Census 2020 apportionments to be released by April 30; redistricting by Sept 30

[muon2]

I just concluded a Zoom call with the Census Bureau. The Bureau's representative said that the Bureau has set a hard date of April 30 for apportionment data. That's consistent with the original pandemic schedule and includes the possibility that it could be released before then, but the schedule made it clear that it wold not be before Mar 27 in any case. If I read between the lines, I would expect it sometime in the last week of Apr.

The Bureau officially has no start or end date for the period to release the redistricting data. The CB is assessing whether to release all states at once, which provides for the most accurate release since everything gets checked the same way. The official said it is highly unlikely that the redistricting data set would be ready before Jul 31. They expect an official timeline in the next few weeks.

The previous administration was going to release concurrent block-level CVAP numbers for Census 2020 based on administrative records. Biden's executive order has stopped that. The only CVAP available for redistricting will be the 2011-2019 5-year ACS data, tied to the 2010 geography. DOJ will have to determine at some much later point what CVAP data is needed to challenge plans made in this cycle.

[Torie]

Why did Biden stop the CVAP numbers? Was there fear that SCOTUS would allow that data to be used for apportionment purposes? I think that you had informed me the CVAP numbers out there are stale and very inaccurate in some places. Without good numbers, how can you get to that "perfect" Goldilocks point that makes you a good VRA warrior?

[Tender Branson]

Why did Biden stop the CVAP numbers? Was there fear that SCOTUS would allow that data to be used for apportionment purposes? I think that you had informed me the CVAP numbers out there are stale and very inaccurate in some places. Without good numbers, how can you get the the Goldilocks point that makes you a good VRA warrior?

The Census Bureau already has to figure out the total population based on 67% self-response and the remaining 33% coming from often dubious efforts to count people, so how should they produce a citizen-only population if citizenship wasn’t even part of the questionnaire ?

Data on non-citizens is very limited and can only be produced by the annual ACS, which is a survey of a few hundred thousand households and no complete enumeration.

All of this is more reason for the US to adopt a CVSR.

[Frodo]

Why did Biden stop the CVAP numbers? Was there fear that SCOTUS would allow that data to be used for apportionment purposes? I think that you had informed me the CVAP numbers out there are stale and very inaccurate in some places. Without good numbers, how can you get the the Goldilocks point that makes you a good VRA warrior?

The Census Bureau already has to figure out the total population based on 67% self-response and the remaining 33% coming from often dubious efforts to count people, so how should they produce a citizen-only population if citizenship wasn’t even part of the questionnaire ?

Data on non-citizens is very limited and can only be produced by the annual ACS, which is a survey of a few hundred thousand households and no complete enumeration.

All of this is more reason for the US to adopt a CVSR.

What does CVSR stand for?  What is it?

[muon2]

Why did Biden stop the CVAP numbers? Was there fear that SCOTUS would allow that data to be used for apportionment purposes? I think that you had informed me the CVAP numbers out there are stale and very inaccurate in some places. Without good numbers, how can you get to that "perfect" Goldilocks point that makes you a good VRA warrior?

My sense is that the decision was largely political in that one side wanted a direct measure of citizenship and the other side was opposed. The fact that the data could be beneficial to either side depending on its application was not material. Note my comment that the CB is punting to DoJ to determine how they might want to prosecute cases with data that doesn't map onto the 2020 Census.

The Census Bureau already has to figure out the total population based on 67% self-response and the remaining 33% coming from often dubious efforts to count people, so how should they produce a citizen-only population if citizenship wasn’t even part of the questionnaire ?

Here's the breakdown I copied of the Non-Response Follow Up which applied to 60.8 M housing units with no self-response.
6.3 M came in with self-response during NRFU.
13.5 M were determined to be vacant
10.3 M were deleted as no longer housing units
Of the remaining housing units 56% were collected by the enumerator at the household, 24% by proxy such as a neighbor or building manager, and 20% by administrative records. That 20% by administrative record was less than planned based on Census 2010. In the end the number of records that had to resolved by imputation was less than 1% of all housing units.

Data on non-citizens is very limited and can only be produced by the annual ACS, which is a survey of a few hundred thousand households and no complete enumeration.

emphasis added

Not entirely true. The federal government has a wealth of records on individuals that point to their citizenship. These administrative records were part of the existing plan to make a more accurate determination than the ACS, but their use was cancelled.

[Tender Branson]

Why did Biden stop the CVAP numbers? Was there fear that SCOTUS would allow that data to be used for apportionment purposes? I think that you had informed me the CVAP numbers out there are stale and very inaccurate in some places. Without good numbers, how can you get the the Goldilocks point that makes you a good VRA warrior?

