, 86 tweets, 16 min read Read on Twitter
"And that points out why he preferred asking the question and getting actual answers from 22.2 million people at 98 percent accuracy, because you simply didn't have administrative records for 35 million people, and the bureau had not yet figured out how to do that estimation."
"He then goes on to say on that same page, more than 10 percent of the American population, some 25 million voting-age people, would need to have their citizen age imputed by the Census Bureau. And so he was making clear that he'd rather go with actual counting than imputation."
A little bit further down:
Govt Attorney: "I think the Secretary fully acknowledged that there was an upside to the request, and the upside was the one the Dept of Justice set forth in his letter, that having citizenship data would help improve Voting Rights Act enforcement. "
He fully understood there was an alternative using administrative records, and he analyzed that alternative in the language that I just read to Justice Breyer, and he understood there was a downside, that adding the citizenship question would potentially increase self-response.."
"...-- decrease -- increase the number of -- decrease the number of self-response rates.

But he found two things with respect to that. First, he found -- and all of this is in his letter -- that he could mitigate that to at least a certain extent with follow-up operations,..."
"... perhaps not entirely but at least to a certain extent, and, secondly, to the extent that materialized, it was the product of illegal activity. "
JUSTICE KAVANAUGH: -- can you explain
how it would improve Voting Rights Act
enforcement?

GENERAL FRANCISCO: Yes, Your Honor. One of the critical elements of Voting Rights Act enforcement is something called Citizen Voting Age Population, or CVAP.
Right now, everything for CVAP comes from the census, with the exception of citizenship. So population, age, race, all of that comes from the census except for citizenship, the C in CVAP.
So a large amount of voting rights litigation focuses on expert witnesses who try to fill in that missing C and try to estimate that missing C through imputation based on the American Community Survey, which goes to just one in 38 households.
And the Department of Justice wanted to get all of the same information from the same database so that critical feature of voting rights litigation, CVAP, all came from the same place.
***I am going to have to stop for a while but I will come back to this. I'm invested now***
***Page 36/103***
***I can also say, after Attorney General finishes his oral argument they could pretty much be done, because Justices Ginsburg, Sotomayor, and Kagan have been arguing New York's case for them.***
And under 6(c), under my -- my friend's on the other side's position, you
actually couldn't even ask the citizenship question on the American Community Survey.
And you also couldn't ask about sex and age on the census itself since all of that information is all also available in administrative records.
But the reason why administrative records are insufficient under 6(c) for any of these purposes is for the simple reason that you don't have them for 35 million people.
***after some exchange between Attorney and Sotomayor***

And in the face of two competing possibilities, either asking the question, getting answers for two-thirds of the people for whom no administrative records existed,
at 98 percent accuracy, or using an estimation model that had not yet been created and had an unknown error rate, the Secretary reasonably chose to go...with the bird in the hand.
***Now to Barbara Underwood for New York***

And he said that adding the question would help voting rights enforcement. But that claim is unsupported by the record as well.
CHIEF JUSTICE ROBERTS: Do you -- do you think it wouldn't help voting rights enforcement? The CVAP, Citizen Voting Age Population, is the critical element in voting rights enforcement, and this is getting citizen information.
MS UNDERWOOD: ...the evidence before him was that it would not give better citizenship information than - that it's the 22 million that the government points to, the 22 million whose citizenship information will be either modeled or the result of the answer to a census question.
***down a ways***
JUSTICE ALITO: On the modeling, there was a lot of talk during the first part of the argument about -- I think it's 22.6 million people who it would -- it is predicted would answer the citizenship question and as to whom there is not administrative data.
And there was an estimate that those answers would be 98 percent accurate. And the comparison then has to be between that 98 percent predicted accuracy rate and whatever the accuracy rate would be for the model.
And is there anything in the administrative record that shows that the model was tested and that it was possible to extract a -- a predicted error rate for the model?
MS. UNDERWOOD: What we have -- what we have is that this -- the model hadn't been generated, but what we have is the Census Bureau saying this is like other modeling that we routinely do. We're confident that we can do it.
JUSTICE ALITO: So, if the Secretary is -- has to choose between two things, and, on one, the Secretary knows there's a 98 percent accuracy rate, and as to the other, the Census Bureau says we're going to create a model, and we don't know how --
we can't give you any statistics, but trust us, it's going to be more accurate than 98 percent, is it arbitrary and capricious for the Secretary to say, I'll go with the 98 percent because that's a known quantity?
MS. UNDERWOOD: If there were no cost to the enumeration, that would be a different question. But, when there is this much uncertainty, then it is arbitrary and capricious to take that kind of risk
...
JUSTICE BREYER: I don't understand uncertainty. I thought the 98 percent -- of course, it's 98 percent. Most people are citizens.

