How can I obtain σ^2 and incorporate it in a margins command? It will probably be something like this:
Code:
margins, expression(exp(predict(xb))*(exp(???)))
Code:
mixed lnyvar i.xvar1 i.xvar2 || id:, reml dfmethod(kroger)
Margins from a mixed model with a log-transformed outcome
I run a mixed model of a log-transformed outcome using ‘mixed’ (and the dfmethod(kroger) option with REML estimation because the sample size is only 36). I want to calculate predictive margins with confidence intervals. The ‘margins’ command produces estimates on the log scale so I cannot just back-transform them as E(Y|X) = EXP(XB). I figure I should instead estimate them as E(Y|X) = EXP(XB)*EXP(σ^2/2), where σ^2 is the variance of the errors.
How can I obtain σ^2 and incorporate it in a margins command? It will probably be something like this:
This is my model command, simplified:
Grateful for your help!
How can I obtain σ^2 and incorporate it in a margins command? It will probably be something like this:
0 Response to Margins from a mixed model with a log-transformed outcome
Post a Comment