Hi there,

I am trying to do Yun's decomposition for a Probit regression using the Oaxaca command (by Ben Jann). I'm having some trouble with weighting though.

When I use the following command:

Code:
svyset cluster [pweight=pweight_cross], strata(dc2001)
oaxaca lfps (education:normalize(educcat11-educcat15)) (age: normalize(agecat21 b.agecat22 agecat23 agecat24 agecat25 agecat26 agecat27 agecat28)) (married: normalize(b.maried11 maried12)) otherhhincpc if working_age==1 & wave==2, by(male) probit pooled relax svy
I get the following comments:
"(model 2 has zero variance coefficients)
(pooled model has zero variance coefficients)"

And then I get no standard errors or t stats at all in my results

However, when I take away the "strata(dc2001)" from my svyset command, my Oaxaca command runs perfectly. I don't have problems with the above weight for any of the other commands I use though. I am very new to using the Oaxaca command, especially with a binary dependent variable. I have read the help file but am struggling to understand what I am doing wrong. Can anyone help me? TIA