Median household income in 2019 ACS

I’m using SPSS to calculate median household income for detailed Asian groups, but also ran numbers for White, Black, Asian, and Hispanic groups so I could check my numbers against published Census data. I’ve read forum discussions explaining why IPUMS USA samples will vary from the ACS datasets used by the Census, but my results were considerably outside the error margins.:
White (my result) 71,210; (census figure) 69,823
Black (my result) 44,500; (census figure) 43,862
Asian (my result) 96,500; (census figure) 93,759
Hispanic (my result) 58,000; (census figure) 55,658

Following advice in forum posts, I’ve selected cases where PERNUM=1 & GQ<3 & HHINCOME [not equal to] 9999999, and then selected for the relevant racial group (e.g., RACE=1 & HISPAN=0 for non-Hispanic White).

Is there something else I’m missing, or can this much variance be due to differences in sample? Thanks.

Try including cases where GQ = 5 in your analyses–these are households under the 2000 definition of households. I am still outside of the margins of error for the table when trying to replicate the table with these, but GQ = 1, 2, or 5 gets me quite a bit closer than the numbers you posted. I will also mention that it isn’t clear to me from this table that the single race rows exclude householders who report Hispanic/Latinx ethnicity. However, I am much less familiar with the tabular data from the Census Bureau than the PUMS data and haven’t been able to turn up clear documentation on this. However, the inclusion of a “White” row as well as a “White alone, not Hispanic or Latino” row in the table makes me curious about any ethnicity stipulation for the single race rows.

Thanks very much!

1 Like