The CPS website recommends that users use the weight FSHWTSCALE when doing any analyses on food security variables. However, it seems that FSHWTSCALE is a household weight. If I’m analyzing food security at the person level, should I use WTSUPP instead of FSHWTSCAL?
Also, does the CPS have the variables PSU and STRATA that are used for survey analyses? I can’t find them on the website.
For person-level analyses with the food security variables, you should use WTSUPP.
The CPS does not release Strata and PSU variables; however, the WTSUPP variable does account for both Strata and PSU. Therefore, if you are using svyset in STATA, then the WTSUPP variable is sufficient. For purposes of calculating standard errors that account for the complex survey design, you might consider using the replicate weights.
Hope this helps.