6

There is currently a legal battle over changing the US census to include a citizenship question.

The plaintiffs argued that participation in the census will be depressed by the addition of the new question, causing a significant undercount. If the government is successful in adding the citizenship question, the census will yield flawed data.
- A C.E.O.’s Plea: Don’t Mess With the Census - NYT Op-ed

The claim certainly makes logical sense, because we expect illegal immigrants to tend to avoid revealing their citizenship status. However, the citizenship question has been on the census before, so we can compare data from those periods to data from other periods where the citizenship question was not on the census. In that comparison we can find disparities that may be explained through lack of participation, as the claim alleges.

So regardless of the logic in the claim, is it true that a citizenship question affects participation in the US census?

An already existing question provides some helpful background information, but not an exact answer to this question.

  • 5
    "the citizenship question has been on the census before, so we can compare data from those periods to data from other periods where the citizenship question was not on the census." I disagree. The political and technological climate is _way_ different now. – Kevin Apr 22 '19 at 20:02
  • 2
    @Kevin Perhaps that challenges the validity of any comparison, but that's a case an answer would have to make. Let's see some data first. –  Apr 22 '19 at 20:11
  • 1
    In the least, a comparison can prove there's lesser participation, as the claim says, regardless of differences in politics and technology. Proving that participation is unaffected is the harder half of the problem where your objection may come in. –  Apr 22 '19 at 20:13
  • 3
    @fredsbend: "a comparison can prove there's lesser participation": How so? Correlation is not causation. Suppose we find that in year X, with a citizenship question, participation was less than in year Y, when there was no question. That will not prove that the citizenship question had anything to do with the different participation rate. A lot of things may have changed between years X and Y (which are necessarily at least 10 years apart), and participation rates can be affected by societal factors that have nothing to do with the questions on the survey. – Nate Eldredge Apr 23 '19 at 01:31
  • @fredsbend The old censuses that asked about citizenship were extremely complete, but census takers went to each house in person. – DavePhD May 03 '19 at 11:39
  • @DavePhD: Is that not the case now in the USA? (Asking as an ignorant foreigner.) – Oddthinking Jan 05 '20 at 02:11
  • People who take insulin are more likely to die of diabetes than people who don't. What does that prove? Unless you have a randomized trial, any comparison of rates is deeply fallacious. – Acccumulation Jan 05 '20 at 06:59
  • 1
    @Oddthinking Now they mail forms, and people should fill the forms and mail them back. If the form is not returned, then there should be a personal visit. – DavePhD Jan 05 '20 at 22:09
  • it certainly affects the participation by non-citizens who're in the country illegally... – jwenting Jan 06 '20 at 05:52

0 Answers0