The Official 2020 Census Thread

[jimrtex]

The Census Bureau thanks the Americans today:

The End of 2020 Census Data Collection, Next Steps, and a Heartfelt Thanks

https://www.census.gov/newsroom/blogs/director/2020/10/the_end_of_2020_cens.html

The Census Bureau had a rather oddly titled news release:

"U.S. Census Bureau Statement on Requests for Additional 2020 Census Metrics and Upcoming Data Quality Measures"

https://www.census.gov/newsroom/press-releases/2020/2020-census-metrics-data-quality-measures.html

It can interpreted as meaning, "quit bugging us, you're not getting any more from us now".

The Census Bureau issued an extraordinary update of self-response rates on Saturday (they had stopped doing weekend updates in June). They also last updated the NRFU completion and coverage completion on Saturday.

https://www.census.gov/newsroom/press-releases/2020/2020-census-all-states-top-99-percent.html

They announced a national completion rate of 99.98% which would amount to about 30,000 housing units (+/- 7500). 99.0% in Louisiana would results in 20,000 incomplete housing units (+/- 1000).

[jimrtex]

I will be posting selected daily reports. This is for September 25, when Massachusetts reached its 2010 response rate.

Thirty-two-day averages (8/24/2010 to 9/25/2010)

Total 1.94 (0.061% per day)
Internet 1.43% (0.045% per day)
Paper+Phone 0.51% (0.016% per day)

At this rate, the USA would reach its 2010 response rate of 66.5% on September 28. It is estimated to take 3.30 days, and there are 6 days remaining.

Two states reached their 2010 response rate in June: MI(15) and WA(30)
Two states reached their 2010 response rate in July:  NV(20) and KY(27)
Eight states reached their 2010 response rate in August: ID(1), CO(17), OR(21), UT(22), VA(24), NH(25), AZ (29) and OH (31).
Thirteen states have reached their 2010 response rate in September: MD(3), MN(3), NJ(7), CT(8), ME(8), IN(11), CA(14), NE(14), AL(15), IL(18), DE(21), FL(22), MA(25)

Massachusetts reached its 2010 response rate on September 25 (gap is 0.1%)

Vermont is projected to reach its 2010 response rate on September 28 (gap is 0.2%)
South Dakota is projected to reach its 2020 response rate on September 28 (gap is 0.2%).

No states are projected to reach their 2010 response rate in the remainder of September, resulting in 27 states (53%) reaching their 2010 response rate during the official enumeration period.

The following eight states would reach within 28 days.

Georgia is projected to reach its 2010 response rate on October 4.
Rhode Island is projected to reach its 2010 response rate on October 6.
New York is projected to reach its 2010 response rate on October 8.
Kansas is projected to reach its 2010 response rate on October 11.
Pennsylvania is projected to reach its 2010 response rate on October 19.
New Mexico is projected to reach its 2010 response rate on October 21.
Alsaka is projected to reach its 2010 response rate on October 23.
Tennessee is projected to reach its 2010 response rate on October 23.

4 states in remainder of October.
7 states + DC in November
2 states in December
2 state in 2021, West Virginia on February March 9

Reached (25)

Green-9: in July (2): MI, WA
Green-8: in August (10): NV, KY, ID, CO, OR, UT, VA, NH (25), AZ, OH
Green-7: in September (13): MD(3), MN, NJ(7), CT(8), ME, IN, CA, NE, AL, IL, DE(21), FL, MA(25)

Projected (26)

Green-6: By September 30 (3): VT(28), SD
Red-3: By October 23, 28 days (8): GA, RI(6), NY, KS, PA, NM, AK, TN
Red-4: October (4): NC, LA, TX, OK
Red-5: November (8): IA, HI, MS, WI, WY, DC, AR, MO
Red-6: December (2): SC, MT
Red-7: 2021 (2): ND, WV

[jimrtex]

The district court in California (collection deadline, etc.) has set an October 27 deadline for the plaintiffs to file an amended complaint, with government reply by November 10.

This will probably drag out past December 31, given that the 9th Circuit and Supreme Court are likely to stay any injunction.

[jimrtex]

I will be posting selected daily reports. This is for Monday September 28, when 4-week rates are reset.

Thirty-five-day averages (8/24/2010 to 9/28/2010)

Total 2.02 (0.058% per day)
Internet 1.51% (0.043% per day)
Paper+Phone 0.51% (0.015% per day)

Twenty-eight-day averages (8/31/2010 to 9/28/2010)

Total 1.48% (0.053% per day)
Internet 1.06% (0.038% per day)
Paper+Phone 0.42% (0.015% per day)

At this rate, the USA would reach its 2010 response rate of 66.5% on September 29. It is estimated to take 1.89 days, and there are 3 days remaining.