The Census Bureau already has to figure out the total population based on 67% self-response and the remaining 33% coming from often dubious efforts to count people, so how should they produce a citizen-only population if citizenship wasn’t even part of the questionnaire ?

Data on non-citizens is very limited and can only be produced by the annual ACS, which is a survey of a few hundred thousand households and no complete enumeration.

All of this is more reason for the US to adopt a CVSR.

What does CVSR stand for?  What is it?

A CVSR is a Central Vital Statistics Registry that most European countries have (in Austria, it’s the ZMR/PSR).

Once people get born, their data is entered immediately (incl. citizenship and place of main residence), when they move it is registered and when they die it is registered.

With such an up-to-date registry you can conduct a census every year at 1/10th the cost of a traditional census (or even less costly).

[Kevinstat]

Both chambers of the New Jersey Legislature and the Virginia House of Delegates are up for election this November.  Those elections will almost certainly have to be held on 2010-era lines.  Fortunately the elections in all three chambers will be for two-year terms.  Lower house members in both states serve 2-year terms.  State Senators in New Jersey are elected for 2-year terms in years ending in '1' and for 4-year terms in years ending in '3' and '5'.  State Senators in Virginia serve 4-year terms, period, but are elected in the year before a Presidential election rather than the year after.  Both chambers in Louisiana and Mississippi are the same as the Virginia Senate.  Kentucky, the remaining state with odd-year Gubernatorial elections, has even-year Legislative elections.

[Former President tack50]

Why did Biden stop the CVAP numbers? Was there fear that SCOTUS would allow that data to be used for apportionment purposes? I think that you had informed me the CVAP numbers out there are stale and very inaccurate in some places. Without good numbers, how can you get the the Goldilocks point that makes you a good VRA warrior?

The Census Bureau already has to figure out the total population based on 67% self-response and the remaining 33% coming from often dubious efforts to count people, so how should they produce a citizen-only population if citizenship wasn’t even part of the questionnaire ?

Data on non-citizens is very limited and can only be produced by the annual ACS, which is a survey of a few hundred thousand households and no complete enumeration.

All of this is more reason for the US to adopt a CVSR.

What does CVSR stand for?  What is it?

A CVSR is a Central Vital Statistics Registry that most European countries have (in Austria, it’s the ZMR/PSR).

Once people get born, their data is entered immediately (incl. citizenship and place of main residence), when they move it is registered and when they die it is registered.

With such an up-to-date registry you can conduct a census every year at 1/10th the cost of a traditional census (or even less costly).

Worth noting such a registry is far from infallible. Many of my friends live in the mainland, in places like Madrid or Barcelona; yet they remain registered in the Canaries for various reasons (most notably, flight discounts to come home)

While you are meant to register at your new house, and a majority of people do it, there is a minority who moves without changing the registration

[Tender Branson]

Why did Biden stop the CVAP numbers? Was there fear that SCOTUS would allow that data to be used for apportionment purposes? I think that you had informed me the CVAP numbers out there are stale and very inaccurate in some places. Without good numbers, how can you get the the Goldilocks point that makes you a good VRA warrior?

The Census Bureau already has to figure out the total population based on 67% self-response and the remaining 33% coming from often dubious efforts to count people, so how should they produce a citizen-only population if citizenship wasn’t even part of the questionnaire ?

Data on non-citizens is very limited and can only be produced by the annual ACS, which is a survey of a few hundred thousand households and no complete enumeration.

All of this is more reason for the US to adopt a CVSR.

What does CVSR stand for?  What is it?

A CVSR is a Central Vital Statistics Registry that most European countries have (in Austria, it’s the ZMR/PSR).

Once people get born, their data is entered immediately (incl. citizenship and place of main residence), when they move it is registered and when they die it is registered.

With such an up-to-date registry you can conduct a census every year at 1/10th the cost of a traditional census (or even less costly).

Worth noting such a registry is far from infallible. Many of my friends live in the mainland, in places like Madrid or Barcelona; yet they remain registered in the Canaries for various reasons (most notably, flight discounts to come home)

While you are meant to register at your new house, and a majority of people do it, there is a minority who moves without changing the registration

That doesn't mean the registry is flawed though ...

People are still registered with their main residence somewhere, unlike in the US where a lot of people go unnoticed.

European bureaucracy is also structured in a way that you have to show up in at least some register somewhere, be it the CPR, the foreigner registry, the housing registry, the tax registry, the school registry, social insurance registry, pension payment registry, car registry or what else.

