Food Security Supplement Weights

I am relatively new to data analysis and am looking to analyze the Food Security Supplement and the December core of the CPS over four years (05, 09, 13, 17) on the household level and am a bit confused on how to use and treat the weights in Stata. I have read the USDA’s technical documentation and the IPUMS pages on weights used in the ASEC, but am quite confused and not able to translate this for the use of the FSS and the core module. Could you advise me on how to set this data in Stata with consideration to the weights (svyset?)? Do I also still need to recalculate these (such as: replace fshwtscale = fshwtscale / 10000)? Furthermore, there are several weight variables in this data set, such as hwtfinl, fshwtscale, wtfinl, fssuppwt to list a few, how do these weights need to implemented into the analysis? Do they have to be integrated into commands for descriptive statistics and regressions?

Any help is greatly appreciated!

I will answer your questions one at a time in the reverse order in which you asked the question.

First, the Food Security Supplement includes two separate household level sampling weights. The first, FSHWTSCALE should be used when analyzing food security scale questions, such as FSRAWSCR, FSRASCH, etc. The second, FSSUPPWT should be used when analyzing all other food security questions.

Second, as you mention, there are a number of other sampling weight variables in the CPS data, see here, and these are designed specifically for when using other sets of CPS variables. Many of these additional weight variables will be automatically included in your data extract, along with a number of other technical variables.

Third, if you downloaded IPUMS CPS data using the provided syntax file, or if you requested pre-formatted data, then the sampling weight will already be appropriately adjusted. Therefore, you do not need to divide the weight by a fixed number.

Finally, in Stata, you can either use SVYSET to declare a single sampling weight for all your analysis or most estimation and summary commands incorporate in-line weight specification. More detail on using sampling weights in Stata is available here and here.