Two states reached their 2010 response rate in June: MI(15) and WA(30)
Two states reached their 2010 response rate in July:  NV(20) and KY(27)
Eight states reached their 2010 response rate in August: ID(1), CO(17), OR(21), UT(22), VA(24), NH(25), AZ (29) and OH (31).
Thirteen states have reached their 2010 response rate in September: MD(3), MN(3), NJ(7), CT(8), ME(8), IN(11), CA(14), NE(14), AL(15), IL(18), DE(21), FL(22), MA(25)

South Dakota is projected to reach its 2020 response rate on September 30 (gap is 0.1%).

No states are projected to reach their 2010 response rate in the remainder of September, resulting in 26 states (51%) reaching their 2010 response rate during the official enumeration period.

The following eight states would reach within 28 days.

Vermont is projected to reach its 2010 response rate on October 2 (gap is 0.2%)
Georgia is projected to reach its 2010 response rate on October 3 (gap is 0.4%)
Rhode Island is projected to reach its 2010 response rate on October 9.
New York is projected to reach its 2010 response rate on October 12.
Kansas is projected to reach its 2010 response rate on October 19.
Pennsylvania is projected to reach its 2010 response rate on October 23.
New Mexico is projected to reach its 2010 response rate on October 23.
Louisiana is projected to reach its 2010 response rate on October 24.

2 states in remainder of October.
9 states + DC in November
3 states in December
2 states in 2021, West Virginia on May 3.

Reached (25)

Green-9: in July (2): MI, WA
Green-8: in August (10): NV, KY, ID, CO, OR, UT, VA, NH (25), AZ, OH
Green-7: in September (13): MD(3), MN, NJ(7), CT(8), ME, IN, CA, NE, AL, IL, DE(21), FL, MA(25)

Projected (26)

Green-6: By September 30 (1): SD
Red-3: By October 26, 28 days (8): VT(2), GA, RI(6), NY, KS, NM, PA, LA
Red-4: October (2): NC, TN
Red-5: November (10): AK, TX, OK, IA, HI, MS, DC, WY, AR, MO
Red-6: December (3): WI, SC, MT
Red-7: 2021 (2): ND, WV

[jimrtex]

I will be posting selected daily reports. This is for September 29, when the USA reached its 2010 self-response rate of 66.5%.

Twenty-nine-day averages (8/31/2010 to 9/29/2010)

Total 1.55% (0.053% per day)
Internet 1.11% (0.038% per day)
Paper+Phone 0.43% (0.015% per day)

The USA reached its 2010 response rate of 66.5%

Two states reached their 2010 response rate in June: MI(15) and WA(30)
Two states reached their 2010 response rate in July:  NV(20) and KY(27)
Eight states reached their 2010 response rate in August: ID(1), CO(17), OR(21), UT(22), VA(24), NH(25), AZ (29) and OH (31).
Thirteen states have reached their 2010 response rate in September: MD(3), MN(3), NJ(7), CT(8), ME(8), IN(11), CA(14), NE(14), AL(15), IL(18), DE(21), FL(22), MA(25)

No states are projected to reach their 2010 response rate in the remainder of September, resulting in 25 states (49%) reaching their 2010 response rate during the official enumeration period.

The following ten states would reach within 28 days.

South Dakota is projected to reach its 2020 response rate on October 1 (gap is 0.1%).
Georgia is projected to reach its 2010 response rate on October 3 (gap is 0.3%)
Vermont is projected to reach its 2010 response rate on October 3 (gap is 0.2%)
Rhode Island is projected to reach its 2010 response rate on October 10.
New York is projected to reach its 2010 response rate on October 11.
Kansas is projected to reach its 2010 response rate on October 20.
Pennsylvania is projected to reach its 2010 response rate on October 21.
New Mexico is projected to reach its 2010 response rate on October 23.
Louisiana is projected to reach its 2010 response rate on October 25.
North Carolina is projected to reach its 2010 response rate on October 26.

