Reply to DC had largest percentage drop in population in nation

Forum Index » Metropolitan DC Local Politics

Reply to "DC had largest percentage drop in population in nation"

Subject:

Emoticons





More smilies

Text Color: Font: Close Marks

[quote=Anonymous][quote=Anonymous][quote=Anonymous][quote=Anonymous][quote=Anonymous][quote=Anonymous][quote=Anonymous][quote=Anonymous][quote=Anonymous][quote=Anonymous][quote=Anonymous][quote=Anonymous]At many people in DC did not fill out census b/c they were  busy, angry, afraid of their immigration status, or remote working somewhere else.[/quote]
Relevance?[/quote]

Not the PP, but they are correct that DC's population is especially hard to count right now. The Urban Institute concluded that DC's population was undercounted by over 14k residents in 2020, larger as a % of population than any state.

https://www.washingtonian.com/2021/11/02/nearly-14400-washingtonians-werent-counted-in-the-census-new-report-says/

It's not a big reach to suspect that the Census population change numbers may also be not as reliable as usual in cities like DC. For example, the Census Bureau isn't even releasing a lot of their usual statistics for 2020 because the data aren't good enough.

https://www.rollcall.com/2021/07/29/census-wont-release-key-annual-survey-because-of-pandemics-impact-on-data/

Without knowing how this particular sausage is made, it's not unreasonable to be cautious about interpreting these numbers also.[/quote]
First of all, I do enjoy this kabuki dance where DC demands to be treated equivalent to a state, but then should not be compared to states when data is bad but still compared to states when people want to create favorable comparisons.

Second, this is honestly a losers game to try to debunk official government statistics. However, to be clear, regardless of whether or not the total number of people is accurate, what is currently being measured is changes in population. These are two different things and the directional shift is very clear and was evident pre-Covid. The DC government itself estimates that population growth for the five years from 2015-2020 declined from the first five years of the decade and averaged a measly 0.6%. The DC government itself estimates that in 2020 they permanently lost over 9,000 net residents. The official Federal government statistics are that the population loss in 2020 was approximately 15,000 and in 2020 it was 20,000.

Argue what the “true” numbers are all you want but the direction is clear and undisputed. [/quote]

Um, I work for a statistical agency. Not on this particular set of statistics, but I'm not completely ignorant of the issues at play, nor of the many data sources that might be used to make estimates like this. I have no doubt that the creators of this data series are doing the best they can with the data they have access to. I'm also not saying that these numbers aren't correct, or even that the changes aren't directionally correct. It's not unlikely that they are, though it's also not unlikely that they're not. I'm just saying that the pandemic has thrown the usual methods that are used to produce these types of statistics for a massive loop, and that the various issues that have arisen disproportionately affect the accuracy of statistics in cities like DC. For that reason, we should take these numbers with a much larger than usual grain of salt, at least until we've had time to observe another year or two. It's way premature to assert that the sky is falling.[/quote]
Caveat, caveat, caveat… “therefore” blah, blah, blah. Okay buddy.

It literally does not matter if DCs current population is 710,000 or 670,000. What matters is the delta over the past 12-24 months. The question is whether DCs population is increasing or decreasing. More people equals broadened tax base and growing economy, less people can signal the inverse.

While there are questions about the methods for deriving the census number, the methods for estimating the delta are different. While I applaud the effort, if as you state you don’t disagree on the direction, I am curious what it is that you are doing except knowingly spreading FUD?[/quote]

I'm stating that I neither agree nor disagree, and that we should take these numbers as a best guess while acknowledging that the data are not currently good enough to make meaningful inferences about differences between urbanized and non-urbanized areas.

I can't give too many concrete details without outing my employer and my work. But to give you a taste of the issues; most statistics like these that aren't based on surveys combine information from multiple sources.  My guess in this case would be that they use data from government records, change of address forms, and/or commerical mailing databases. Sources are often a snapshot of information from a different time than the one that you really want to measure, and there are often multiple addresses associated with a single person. Sometimes you might not get really good information about where a person lived at a particular time until well after they lived there (like the address where you filed your taxes for last year). So, you look at the records you have for each person and try to decide which one is the best indicator of where they really were at the time you're trying to measure using the data you have.