MS. UNDERWOOD: Right.
JUSTICE BREYER: The people who are citizens are not going to -- you know, they'll say they're citizens. All you'd ever expect are a few percent who are not citizens.
Then I have on pages J -- Joint Appendix 882 through 884 Mr. Abowd's testimony where he unequivocally says three times that in -- in -- not 98 percent -- in respect to those people who are not citizens,
the administrative model will be more accurate than just asking the census question, and if you add the census question, then you look to it for the answer, you will discover that you are less accurate in respect to non-citizens.
Now he says that. That's why I asked the Solicitor General what is their contrary to that, and he gave me things to look at. And I would say the most contrary thing, which I want to ask you about, is Dr. Abowd, at the trial, said,
***Emphasis***

"There's no credible quantitative evidence that the addition of a citizenship question will affect the accuracy of the count." All right? That's what he said.

MS. UNDERWOOD: Yes.
JUSTICE BREYER: Now I'm sure in their reply brief they pointed right to that. You can't simply ignore it. And so I want to hear what your answer is to that, which the government says is contrary evidence.
MS. UNDERWOOD: If you look at that testimony in context, it is perfectly clear that what he is saying is that he didn't have enough evidence for a firm quantitative statement, meeting scientific standards. He actually defined the term "credible quantitative evidence."
***Down a ways***

MS. UNDERWOOD: -- while I think there is good evidence, and nothing contrary, that this 22 million would be more accurately identified by the modeling than by the census,
I think it is sufficient for this purpose to treat it as somewhat uncertain because it is uncertainty with respect to the discretionary part of what the Census Bureau does; namely, collect extra information.
The core function of the census, not of the Census Bureau in all its actions, but of -- on the census form
JUSTICE KAVANAUGH: But the -- the United Nations recommends that countries ask a citizenship question on the census. And a number of other countries do it. Spain, Germany, Canada, Australia, Ireland, Mexico ask a citizenship question.
And the United States has asked a citizenship question, as you know, in one form or another since 1820, excluding 1840. And, again, long form at times, in more recent times, and then on the ACS since 2005.
The question is, does that international practice, that U.N. recommendation, that historical practice in the United States, affect how we should look at the inclusion of a citizenship question in this case?
MS. UNDERWOOD: The same guidance from the U.N. also says to be careful to test questions to make sure they don't interfere with the enumeration. It says you need to make a judgment in context.
It may be that those countries either haven't examined or don't have the problem that has been identified -- the problem of depressing the enumeration that the United States has.
It's certainly something to look at, but --

JUSTICE KAVANAUGH: But you agree it's very -- it's a very common question internationally?
MS. UNDERWOOD: Well, it is certainly useful information for a country to have. And I'm not suggesting at all that that information shouldn't be collected.

***Sorry but I've got to leave you on this cliff hanger. Only page 56 if you want to read ahead! Hard not to post it all!**
***AAAAND we're back!***

The question is whether it should be collected on the very instrument that is -- whose principal function is to count the population, when we have such strong evidence that it will depress that count, make it less accurate, & make it less accurate in a--
CHIEF JUSTICE ROBERTS: Well, the principal purpose -- you're -- you're right, the principal purpose is to count the population, but we've had demographic questions on the census -- I don't know how far back, but, certainly, it's quite common.
MS. UNDERWOOD: That's -- that's correct, but we have no evidence about--

CHIEF JUSTICE ROBERTS: Sex, age, things like that. You go back and it looks -- you know, do you -- do you own your house? Do you own a radio? I mean, the questions go quite beyond how many people there are.
JUSTICE ALITO: ...A lot of your argument and a lot of the district court's argument seems to hinge on this prediction that there will be 5.1 percent fewer responses if the citizenship question is included on the census.
But that seems -- that is based, as I understand it, on the fact that non-citizens are somewhat less likely to complete the ACS, which includes the citizenship question, than are citizens. Am I right in understanding that? That's fundamentally where that comes from?
MS. UNDERWOOD: It's not about not completing. It's about not -- I mean, it's not about skipping questions on a form.

JUSTICE ALITO: Not -- not responding.

MS. UNDERWOOD: It's about not responding. Yes.
JUSTICE ALITO: That's correct. Okay. They are somewhat less likely to respond to the ACS than are --

MS. UNDERWOOD: The ACS in one study and the long form in another.
JUSTICE ALITO: Okay. But what jumps out is the fact that citizens and non-citizens differ in a lot of respects other than citizenship. They differ in socioeconomic status. They differ in education.
They differ in language ability. of a statistician to wonder about the legitimacy of concluding that there is going to be a 5.1 percent lower response rate because of this one factor. But maybe there is something more there.

So what -- what does that analysis miss?
MS. UNDERWOOD: The strong -- well, a couple of things. The strong empirical evidence that is the basis for that judgment, ... the government has other things to say but does not contest this decline...
...is a retrospective review of comparing in one case for 2010 the short form census and the ACS, and in 2000 it was to compare the short form and the long form census. It's a comparable comparison. In each case the longer one had a citizenship question on it.
In each case everyone, population groups notwithstanding, there was a decline from the short form to the long form. But there was a much greater decline among Hispanics and non-citizens.

