I am attempting to estimate the average gross rent in Miami Dade County, Florida. Based on my understanding of the data, the hhwt variable provides the number of households in the general population which the sample represents, and should therefore be used to weight the rentgrs variable. For instance, if I was running this calculation in Stata for the entire US, I would first filter my data to records where pernum=1, and then run the following:
mean rentgrs [pweight=hhwt]
I believe this would provide an accurate point estimate for the mean, with imprecise standard errors (which is another issue).
My question is whether this same approach is correct for subregions? I am slightly confused by what is meant by “general population” in the weighting definition. Online resources seem to suggest that weighting is based on the demographic characteristics of the head of household, and how similar these are to the underlying population. If weighting is intended to reflect the entire US population, the underlying demographic weighting wouldn’t seem to be a good reflection of a subregion such as Miami-Dade.