Confusion about generating estimates for *households* and not *individuals* in MDAT

Hi,

I'm currently trying to compile data for the percent of households receiving food stamps/SNAP in Dane County Wisconsin, broken out by Hmong alone, All Asian (alone or in combination with other races), and Non-Asian, using 2017-2021 ACS. The data for Hmong alone is not available on data.census.gov, hence the use of the MDAT. However, when I run this in MDAT, it gives me counts for individuals, not households, even though Housing Unit Weight is selected. E.g., for the "Total" it gives me 555,435, which is the population of Dane County, not the number of households. Is there some sort of setting or checkbox that I'm missing? Or am I misunderstanding the Housing Unit Weight? Thanks for any assistance!

Parents
  • Perhaps query this in iPUMS or by merging the housing and person records PUMS SAS files?  I like MDAT, and maybe you can yield your desired results in MDAT by tweaking your setup, but I find the other PUMS sources more flexible.  

Reply
  • Perhaps query this in iPUMS or by merging the housing and person records PUMS SAS files?  I like MDAT, and maybe you can yield your desired results in MDAT by tweaking your setup, but I find the other PUMS sources more flexible.  

Children
  • Dear Brittany,

    When you are using PUMS variables that are based on the "person" data set you need to combine the person variable across the household. If you use the  FTP files there are 2 files for each PUMA a person file and a household file you need to decide how to handle the person variable  Also when you look at the PUMS codebook there are 2 sections one for household variables and one for person variables.  When you want a household variable that is based on person variables for people in the household you need to determine how to combine the data for (across) the members in the household.  You will need to use some type of logic that involves the SPORDER person variable.  The SNAP PUMS variable is a household variable. "Yes" means that someone in the household receives SNAP. Keep in mind that there might be two or more families in the household (see RELSHIPP variable).  To define a Hmong household do you mean there is at least one Hmong person in the household or do you mean that all the household members are Hmong?  You need to decide.  The only way to do the logic is to use some type of statistical package like SAS or R. I don't use the MDAT interface I use the API but you might not be able to do the logic through the MDAT interface.

    Best Dave Dorer