st: y-standardisation in logistic multilevel models with gllamm

[Tina Schmid <schmid@soziologie.uzh.ch>]

Dear all

When comparing coefficient X1 in binary logistic regression across 
models (same sample, but different covariates), y standardization of the 
coefficient is required (see for example Mood 2010, Logistic Regression: 
Why we cannot do what we think we can do, and what we can do about it, 
ESR, 26 (1), 67-82). So far, so good.
Now we want to compare a coefficient X1 across  binary logistic 
multilevel models using gllamm, e.g. model A with y x1, whereby x1 is a 
covariate on the lowest level, model B with y x1 x2 x3, whereby x2 and 
x3 are measured on the lowest level (x2) and the second level (x3).

We are somewhat puzzled how to do the y-standardisation in a multilevel 
model.
Is it possible to y-standardise the regression coeffiecients in gllamm 
and how can we do it?
Can we use the standard deviation of y* on the lowest level only?

And, imaginge we want to compare a coefficient from the second level 
covariate across  models, e.g. model A with y x2, whereby x2 is a 
covariate on the second level, model B with y x1 x2 x3, whereby x1 and 
x3 are measured on the lowest level (x1) and the second level (x3). Can 
we use the standard deviation of y* (lowest level) or do we have to 
calculate a second level standard deviation of y*?  (In case: how can we 
do it?).

We appreciate any comments.

Thanks a lot in advance

Tina Schmid and Klaus Haberkern

--
_________________________________
lic. phil. Tina Schmid
Soziologisches Institut
Universität Zürich
Andreasstrasse 15
CH-8050 Zürich

Tel.: ++41-(0)44-635 23 47
E-Mail: schmid@soziologie.uzh.ch
Internet: www.suz.uzh.ch/ages

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/