Message-ID: <135954601.4889741446146704255.JavaMail.httpd@webmail-38.iol.local>
Date: 2015-10-29T19:25:04Z
From: Davide Guido
Subject: model building
Hello Everyone!
I have a dataset with 150.000 statistical units (subjects) and 5 variables:
- Binary outcome (0/1) (y)
- municipality (string) (25 small areas)
- gender
- age
- copper concentration (in ppm) (25 level, one by municipality)
The last one, i.e. copper concentration, has been revealed per municipality
(25 levels) and it is defined as municipalty mean of the different municipal
sampling sites. I'm interested to the (conditional) copper effect on outcome
and I have tried to specify this GLMM:
gLMM <- glmer (y ~ gender + age + copper + (1 | municipality), family="
binomial", data=datiSM)
Is it correct fit a model containing both disaggregated and aggregated
variables?
Unfortunately, I cannot measure the copper at disaggregated level (by
subject).
Thanks in advance
Davide