Austria for example used and will use and examine about 20 main registers in it's 2011 and 2021 register census to weed out "Karteileichen" (database bodies, or people that do not exist or are not in the country any longer). After comparison of all of those registers in 2011, slightly less than 1% of the ZMR reference date stock was flagged as "needs to be checked further" because of evidence that this person doesn't live any longer in the country or at all. Of those 70.000 clearing cases, a governmental letter was sent to their main residence and about 60% of them were actually returned after several letters. That left an error rate of about 30.000 out of 8.4 million people, or about 0.4% of the total population.

These are mostly people who went abroad, but didn't de-register in their town.

As a result, Austria now clears its ZMR central population register frequently and produces an annual mini-census, without the quality checks mentioned above. That minimizes the cumulative errors for a 10-year time span. The 2021 Census will cost only 10 Mio. €, instead of almost 500 Mio. € for the last traditional paper census in 2001.

These population registers therefore allow for a more competent census.

In Spain, it's pretty much the same thing as here, but I guess Spain still uses the traditional census despite having a national population register as well.

It's also easier for elections: with a CPR, you can easily screen for every voting-age and voting-eligible person in the country, who are then automatically registered to vote.

[Former President tack50]

These population registers therefore allow for a more competent census.

In Spain, it's pretty much the same thing as here, but I guess Spain still uses the traditional census despite having a national population register as well.

It's also easier for elections: with a CPR, you can easily screen for every voting-age and voting-eligible person in the country, who are then automatically registered to vote.

Yeah, this is all correct, I guess you are right

For the record, like you say Spain does do a more traditional census as well I think

However it is worth noting that Spain's apportionments get done by the CPR based off your registration, which update on a yearly basis. This is also why reapportionment here happens basically every time an election gets called. And like you say, it also allows for automatic voter registration at whichever adress you are registered on.

This does create some niche scenarios where people who move shortly before election day have a hard time figuring out where they have to vote but I guess it's a very small number of cases.

[Tender Branson]

These population registers therefore allow for a more competent census.

In Spain, it's pretty much the same thing as here, but I guess Spain still uses the traditional census despite having a national population register as well.

It's also easier for elections: with a CPR, you can easily screen for every voting-age and voting-eligible person in the country, who are then automatically registered to vote.

Yeah, this is all correct, I guess you are right

For the record, like you say Spain does do a more traditional census as well I think

However it is worth noting that Spain's apportionments get done by the CPR based off your registration, which update on a yearly basis. This is also why reapportionment here happens basically every time an election gets called. And like you say, it also allows for automatic voter registration at whichever adress you are registered on.

This does create some niche scenarios where people who move shortly before election day have a hard time figuring out where they have to vote but I guess it's a very small number of cases.

That's indeed a very small number of people.

And even if you are registered with your main residence in the Canaries or Vorarlberg and move to Madrid or Vienna around the election, you can request a postal ballot and it's sent to your new address.

Most people often just move a few km and can drive into their neighbouring former town to vote.

[Gass3268]

Both chambers of the New Jersey Legislature and the Virginia House of Delegates are up for election this November.  Those elections will almost certainly have to be held on 2010-era lines.  Fortunately the elections in all three chambers will be for two-year terms.  Lower house members in both states serve 2-year terms.  State Senators in New Jersey are elected for 2-year terms in years ending in '1' and for 4-year terms in years ending in '3' and '5'.  State Senators in Virginia serve 4-year terms, period, but are elected in the year before a Presidential election rather than the year after.  Both chambers in Louisiana and Mississippi are the same as the Virginia Senate.  Kentucky, the remaining state with odd-year Gubernatorial elections, has even-year Legislative elections.

Virginia will probably have elections in 2021, 2022, and 2023.

[Sol]

This is random, but has Virginia ever considered moving its elections to normal years?

[Kevinstat]

Both chambers of the New Jersey Legislature and the Virginia House of Delegates are up for election this November.  Those elections will almost certainly have to be held on 2010-era lines.  Fortunately the elections in all three chambers will be for two-year terms.  Lower house members in both states serve 2-year terms.  State Senators in New Jersey are elected for 2-year terms in years ending in '1' and for 4-year terms in years ending in '3' and '5'.  State Senators in Virginia serve 4-year terms, period, but are elected in the year before a Presidential election rather than the year after.  Both chambers in Louisiana and Mississippi are the same as the Virginia Senate.  Kentucky, the remaining state with odd-year Gubernatorial elections, has even-year Legislative elections.