2 states in remainder of October.
7 states + DC in November
4 states in December
2 states in 2021, West Virginia on May 11

Reached (25)

Green-9: in July (2): MI, WA
Green-8: in August (10): NV, KY, ID, CO, OR, UT, VA, NH (25), AZ, OH
Green-7: in September (13): MD(3), MN, NJ(7), CT(8), ME, IN, CA, NE, AL, IL, DE(21), FL, MA(25)

Projected (26)

Green-6: By September 30 (0):
Red-3: By October 26, 28 days (10): SD, GA, VT(2), RI(6), NY, KS, NM, PA, LA, NC
Red-4: October (2): TN, AK
Red-5: November (8): TX, IA, OK, HI, MS, DC, WY, AR
Red-6: December (4): MO, WI, SC, MT
Red-7: 2021 (2): ND, WV

[jimrtex]

The Census Bureau held a press briefing on completion of field operations and things coming forward in the future.

https://www.census.gov/newsroom/press-kits/2020/2020-census-operational-press-briefing-october21.html

The audio lasts 1 hour. A transcript is expected to be released:

https://broadcast.census.gov/comm/2020census/audio/op-brief/20201021-op-brief-2020-census.mp3

Things I found interesting:

The press officer, opened by saying field operations had been completed "back on October 15" as if it were several months ago.

NRFU resolved 1.92 cases/enumerator-hour compared to 1.01/hour in 2010. This was attributed to use of electronic devices, which included optimization of assignments, and recording of enumeration, vs. paper forms in 2010 and enumerators being given a list of targets.

There were 435K field enumerators total, with a peak active of 235K on August 28. There were 14K field supervisors (roughly 3600 NRFU cases per supervisor, 57 field supervisors per ACO, and 213 housing units per peak enumerator.

They were particularly proud that there had been zero downtime for their Internet operations, especially after there had been suggestions that this might become  healthcare.gov 2.0.

9 million of the 99 million self-responses were triggered by NRFU operations.

There were 400,000 partners (government agencies and private sector groups).

There were 53,000 locations associated with the homeless county, including 37,000 locations where people might be sleeping rough (the other 16,000 were soup kitchens, homeless shelters, and the like). This compares to 19,000 and 8000 in 2010. Apparently, there was a significant inventory since the Spring.

There were 215K group quarters compared to 166K in 2010.

The jump in completions in Shreveport, LA ACO was do to use of high quality administrative records (IRS, SSA, Medicare, 2010). Apparently, after the first knocking, they held up repeat visits if they had high quality administrative records. If they had a chance, they would go back later, but if they were not able, they'd stick with the administrative records.  This likely was the cause of closeout in other ACO's as well where completion jumped from 98% to 99% or so, to 99.9% in the last few days.

In Louisiana, they got a list of motels being used for evacuees from Laura, and enumerated people in those locations.

Processing for the apportionment count is being separated from the redistricting data, in an effort to meet the December 31 deadline. In addition operations will continue on weekends and holidays.

It sounds like they have extremely shoe-horned this into meeting the deadline, and they would not commit to actually meeting it. They seemed even less certain about meeting the deadline for the redistricting data.

There is a group outside the 2020 Census group working on the the "President's memo" which is how it was referred to by everybody. It is unknown how or if the two sets of numbers will be presented.

There were a few questions apparently intended for political purposes.

One questioner wanted to know how many enumerators became infected with COVID-19 and how many homeowners they infected. Another wanted to know how they could possibly begin processing before the final mailed-in paper forms straggled in. Another wanted to know that "since it has been reported" that enumerators had been instructed to cut corners, how would they account for those that were missed. The Census Bureau Fontenot said that imputation had been used since 1990. There was a correction added to the website, saying that imputation had been used since 1960.

[Tender Branson]

What would happen for example if the Census Bureau finishes the apportionment counts in early to mid-January and present one set with the normal population state counts and one set with some dubious removals of non-citizens, using patchy administrative data ?

Assuming Biden takes office a few days later, which set would be used for apportionment for the next 10 years ?

If the non-citizen set of data would only include people in ICE centers about to be deported, I could live with it. But not if there's a more broader non-citizen purge based only on some crappy estimates, instead of actual enumeration of said non-citizens.

[jimrtex]

What would happen for example if the Census Bureau finishes the apportionment counts in early to mid-January and present one set with the normal population state counts and one set with some dubious removals of non-citizens, using patchy administrative data ?

Assuming Biden takes office a few days later, which set would be used for apportionment for the next 10 years ?

If the non-citizen set of data would only include people in ICE centers about to be deported, I could live with it. But not if there's a more broader non-citizen purge based only on some crappy estimates, instead of actual enumeration of said non-citizens.

This case may be of interest.

Franklin v Massachusetts

Congress may always change the apportionment law.

This is a transcript of the press briefing. Note that they said that they would not directly address any ongoing litigation.

Transcript: 2020 Census Operational Press Briefing — October 21, 2020 (PDF)

[Tender Branson]

Enumeration is now completely over.

Paper questionnaires postmarked by Oct. 15 were received until Oct. 22 to be processed.

