my first steps to estimate a multivariate multiple membership model using brms in R.

Most of my independent and dependent variables are averaged scores from multiple items that have a likert-type rating scale (scores 0 to 4). Two variables are standardised achievement scores and can thus also take on negative values. Most variables are moderately skewed and some also have kurtosis. My guess is that I have to use the skew_normal family. Is that right? Does that only take into account the skewness or also the kurtosis?

My next question would then be what link to choose. I found this specification under Usage on https://rdrr.io/cran/brms/man/brmsfamily.html:

skew_normal(link = “identity”, link_sigma = “log”, link_alpha = “identity”)

Are these the default options or these the only options for the skew_normal family?

Thanks,

Pia