Virginia will probably have elections in 2021, 2022, and 2023.

Can you explain what you mean here?  I mean probably almost everywhere has something on the ballot every year.  If you're saying the Virginia House and/or Senate will be up in 2022 for a 1-year term under new lines, I had heard something about that but I have my doubts.  Were Virginia Senate terms commuted in the early 2000s (decade), when redistricting wasn't delayed, just the Senate wasn't to be up until 2003 after they were last up in 1999?  If Virginia was due to elect people to 4-year terms this year I could see commutation happening (although terms weren't commuted in Mississippi after legislators (both houses) were elected to 4-year terms in from old districts in 2011 (they served until late 2015 or early 2016)).  But my guess is that the State Representative elected this November and the State Senators elected to 4-year terms in November 2019 both serve until late 2023/early 2024.

[Gass3268]

Both chambers of the New Jersey Legislature and the Virginia House of Delegates are up for election this November.  Those elections will almost certainly have to be held on 2010-era lines.  Fortunately the elections in all three chambers will be for two-year terms.  Lower house members in both states serve 2-year terms.  State Senators in New Jersey are elected for 2-year terms in years ending in '1' and for 4-year terms in years ending in '3' and '5'.  State Senators in Virginia serve 4-year terms, period, but are elected in the year before a Presidential election rather than the year after.  Both chambers in Louisiana and Mississippi are the same as the Virginia Senate.  Kentucky, the remaining state with odd-year Gubernatorial elections, has even-year Legislative elections.

Virginia will probably have elections in 2021, 2022, and 2023.

Can you explain what you mean here?  I mean probably almost everywhere has something on the ballot every year.  If you're saying the Virginia House and/or Senate will be up in 2022 for a 1-year term under new lines, I had heard something about that but I have my doubts.  Were Virginia Senate terms commuted in the early 2000s (decade), when redistricting wasn't delayed, just the Senate wasn't to be up until 2003 after they were last up in 1999?  If Virginia was due to elect people to 4-year terms this year I could see commutation happening (although terms weren't commuted in Mississippi after legislators (both houses) were elected to 4-year terms in from old districts in 2011 (they served until late 2015 or early 2016)).  But my guess is that the State Representative elected this November and the State Senators elected to 4-year terms in November 2019 both serve until late 2023/early 2024.

I'm taking that from here:

https://twitter.com/ChazNuttycombe/status/1354508389682991104?s=20

[Kevinstat]

Both chambers of the New Jersey Legislature and the Virginia House of Delegates are up for election this November.  Those elections will almost certainly have to be held on 2010-era lines.  Fortunately the elections in all three chambers will be for two-year terms.  Lower house members in both states serve 2-year terms.  State Senators in New Jersey are elected for 2-year terms in years ending in '1' and for 4-year terms in years ending in '3' and '5'.  State Senators in Virginia serve 4-year terms, period, but are elected in the year before a Presidential election rather than the year after.  Both chambers in Louisiana and Mississippi are the same as the Virginia Senate.  Kentucky, the remaining state with odd-year Gubernatorial elections, has even-year Legislative elections.

Virginia will probably have elections in 2021, 2022, and 2023.

Can you explain what you mean here?  I mean probably almost everywhere has something on the ballot every year.  If you're saying the Virginia House and/or Senate will be up in 2022 for a 1-year term under new lines, I had heard something about that but I have my doubts.  Were Virginia Senate terms commuted in the early 2000s (decade), when redistricting wasn't delayed, just the Senate wasn't to be up until 2003 after they were last up in 1999?  If Virginia was due to elect people to 4-year terms this year I could see commutation happening (although terms weren't commuted in Mississippi after legislators (both houses) were elected to 4-year terms in from old districts in 2011 (they served until late 2015 or early 2016)).  But my guess is that the State Representative elected this November and the State Senators elected to 4-year terms in November 2019 both serve until late 2023/early 2024.

I'm taking that from here:

https://twitter.com/ChazNuttycombe/status/1354508389682991104?s=20

Okay, thanks.  The post doesn't say that State Senate seats will be up next year (possibly because that was due to be a 4-year term under the current lines even before COVID-19, and no non-special State Senate elections will be held there this year, using the old lines even though they would normally use new ones).

[Frodo]

Why did Biden stop the CVAP numbers? Was there fear that SCOTUS would allow that data to be used for apportionment purposes? I think that you had informed me the CVAP numbers out there are stale and very inaccurate in some places. Without good numbers, how can you get the the Goldilocks point that makes you a good VRA warrior?