***I promise I'm getting to the point, this was all too important to cut more than this***
JUSTICE GORSUCH: But, counsel, doesn't Justice Alito have a point, to the extent that there could be multiple reasons why individuals don't complete the form. ...But we don't have any evidence disaggregating the reasons why the forms are left uncompleted. What do we do with that?
I mean, normally we would have a regression analysis that would disaggregate the potential cause and identify to a 95th percentile degree of certainty what the reason is that persons are not filling out this form and we could attribute it to this question.
We don't have anything like that here.

So what are we supposed to do about that?

... ... ...
CHIEF JUSTICE ROBERTS: Are there other -- are there other questions on the census for which the administrative records provide more accurate information?

MS. UNDERWOOD: There is nothing in the record about that.

CHIEF JUSTICE ROBERTS: Well, then I don't want to hear about it.
ORAL ARGUMENT OF DALE E. HO ON BEHALF OF RESPONDENTS NEW YORK IMMIGRATION COALITION, ET AL.

... ... ...
***Page66***

Second, the Secretary asserted that adding the question would improve the Bureau's imputation of citizenship for people who lack -- for people for whom the government lacks any such records.
But the government has conceded that that was not true either. At page 34 of their opening brief, they acknowledge that the Census Bureau determined that if the question is added, the imputation process will become less accurate.
***Here we go***

And here's why: The accuracy of imputation depends upon the accuracy of existing data. Federal administrative records are based on a person's legal documents of their citizenship, and, thus, are quite accurate and reliable for this purpose.
But the citizenship question is not. The evidence shows that non-citizens respond to the question inaccurately one-third of the time. So, if the question is used, the data that's used for imputation will be contaminated by those incorrect responses,
making the output of the imputation process less accurate, making the data less accurate, and, again, harming the Secretary's stated purpose of improving the accuracy of citizenship information.

... ... ...
MR. HO: That's right. Adding a citizenship question to the census, I'm sorry, is not helpful for Voting Rights Act purposes because responses to the question are inaccurate so frequently for non-citizens.
Citizenship matters in the Voting Rights Act context when you're dealing with a population in which there's a large number of non-citizens.
The VRA requires the drawing of districts in which minority voters constitute a majority sometimes under some circumstances. Now, under normal circumstances, voting age population data will be sufficient for that purpose if citizenship rates are high.
But, if the minority group has relatively low citizenship rates, for example, as is the case with Hispanic populations in some circumstances, then you need citizenship data to make sure that you're drawing a district in which minority voters are, in fact,
...a majority of the population. And data that's wrong one-third of the -- the time with respect to non-citizens just doesn't help you draw districts --
JUSTICE GORSUCH: ...some of the states who are now Respondents before us have in litigation, including in this Court, argued that ACS data should not be relied upon for purposes of citizenship or other purposes, that the census data is more accurate.
What do we do about that? It seems to me like you kind of put the government in a bit of a Catch 22. You say they shouldn't use the census, except for in later litigation when they have to use the census.
***Basically saying that you can't have it both ways. The question should be asked for a variety of reasons and you have multiple cases arguing that it shouldn't be in either.***
JUSTICE GORSUCH: So -- so the states that said previously that wasn't enough now are going in all future litigation to bind themselves to accept that it is enough? Are you prepared to say that?
MR. HO: -- Justice Gorsuch, we've never taken -- our clients have never taken that position, and I -- I'm not aware of my organization ever taking that position in litigation.
JUSTICE GORSUCH: And how -- how about the under-reporting or the folks who stop and break off answering the long form and -- and we're asked to believe that that's solely attributable to this question?
We have a whole bunch of states that say that, in fact, the break-off rate because of that question, at that question, is something like 0.36 percent. ... So that it's very difficult to understand why that question would be the cause of people stopping answering,
whereas another possible explanation that hasn't been explored, as I understand it at least, is the length of the form itself may deter those with less means and less time to fill them out, just as simple as that, and we don't know.
And what do we do with the fact that we don't know?
MR. HO: ...And the number on the break-off rates for the Internet ACS survey, which I believe Your Honor was referring to, they showed that Hispanics were actually eight times as likely to break off in responding to the ACS upon encountering the citizenship question.
***I am going to have to come back to this again tomorrow. I am sorry for the interruptions. I have a big day tomorrow, on a personal level, and should get some sleep. I hope I have chopped this down some for you. Cutting off at page 78/103. Good Night all***
Missing some Tweet in this thread?
You can try to force a refresh.

Like this thread? Get email updates or save it to PDF!

Subscribe to Headsnipe011
Profile picture

Get real-time email alerts when new unrolls are available from this author!

This content may be removed anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

Try unrolling a thread yourself!

how to unroll video

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" @threadreaderapp unroll

You can practice here first or read more on our help page!

Follow Us on Twitter!

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just three indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3.00/month or $30.00/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!