Dear Lorna,
I've done some work with the SPM. Have you looked at https://www.census.gov/content/dam/Census/library/working-papers/2020/demo/SEHSD-WP2020-09.pdf yet? The SPM data file combines data from the ASEC Microdata (N= approx 95,000 households with the ACS microdata (3.5 million households). The ASEC has much more detail on sources of income. Are you looking at the SPM variable SPM_resources or SPM_totval for income or, once you have merged the files, ACS PUMS variable HINCP ? You might make some scatter plots for these 3 variables to get an idea of how things vary with the different variables. I assume that to "pull over" the ASEC detailed income variables 'into' the ACS PUMS data a statistical "model" is used for the matching. Although I haven't located the detail methodology paper yet. Information on how the ACS computes income for the POVPIP PUMS variable can be found here https://www.census.gov/topics/income-poverty/poverty/guidance/poverty-measures.html . The SNAP ACS question is a Yes/No question IN THE PAST 12 MONTHS, did you or any member of this household receive benefts from the Food Stamp Program or SNAP (the Supplemental Nutrition Assistance Program)? Do NOT include WIC, the School Lunch Program, or assistance from food banks. I would investigate how the Household Income on the ACS is computed. I think that it is the personal income summed across RELATED persons in the household. (This tends to work against your observation. I looked at the WA SNAP form (you are an expert on that whereas I am not). The question that I see is 10. How much money do you expect your household to get this month? The ACS asks several detailed questions about different sources of income over the last 12 months. This includes many cash benefits. I would think that the ACS might report higher incomes because the ACS asks about a list of individual income sources where as the SNAP application asks about a lump sum monthly amount. I don't know if there is research on the (statistical) bias between these different questions. The scatter plots might help with this. In addition to plotting the ACS "Household Income" PUMS variable you might plot the individual components of the ACS household income and do some simple linear regressions and put the regression line on the scatter plot.
Best,
Dave Dorer
------------------------------
David Dorer
Dorer Community Service Foundation
------------------------------
Original Message:
Sent: 05-06-2026 09:49 AM
From: Lorna Holzworth
Subject: SPM data
Hello, has anyone investigated the SPM data much? I downloaded it and compared some of the figures to our own (dshs in Washington). Specifically, I looked at the SNAP incomes and recipients reported. The total recipient figure is close to our actual data, but the income/issuance is way high, about double, in the SPM data.
Any ideas?
Lorna
------------------------------
Lorna Holzworth
Rresearch Investigator
State of Washington
------------------------------