2

I'm trying to download Census data using R and the Census API. There's a list of variable names at https://api.census.gov/data/1990/sf3/variables.html

However, most of the variable descriptions aren't very informative. Is there a codebook somewhere that explains what each measure is actually measuring? Edit, for example the variable P1200008 is described as "INC BELOW POV[14] Under 5 years" - I assumed this meant "number of people under 5 in households with incomes below the poverty line" but that isn't returning expected values.

Here's the specific data I'm trying to get: I want to end up with county-level poverty rates. The total population numbers (the POP100 variable) seem right (they add up to 240 million or so) but the poverty rate ends up being around 2% when it should be 13%.

     get <- c("POP100", "P0080001", "P1200008", "P1200009", "P1200010", "P1200011", "P1200012", "P1200013", "P1200014",
                "P1190036", "P1190037", "P1190038", "P1190039", "P1190040", "P1190041", "P1190042")


        census <- getCensus(name = "sf3", vintage = "1990", key = key, vars = get, region = "county:*")
        colnames(census) <- get.labels

#poverty rate for county level rows
        census$poverty <- (sum of poverty variables) / census$POP100
#overall poverty rate, should be around 12%
rate <- (sum of poverty columns) / (sum of POP100 column)

Second edit: I was able to find the specific data I was after on an archived USDA page here: https://wayback.archive-it.org/5923/20110904012456/http://ers.usda.gov/Data/Povertyrates/1989_1999/PovListpct.asp?st=NY&view=Percent. Obviously that doesn't answer the general question, so if anyone can answer that (for myself and posterity) I'd be thankful!

  • did you look through the 582 page PDF census.gov/prod/cen1990/cp2/cp-2-1.pdf (link from here census.gov/censusexplorer) – philshem Mar 18 '18 at 14:33
  • Using Decennial Census Data is a deep dive into terms and methodology. Could you tell us what specific measurements you are interested in to give us a better idea of which items might interest you most? – Kotebiya Mar 18 '18 at 14:42
  • @Kotebiya, I'm trying to get county level poverty rates. I tried the variables P1200008-P12000014 (number of people in certain age brackets below poverty in each county) but the total adds up to being too low. Either they're not weighted somehow or I'm misunderstanding what they measure. – Ryan Parsons Mar 18 '18 at 20:35
  • @philshem, I hadn't seen that PDF but I'm not sure it has what I'm looking for. The variable names in the page I linked don't show up in the PDF (unless I'm looking for the wrong thing) – Ryan Parsons Mar 18 '18 at 20:38
2

I think I found the culprit for your discrepant numbers. In addition, I have an alternative and a recommendation.


The Culprit: I believe you are using Table P120 from Summary Tape File 3A.

P120. POVERTY STATUS IN 1989(2) BY AGE(7) [14]

Universe: Persons of Hispanic origin for whom poverty status is determined

I checked the documentation at this page, which led me to this series of supporting documentation.

I found the contextualization for the table you are using in the TBL_MTX.TXT (aka TABLE MATRIX SECTION) file.


Alternative Solution:

I recommend using Table P117 instead.

P117. POVERTY STATUS IN 1989(2) BY AGE(12) [24]

Universe: Persons for whom poverty status is determined Income in 1989 above poverty level:


Recommendation: I recommend not using total population as your denominator for poverty. The Census Bureau recognizes difficulty in measuring poverty for certain populations (e.g. the institutionalized, active duty armed forces, foster children under age 16). These days, the Census Bureau would put the universe for the subject you would be measuring. Here, it seems they do not do that.

To make your denominator, I would recommend summing all 24 items (P1170001 -- P1170024) in the table to calculate the "population for who poverty status is determined".

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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