There may be additional quality checks and assessments, but that's not actual enumeration. Just to see if the data that was gathered is of high quality and how it relates to Census 2010 and Census 2000 data quality.

[jimrtex]

Enumeration is now completely over.

Paper questionnaires postmarked by Oct. 15 were received until Oct. 22 to be processed.

There may be additional quality checks and assessments, but that's not actual enumeration. Just to see if the data that was gathered is of high quality and how it relates to Census 2010 and Census 2000 data quality.

In the district court case, the plaintiffs and Census Bureau have filed their interpretations of the Supreme Court's stay.

The Census Bureau argues that the stay recognized the statutory obligation of the Census Bureau to meet the December 31 deadline that has been set by Congress, and that any harms caused to the plaintiffs by post-processing are speculative and can be litigated after the apportionment counts have been released (this has been the past practice where states (e.g. Montana, Massachusetts, Utah) have found themselves on the short end of the count, and sued.

The plaintiffs argue that the court should hold compressed procedure that would distract the Census Bureau from meeting deadlines and conclude with a trial ending on Christmas Eve.

Meanwhile in the 9th Circuit the Census Bureau's response frames the issue as:

Whether the district court erred in enjoining the Census Bureau from conducting the final phases of the decennial census on a schedule designed to meet the unambiguous and valid statutory deadlines governing the reporting of the census to the President, and subsequent reports to Congress, for purposes of apportionment, and to States, for purposes of redistricting.

Note that the statute requires the Secretary of Commerce (i.e. Census Bureau) report the apportionment population to the President by December 31, 2020 (9 months from Census) and for the President to calculate the apportionment and report it to Congress by January 10, 2021. Redistricting data must be reported to States by March 31, 2021 (12 months from Census). Congress also set the April 1, 2020 Census Date.

[Tender Branson]

Bureau update on their processing work:

"Importantly, we’ve not uncovered anything so far that would suggest that the 2020 Census will not be fit for its constitutional and statutory purposes."

https://www.census.gov/newsroom/blogs/random-samplings/2020/11/update-2020-census-data-processing-and-quality.html

[Frodo]

Bureau update on their processing work:

"Importantly, we’ve not uncovered anything so far that would suggest that the 2020 Census will not be fit for its constitutional and statutory purposes."

https://www.census.gov/newsroom/blogs/random-samplings/2020/11/update-2020-census-data-processing-and-quality.html

I wouldn't trust anything coming out of the Census Bureau:

Census takers say they were told to enter false information

Just how truly reliable are their numbers?  And why should anyone trust them now?  

Honestly, given how hopelessly compromised this year's census has proven to be, we need a redo next year. 

[Tender Branson]

Bureau update on their processing work:

"Importantly, we’ve not uncovered anything so far that would suggest that the 2020 Census will not be fit for its constitutional and statutory purposes."

https://www.census.gov/newsroom/blogs/random-samplings/2020/11/update-2020-census-data-processing-and-quality.html

I wouldn't trust anything coming out of the Census Bureau:

Census takers say they were told to enter false information

Just how truly reliable are their numbers?  And why should anyone trust them now?  

Honestly, given how hopelessly compromised this year's census has proven to be, we need a redo next year. 

A) there is no redo next year.

B) while the US Census is always of lower quality than say a Census here in Europe because there is no central population register, the self-response this year is somewhat better than in 2010. Also, these instructions to enter false information on households where nobody was home are probably more isolated cases. And if not, it's a failure of design because of the lack of a population register. In EVERY 10-year census, there are ca. 10% of all households that cannot be enumerated because nobody is home or doesn't want to answer. And where census takers need to submit dubious information from landlords or neighbours. 2020 is not worse in this regard than 2010 or 2000.

[Tender Branson]

Bureau update on their processing work:

"Importantly, we’ve not uncovered anything so far that would suggest that the 2020 Census will not be fit for its constitutional and statutory purposes."

https://www.census.gov/newsroom/blogs/random-samplings/2020/11/update-2020-census-data-processing-and-quality.html

I wouldn't trust anything coming out of the Census Bureau:

Census takers say they were told to enter false information

Just how truly reliable are their numbers?  And why should anyone trust them now?  

Honestly, given how hopelessly compromised this year's census has proven to be, we need a redo next year. 

Statement:

NOVEMBER 09, 2020

The U.S. Census Bureau is very proud of the hundreds of thousands of census takers who performed their duties professionally and efficiently and were responsible for one of the highest national address resolution rates in census history — 99.98%. The Census Bureau knows of no attempts to systemically falsify respondent information. We employed new information technologies and safeguards to prevent and identify mistakes or misreporting of data. We also conducted extensive follow-up quality assurance interviews.