The Census Bureau already has to figure out the total population based on 67% self-response and the remaining 33% coming from often dubious efforts to count people, so how should they produce a citizen-only population if citizenship wasn’t even part of the questionnaire ?

Data on non-citizens is very limited and can only be produced by the annual ACS, which is a survey of a few hundred thousand households and no complete enumeration.

All of this is more reason for the US to adopt a CVSR.

What does CVSR stand for?  What is it?

A CVSR is a Central Vital Statistics Registry that most European countries have (in Austria, it’s the ZMR/PSR).

Once people get born, their data is entered immediately (incl. citizenship and place of main residence), when they move it is registered and when they die it is registered.

With such an up-to-date registry you can conduct a census every year at 1/10th the cost of a traditional census (or even less costly).

Gotcha, thanks. 

[100% pro-life no matter what]

Are the projections typically accurate, or is there a realistic chance of surprises?

[Roronoa D. Law]

Are the projections typically accurate, or is there a realistic chance of surprises?

If I'm not mistaken the only that is still in the air is CA/MN losing a seat and MT gains a seat.

[jimrtex]

This is random, but has Virginia ever considered moving its elections to normal years?

At the time of the Constitution most state legislatures were elected for one-year terms. In Rhode Island it was 6-months. In the Federalist it was argued that two-year terms were warranted for the federal House of Representatives because of travel times, and the need to become acquainted to the national situation.

Most state legislatures met for a very short while (a few weeks) in the Winter, when the dirt roads were frozen rather than mud, and farmer-representatives might be able to take off, if they had someone to milk the cows. Ordinarily, they would be repairing equipment before the Spring planting.

Elections were simple then. Men folk, or those who owned enough property, would meet, take a chaw of tobacco, a swig of hard cider, and cast their vote. Sometimes by paper ballot, other times viva voce. This could be done a few weeks before the legislature met.

Later, terms were increased to two years or even four years. What year elections were held might depend on when the change took effect. If a constitutional amendment was approved in an even year, they new terms might take effect in the odd-year. This may be the case for executive offices, and explain why some states such as North Carolina and Washington hold gubernatorial elections coincident with presidential elections which seem weird if not abnormal to me.

It wasn't until the 1872 that congressional elections were required to be in November - and this was not universally implemented until almost a century later. States which might have held state elections in September or October, might move these elections to match the federal schedules. This might simply be a matter of political expediency as any real economy.

[cinyc]

Are the projections typically accurate, or is there a realistic chance of surprises?

If I'm not mistaken the only that is still in the air is CA/MN losing a seat and MT gains a seat.

That's incorrect. CA is losing a seat. MN losing a seat is also likely. I think the main bubble is NY losing 2 seats vs AL losing 1.

[jimrtex]

Are the projections typically accurate, or is there a realistic chance of surprises?

If I'm not mistaken the only that is still in the air is CA/MN losing a seat and MT gains a seat.

That's incorrect. CA is losing a seat. MN losing a seat is also likely. I think the main bubble is NY losing 2 seats vs AL losing 1.

I think the "surprise" would be more in longer term trends. I doubt anyone in 2011 would say that California would lose population in absolute numbers in 2020, or that the decline in population growth in New York would decline and go negative to such an extent that the decade will be negative.

Another factor is the overall decline in both national and state growth. You might estimate a state will grow 60% and find out that it was actually 70%, especially if that 10% represents a few 10,000 persons. It is much less likely to estimate 5% and find out that it was actually 15%.

The Census Bureau has also refined their estimation procedures, figuring out in particular how to estimate migration, both interstate and international, getting access to tax and social security records (that have been anonymized). Birth and death records are reasonably accurate, but there is not a direct measurement of migration.

Some state estimates have proven to be more accurate because they are based on the number of houses. You can get a quite accurate count of houses in Arizona because they are in the property tax records. This gives an estimate of migration to Arizona because people live in Houses.

[hyouzel the predictor]

2020s Congressional patch notes incoming

[Epaminondas]

Are the projections typically accurate, or is there a realistic chance of surprises?

If I'm not mistaken the only that is still in the air is CA/MN losing a seat and MT gains a seat.

That's incorrect. CA is losing a seat. MN losing a seat is also likely. I think the main bubble is NY losing 2 seats vs AL losing 1.

That's also the impression I was under.
Turns out that with Tenney winning in NY-22, the partisan balance wouldn't really be affected as a GOP seat will be cut either way.

[muon2]

I just received an announcement from the Census has set a timeline for release of redistricting data. The redistricting data will be released by Sept 30 and will be released to all states at the same time.