CPS 2022 Food Security Supplement - Trying to generate Food Security/Insecurity Proportion, SE, and 90%CI for households with children.

[Ana Poblacion]

UNIVERSE: Using the CPS Dec 2022 data, I generated a dataset with a restricted sample to 1) one respondent per household (reference person - perrp==40 | perrp==41), 2) households with supplement interview (food security responses - hrsupint==1), and 3) households with children (hrfs12mc == 1 | hrfs12mc == 2 | hrfs12mc == 3). This dataset matches the unweighted sample depicted in the CPS 2022 codebook (n=8,306 households). This dataset also contains the same number of expanded households (n=37, 235,419) and food insecurity weighted prevalence as the USDA 2022 report (FI=17.3%).

OBJECTIVE: To generate proportion, SE, and 90%CI I must use sample weights.

PROBLEM 1: The "svy" command requests sampling weight, a PSU variable, and a Strata variable. The only supplement weighting variables for 2022 CPS Food Security Supplement are HHSUPWGT and PWSUPWGT. Should I use HHSUPWGT as PSU and PWSUPWGT as strata variables? What about sampling weight? Should I use "_n or empty"? I am trying to replicate SE and 90%CI of the figure shown on page 16-10 and 16-11 of the Census Bureau codebook (SE 0.35, 90%CI 82.1-83.3) using Stata's proportion code with sample expansion, however without success. I am clearly doing something wrong when declaring survey design for dataset.

PROBLEM 2: Another issue I am encountering is with "singleunit" option (certainty, scaled, or centered), if missing SE are not generated.

I appreciate in advance any help I can get. Best regards, Ana

[George Ford]

see page 23.

https://www.bgsu.edu/content/dam/BGSU/college-of-arts-and-sciences/center-for-family-and-demographic-research/documents/Workshops/2021-introduction-to-CPS.pdf

what you weight by depends on what you want - a person weighted or household weighted mean.

[Ana Poblacion]

Thank you George Ford for your message. I truly appreciate you trying to help me. Unfortunately the link you provided - and Stata coding - is for the March dataset and not the December supplement dataset. For better or worse, food insecurity data has many particularities - not the same as the core CPS survey.
IPUMS, for reasons I don't understand, collapses some food security questions together and it is impossible to separate them into original responses. For that reason, I downloaded the 2022 CPS Food Security December supplement from the Census Bureau instead. Unfortunately, IPUMS dataset and Census dataset don't have the same variables, or variable names - especially when it comes to weights - which makes difficult to replicate IPUMS coding.
I have been fighting this issue for two weeks now and I seem to have reached a dead end of resources that can help me figure this out.
This Census Bureau resource page is where I am relying my knowledge: https://www.census.gov/data/datasets...-security.html. However, even the technical documentation isn't clear on how to deal with weights for the FS supplement. As I mentioned above attachment 16 (page261 of pdf) is my best bet, but I am still having difficulties finding out how to manage weights in this dataset.
Thoughts?
Best regards,
Ana

[Ana Poblacion]

I finally found what I was looking for. Thank you!

[Leonardo Guizzetti]

I’m glad you were able to solve your problem. If you wouldn’t mind, could you post what you found here in order to close the loop on the question. It may also help people who later find this thread with a similar question.