BRMS Multivariate Syntax Question

Modeling
,
1

Hi,

I am currently fitting this multivariate model with brms:

bf(mvbind(var_1, var_2,var_3) ~ Condition + (1|Subject) + (Condition|q|Region)) + set_rescor(FALSE)

With this model both the random slope (Condition) and random intercept (Region) have correlation terms for each of the three response variables. Is it possible to include correlation terms for the intercept (Region) but not the slope? I suspect this would require uncorrelating the slopes and intercepts as well.

Thanks!

-Tyler

2

I think you could use

... + (1 | q | Region) + (0 + Condition | Region)

which would fit the correlation among Region intercept terms, but neither the correlation among slope terms nor the correlation between the intercept and slope.

3

Thanks, that did the trick!

-Tyler