Dear all,

I am currently working with the user written program -ineqrbd- (for regression-based decomposition, available at ssc). As my data set comes with probability weights, I am a bit stuck now. According to the help file, ineqrbd does not support pweights. After a quick look in the ado-file, I assume that this stems from the fact that ineqrbd uses the summarize and correlate commands which do not support pweights either. However, as pointed out here, you can use the summarize command with pweights by "pretending" them to be aweights and you will still get the right point estimates and SD. Same holds true for the correlate command. With the regress comand, you get the right estimates for the coefficients, however, you will end up with the wrong std. errors.

Thus, I was wondering whether using my pweights as aweights would lead to the correct results, apart from the regression output which could be omitted anyway and replaced by running the regression manually using pweights.