[contrahax]

I think there is a case for the opposite trend as well - anecdotally I and many others I know moved from NY to another state after responding to the census as a resident of NY. From what I observed (and when I responded) most of the census outreach in NYC was early in 2020 before the pandemic really hit. USPS published some data using change of address information that shows NY lost a good number of residents during 2020, how many of those responded to the census before or after they moved is a toss up though. No conclusion, just adding some context.

[CameronNemo]

Yeah this is such a cluster. Makes you think that maybe we should not be doing point in time counts every ten years. That can have a huge impact on entire generations of people. American Community Surveys run far more often, showing that we can potentially change our ways.

[danso]

Sure, but (ignoring the Constitutional decennial mandate, which also requires Census subjects to be defined 3 years before the actual counting [0]) what's a cost-effective alternative? The budget for the 2020 Census was $7+ billion [1].

And while the ACS annual surveys are considered accurate, what would be the right period for changing Congressional seat counts? It obviously can't be annual (since House terms are every 2 years).

[0] https://www.census.gov/programs-surveys/decennial-census/abo...

[1] https://www.census.gov/about/budget.html

[xnx]

It's also astounding that this is the standard method when Apple/Google/Facebook/AT&T/Verizon could trivially give an instantaneous and probably more accurate count.

[danso]

None of those companies are likely to uniformly count all of the things that the Census purports to do, such as race/ethnicity and housing status.

[codezero]

Even in California when we saw folks moving away I had a census worker on foot knocking door to door trying to ask about who lived in the various apartments - and at the time I know a lot of folks had moved away even if temporarily.