Possible to unpool intercepts in rstanarm model?

[Lizzie]

I have a simple lmer syntax question I cannot seem to find the answer to, which is stymieing an rstanarm model I want to try.

I’d like to try a model with pooled slopes (pooled by some grouping factor, g) and *unpooled* intercepts (I realize there are some inference issues with this approach but I think it may be best for my ultimate goal with the model).  

I understand that:

stan_lmer(y ~ x + (x | g))

should give me pooled intercepts and slopes and that:

stan_lmer(y ~ x + (1 | g))

should give me pooled intercepts and unpooled slopes, but I cannot seem to find in vignettes, manuals or queries online how to request pooled slopes without pooled intercepts.

Sorry for the basic question, but any help would be much appreciated!

Lizzie

[Ben Goodrich]

On Thursday, January 5, 2017 at 11:53:42 AM UTC-5, Lizzie wrote:

I understand that:

stan_lmer(y ~ x + (x | g))

should give me pooled intercepts and slopes and that:

stan_lmer(y ~ x + (1 | g))

should give me pooled intercepts and unpooled slopes, but I cannot seem to find in vignettes, manuals or queries online how to request pooled slopes without pooled intercepts.

stan_lmer(y ~ x + (0 + x | g))



 

[Lizzie]

How very logical. Thanks!