The Census Bureau takes falsification allegations very seriously. Intentional falsification of respondent information by a Census Bureau employee is a serious federal offense, will be fully investigated, and referred for prosecution, if appropriate.

Some alleged incidents reported to the media may represent employment-related disputes and/or misunderstandings of operations. For instance, a household refusing to answer the 2020 Census to one census taker would likely have been reassigned to another census taker or supervisor, which reassignment would not be apparent to the first census taker. The concerns expressed by the initial census taker may not reflect the follow-up work done to resolve the status of the address.

We continued working cases in all areas of the country until the end of data collection. If we did not get at least the basic information for the apportionment count and could not resolve the status of the address, the case was assigned a final status code that showed this information was not available. As we have previously stated, the number of unresolved addresses was exceedingly low — approximately 0.02% nationwide — which will be ultimately resolved by well-accepted methods.

If you have any information or allegations to share with the Census Bureau in regards to data collection, please email rumors@census.gov.

https://www.census.gov/newsroom/press-releases/2020/data-collection-statement.html

[brucejoel99]

Oh sh*t:

https://twitter.com/ElectProject/status/1329493100423761920

[Tender Branson]

Awesome.

Trump has meddled enough already and should never get his hands on those numbers.

[°Leprechaun]

I am wondering what people think about how badly the census will hurt Democrats.

With a razor thin majority in the House, it could make it extremely difficult for them in 2022, right?

[Zaybay]

I am wondering what people think about how badly the census will hurt Democrats.

With a razor thin majority in the House, it could make it extremely difficult for them in 2022, right?

Why the automatic assumption that the census will hurt the Democrats?

[°Leprechaun]

I am wondering what people think about how badly the census will hurt Democrats.

With a razor thin majority in the House, it could make it extremely difficult for them in 2022, right?

Why the automatic assumption that the census will hurt the Democrats?

It would seem to be the case. Am I wrong?

[Tender Branson]

I am wondering what people think about how badly the census will hurt Democrats.

With a razor thin majority in the House, it could make it extremely difficult for them in 2022, right?

Why the automatic assumption that the census will hurt the Democrats?

It would seem to be the case. Am I wrong?

Yes.

The 2020 Census has been largely a success, despite everything this year and the quality is in line with those conducted in 2000 or 2010.

There is still a Supreme Court ruling coming in a few days if the Census Bureau needs to remove non-citizens from the reapportionment data or not ... and if yes, to which extent.

But it's pretty unlikely that the Census Bureau has the necessary data to remove a large part of non-citizens. Maybe they can be ordered to remove people in ICE detention.

[Zaybay]

I am wondering what people think about how badly the census will hurt Democrats.

With a razor thin majority in the House, it could make it extremely difficult for them in 2022, right?

Why the automatic assumption that the census will hurt the Democrats?

It would seem to be the case. Am I wrong?

Unless you have any evidence or any argument for it to be disadvantageous to Democrats, yes.

[Tender Branson]

The only problem for Dems between 2020-30 are the population growth dynamics in the states between 2010-20.

States won by Biden will likely lose 4-8 EV with the new numbers, Trump states will likely net the same amount.

But that has nothing to do with the Census process and quality itself.

[Tender Branson]

In 2 weeks, the Census Bureau will release updated April 1, 2020 population estimates - to compare them to the actual Census count (this will be out in January, February or March).

https://www.census.gov/newsroom/press-releases/2020/news-briefing-demographic-analysis.html

[Mr.Phips]

Bureau update on their processing work:

"Importantly, we’ve not uncovered anything so far that would suggest that the 2020 Census will not be fit for its constitutional and statutory purposes."

https://www.census.gov/newsroom/blogs/random-samplings/2020/11/update-2020-census-data-processing-and-quality.html

I wouldn't trust anything coming out of the Census Bureau:

Census takers say they were told to enter false information

Just how truly reliable are their numbers?  And why should anyone trust them now?  

Honestly, given how hopelessly compromised this year's census has proven to be, we need a redo next year. 

You would need Congressional approval for a redo, which McConnell would never allow a vote on in the Senate.

[Bootes Void]

The only problem for Dems between 2020-30 are the population growth dynamics in the states between 2010-20.

States won by Biden will likely lose 4-8 EV with the new numbers, Trump states will likely net the same amount.

But that has nothing to do with the Census process and quality itself.

for the 2024 election, Texas and Florida will gain some EVs which do lean republican for now. NC,Ga and AZ will gain seats as well but those are tossups. as for dems NV,OR,WA,CO will gain seats as they are growing as well.As a whole, it might be a slight loss or a wash for Biden