I want to use inverse probability weighting in some regressions and to estimate some weighted means from a non-representative sample. I plan to estimate a probit model for probability of selection into the sample, where the non-representative sample is "selected" and microdata from the ACS represents the broader population. The ACS microdata itself has a weight variable that should be used to ensure that it is representative. Thus, I need to run a weighted probit. I plan to do this using the feglm function from the fixest package. My question is... if I need to apply weights to the microdata from the ACS to ensure it is representative of the population, what weight should I assign to the data from my sample?
For additional context, the ACS weighting variable is PERWT, which indicates how many people in the population are represented by a given microdata observation link
