How to quantify how specific independent variables influence uncertainty of prediction?

I have a mixed model built using brms that has many variables. It looks something like this:

brm(Y ~ (1 + A + B + C + D | E) + A + B + F + G + H + I + J, data=Data)

I am trying to get a sense of which independent variables cause the dependent variable to have a wide variance in its distribution (ie. a large difference between Q2.5 and Q97.5). For example, are the predicted Y values more uncertain when A is very low, F is very high, etc? Any advice or references on how to go about quantifying this?

Plot the posterior predictive distributions fixing all but one variable to a typical value.

1 Like