Skip to content
Prev 18772 / 20628 Next

Why lme4 doesn't throw an error for illegitimate an model

Very interesting Phillip! I'll think about your final question! But, in
your terminology, what is the relationship between "adjustments", "random
effects", and "shrinkage" (of cluster estimates towards the total cluster
mean)?

On the other hand, let's now use an individual-level predictor (i.e.,
"ses": varies within & across clusters) in the random part of the model
WHILE excluding that predictor from the fixed part.

what does `ses` in the output of `VarCorr(m43)` below represent?

hsb <- read.csv('
https://raw.githubusercontent.com/rnorouzian/e/master/hsb.csv')

m43 <- lmer(math ~ female*minority + (ses | sch.id), data = hsb)
Groups   Name        Std.Dev. Corr
 sch.id   (Intercept) 2.1961
          ses         2.0963   -0.285
 Residual             5.9732
On Wed, Oct 7, 2020 at 6:29 PM Phillip Alday <phillip.alday at mpi.nl> wrote: