ACS Data Users Group

Hello -- I am trying to look at changes in single parent households by place, specifically Tacoma, WA. I'm using the 2021 5-year estimates, and typically I prefer to use the detailed tables so I pulled data from B11005. My colleague pointed out the estimates were different when looking at DP02. Out of curiosity I also compared S1101, which from what I understand should be including the same data shown in B11005. Here are the details of what I'm seeing for Tacoma, WA in the 2021 5-year estimate:

S1101 -- Male householder, no spouse present, family household; Households with own children of the householder under 18 years: 2,485

B11005 -- Households with one or more people under 18 years; Male householder, no spouse present: 2,832

DP02 -- Male householder, no spouse/partner present; With children of the householder under 18 years: 1,282

Can someone help me understand the differences here? I found some documentation describing the nuances in what "children of the householder" means, but it seems like there must be something else contributing to such a wide range of estimate.

Thank you in advance.

I can't offer you a clear answer, but the first thing I do is check universes, and all 3 of those tables start with 129,870,928 for the US, so they match. Just because the totals match doesn't mean they have the same variations of measures, it might be the non-family households. I just try to add up various pieces to see if I can match the highest number 2,832 in each table. A parent-child relationship is different than an adult-child one. The naming standards are not great and when I see "no spouse present" vs "no spouse/partner present" it indicates a difference, but I always have to do the math (assuming all the variables are there) to understand what each table is saying. Regardless of the documentation, which is also not always a home run, I always try to use a second table to match whatever measure I'm looking at to help assure me I'm not unfortunately using that one-off term or measure and then I just run with it. I've done that with slightly more than 100 measures and I probably have a few that might be different than I think they are, but it's very time-consuming.

Good Luck.

Thank you, all good suggestions. I hadn't thought to check an additional Detailed table to validate measures / definitions. I typically prefer the categories in B11012 when looking at household type, but the measures change drastically for the 2011 5-year estimates so I stopped using that table when looking at change over time.

Both B11012 and B11005 list Households as the table universe and have the same geography restrictions, but B11012 has the 1,282 estimate for Male householder, no spouse/partner present; With children of the householder under 18 years, which matches the DP02 numbers, but is less than half of the estimates given in B11005 and S1101.

To add another table to the mix, I checked B22002 which has an estimate of 2,832, matching the B11005 estimate. Both of those tables have the same description (Households with one or more people under 18 years; Male householder, no spouse present), so perhaps the estimate doubles with the inclusion of partners?

Not sure about the drastic change, but I think B11012 is a better table. We are weeks away from ACS5 2022, so maybe wait for the update and see if it's viable. I think the national number for Married with kids 18.6% and Single Parents at 6.4% so about 1/3. So there is a big diff. Note Single parents is not a measure that I could find, so I had to calc it.

I am clearly not ready for the new ACS5, and may have to spend my Christmas week sorting out all my measures. That's 87 for every County and State and for 2021 and 2022 including variances vs US. Just a bit over 1.8 million values, which should make for at least a few good dashboards.

Assuming I can do it (Early Jan). I will have every county and state Married, CoHab, Male, Female, and Single-parent households and the difference vs last year and vs US (ACS5). Maybe everything you want and more. The data for parent households (I have extra stuff) is only 2 URLs (for county), and all you have to do is change 2021 to 2022 (when its available), although there is a fair amount of cleanup and added calcs to make it a great data set (my 2 weeks of effort). It takes 12 pulls to get all the raw data (12 minutes), then I add 2 weeks of grunt work.

api.census.gov/.../acs5

As Tom said check the universe for each table. One table B11005 is people under 18 years the other (B11012 ) is children under 18 years. "Children" means children of the "head of household". These are two different things. For example foster children would count in B11005 but not B11012. I'm not sure where adopted children fit it these tables. Carefully check the definitions here www.census.gov/.../subject-definitions.html