mixed with log-transfered dependent variable: back to non-log

[marta]

Hi, I have a question regarding multilevel model when I use log-transformation
on the dependent variable.

generate ln_depvar= ln(depvar)
mixed ln_depvar c.c_age##i.bmicat c.c_age##i.childcat c.c_age##mamm || id: c_age, reml covariance(unstructured)

Is there a way to change the above code so I can back log values of the output of the mixed command ?
Thank yuo
Marta

[Joseph Coveney]

How about not transforming the response variable in the first place?

meglm depvar c.c_age##i.bmicat c.c_age##i.childcat c.c_age##mamm || id: c_age, ///
    family(gaussian) link(log) covariance(unstructured)

Either that, or continue to work in the log-transformed metric after fitting with mixed .. . reml, that is, don't back-transform.

[marta]

thank you Joseph