In normal times, any error with this is just a level effect, so maybe not such a big deal. But with a pandemic, your usual priors about where a person "really" lives when you see multiple addresses for them probably shouldn't work the same way. For example, people might put in a mailing change of address form for a lengthy stay with family or friends even if they don't intend to relocate permanently. People with second homes may be more likely to be reassigned to their second home even if they intend to return to their primary residence. People who stopped working because of the pandemic might no longer have tax records with their address on them. Those things affect the change estimates, not just the level estimates, because they're occuring much more often than before. Add in the specifics of 2021 (widespread vaccination in the Spring and an anticipation of return to office that was delayed by the Delta and Omicron waves), and you end up with a year where population changes are especially hard to estimate.[/quote]
This post is honestly a pretty sad waste of time. Lots of conjecture and no facts. Would love to see you tell your colleagues in the Census Bureau that you don’t trust the quality of their work on no other basis than because it doesn’t feel right to you. Like I said, spreading FUD. [/quote]

What makes you think they don't think the same thing? When a release is required by statute or policy, you do the best you can with the resources you have. That doesn't mean that the output is beyond reproach.[/quote]
And yet, you have no evidence as the basis for your claim about the quality of the data collection for population migration. You have absolutely nothing. Again, this is pretty sad. [/quote]

Ok, I finally had the time to review the methodology release that came along with these statistics. This is as much information as literally anyone will be able to provide you without undergoing a months long review process:

https://www2.census.gov/programs-surveys/popest/technical-documentation/methodology/2020-2021/methods-statement-v2021.pdf

The issues are basically just as I described; the authors are doing their absolute best with an unprecedented situation and a methodology that was never designed to be robust to the effects of a pandemic.

- The baseline levels themselves have potentially significant issues. Many of the source numbers from the Decennial that the authors would normally use still haven't been approved through the usual quality control channels. The authors developed a workaround that makes use of some of the data sources that are normally used for quality control only. This methodology is brand new so it's hard to know how it's affecting baseline estimates, or how closely we can compare to previous years (p. 3).
- The data sources used to observe migration are only a few, and they are very laggy. The primary for working-age people is tax return data, which is only observed once a year and actually reflects where people were in the previous year. NBD in most years, but estimating 2021 population changes by where people spent 2020 and 2019 is problematic. The effect is to make it look like 2020 population movements persisted into 2021, even though they may well have stopped or even reversed by now as renters return. The authors probably don't have access more timely data sources that would resolve this issue, and even if they did it would be difficult to incorporate them and compare to previous years without an extensive quality control analysis (p. 6).
- The methodology assumes that the migration rate of people in their data sources (people who file tax returns) is the same as the migration rate of people who don't. In most years, not much of a problem. In a pandemic, most migration is likely of people who are working remotely without switching jobs. Since non-earners are less likely to migrate in this circumstance, this has the effect of overstating the amount of outmigration from places where employees left while doing WFH. That is, it makes the situation in DC and other high-cost cities look worse than it is (p. 7). If WFH folks return, this will then have the opposite effect, making DC look like it's growing faster than it is.

I'm not going to belabor this any further--not least because I don't think you're arguing in good faith--but I think the point is made. None of this is an indictment of the authors of this data series, but I think it's important to recognize that we really shouldn't have nearly as much confidence in these numbers as we might normally, and most of the conclusions being drawn on this thread are therefore premature. Don't be surprised to see headlines 12-24 months from now claiming that DC has the fastest population growth rate in the country once some of these data limitations and statistical issue work their way out, which we will also have to take with a similar grain if salt. We'll have a better sense of the long-term effect of the pandemic in a couple of years, but in the meantime we should all recognize that the pandemic has made it a lot harder to do even seemingly simple statistical analysis in many cases.[/quote]
Let me repeat. You are filibustering because the you have no other recourse to actual evidence to defend your claims regarding the quality of migration estimates. You have a bunch of assumptions about the impact of the pandemic and the population of movers but nothing else.

And let’s not forget that this started because you were citing issues with the census and not the estimate of movers. As I’ve pointed out, it’s apples and oranges and the actual total population is completely irrelevant to what is being measured here. [/quote]

Options

	Disable HTML in this message
	Disable BB Code in this message
	Disable smilies in this message

Review message

Search Recent Topics Hottest Topics