2

The US census provides the number of housing units, which is supposed to represent individual dwelling units, not structures. The American Community Survey provides the number of units by structure type. My question is regarding the data for housing structures with more than one unit. I am looking at b25024 data set:

B25024e1 UNITS IN STRUCTURE - Universe: Housing units - Total: -- (Estimate)

B25024e2 UNITS IN STRUCTURE - Universe: Housing units - Total: 1, detached -- (Estimate)

B25024e3 UNITS IN STRUCTURE - Universe: Housing units - Total: 1, attached -- (Estimate)

B25024e4 UNITS IN STRUCTURE - Universe: Housing units - Total: 2 -- (Estimate)

B25024e5 UNITS IN STRUCTURE - Universe: Housing units - Total: 3 or 4 -- (Estimate)

B25024e6 UNITS IN STRUCTURE - Universe: Housing units - Total: 5 to 9 -- (Estimate)

B25024e7 UNITS IN STRUCTURE - Universe: Housing units - Total: 10 to 19 -- (Estimate)

B25024e8 UNITS IN STRUCTURE - Universe: Housing units - Total: 20 to 49 -- (Estimate)

B25024e9 UNITS IN STRUCTURE - Universe: Housing units - Total: 50 or more -- (Estimate)

Here is a description of that data:

2 or More Units--These are units in structures containing 2 or more housing units, further categorized as units in structures with 2, 3 or 4, 5 to 9, 10 to 19, 20 to 49, and 50 or more units.

When I look at the data, it doesn't make sense. All the total units in B25024 add up to the total housing units reported by the census... great! But if you look at one of these categories, such as B25024e8 (Total 20 to 49), there are some block groups with values like "11" or "12" or "2".

How can you have a value in the 20-49 units category that is less than 20 units? Is it possible that the original data was at the tract level and got split to block groups?

4
  • 3
    You should ask this the authors of the census methodology.
    – lynxlynxlynx
    Commented Oct 4, 2017 at 20:48
  • If this is open data then I think it is better researched/asked at the Open Data Stack Exchange
    – PolyGeo
    Commented Oct 4, 2017 at 21:13
  • @PolyGeo it is publicly available data... so should I close this question and ask it in Open Data then?
    – farrenthorpe
    Commented Oct 4, 2017 at 21:56
  • I'll migrate it there for you.
    – PolyGeo
    Commented Oct 4, 2017 at 21:57

1 Answer 1

1

I was able to get an answer to this question by speaking directly with a census employee who is an expert in the data category.

He agreed that in the B25024e8 (Total 20 to 49) data-set (for example), the minimum value should be 20. However, he acknowledged that there are some block groups with values like "11" or "12" or "2". He explained that this is because the people who took the survey must have checked the wrong box. So, for instance, if they lived in a 2-unit structure, but checked the 20-49 unit structure box on the survey, the data report would retain the error. He explained that they do not have information on the types of structures within each block group, so they depend entirely on the survey responses. Thus, there is potentially a large margin of error in the reported data-sets. This is why "margin of error" (e.g. B25024m8 data-set) is provided along with the actual estimate.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.