Balkinization: Will Congress fund the Census?

Comments:

Pam Karlan has addressed how a failure to reapportion in the 1920s led to various modern era debates in voting and districting: http://lsolum.typepad.com/legaltheory/2017/08/karlan-on-congressional-failure-to-reapportion-after-the-1920-census.html

I won't include a link but at first thought this post might be related to another contributor of this blog who argues the current apportionment process violates the second section of the 14th Amendment. He also is working on an article on it.

# posted by

Joe : 10:10 PM

This comment has been removed by the author.

# posted by

Bart DePalma : 11:45 PM

This comment has been removed by the author.

# posted by

Bart DePalma : 11:46 PM

The dispute is not whether to fund the census, but rather the GOP's desire to perform actual census head counts face-to-face, by mail or now online and the Democrats recent proposals to use statistical sampling methods to estimate the population.

The Democrats are justifiably concerned illegal aliens, criminals and folks suspicious of the government in general who are more likely to reside in their blue cities will avoid participating in an actual census.

The Republicans are justifiably concerned a bureaucracy dominated by Democrats is likely to game any statistical sampling to overcount Democrats much like Democrat media polls routinely do before elections.

Constitutionally, however, the GOP is on more solid ground. A census is a head count, not a polling sample.

# posted by

Bart DePalma : 11:48 PM

"There’s no way to know exactly how many Americans would be overlooked. But statisticians fear that if we do move to an electronic method of enumeration, many of them might be poor people who don’t have easy access to computers. And undocumented immigrants, ..."

Technically, of course, undocumented immigrants are incapable of being Americans who are overlooked. But I get your meaning.

An online census wouldn't just be inaccurate because of people who didn't have access to computers, or justly feared being found. It would be terrifyingly subject to being hacked. It has to be a non-starter, and the fact that the head of the Census proposed doing is is a real red flag that there are internal issues at the Census bureau that need to be addressed, the same sort of issues that the previous proposal to replace "actual enumeration" with estimates implied.

We need a head of the Census who actually wants to do the job the Census is constitutionally assigned, and not something different, easier, and much more error prone.

# posted by

Brett : 5:30 AM

Technically, of course, undocumented immigrants are incapable of being Americans who are overlooked.

In NYC, we have a city id, and you can obtain them without having the proper documentation necessary to enter the country. These people repeatedly have valid "documents" that show who they are.

They are also capable of "being Americans." As like police trying to investigate crimes without scaring off some possibly long term undocumented resident not bothering anyone, census counters can do their job there. The same with the other groups cited in the linked piece. This would require a certain mentality and care that might not be present in this administration.

The Supreme Court in 1999 said the law in place required actual enumeration for apportionment purposes and not sampling. Department of Commerce v. United States House. One dissenting opinion noted:

The words “actual Enumeration” require post-1787 apportionments to be based on actual population counts, rather than mere speculation or bare estimate, but they do not purport to limit the authority of Congress to direct the “Manner” in which such counts should be made.

Another cited history in which some sort of "form of statistical estimation to adjust or correct its actual headcount" was used at least for about 80 years. The word "enumeration" means some sort of "counting" and merely doing a head count of 320 million people is not the only way to do that.

The article explains how at some point MAIL was used, a questionnaire sent to be filled out and returned. There are limitations to that including for those with literary issues and so forth. But, it's a useful way to do much of the work. We use computers now to do what used to be done by mail. For instance, it is quite common now to do file you taxes electronically.

If taxes, with all the information there, can be done electronically, not sure why it is horrifying to think the census would be. But, the article didn't say it would merely be done, full stop. It was a new and was going to be test ran. And, like for taxes, computers wouldn't be the only way it was done.

# posted by

Joe : 10:26 AM

I'm kind of with Brett in my skepticism of on line systems. They're too easy to hack, plus they have the access problems others have noted. It would take a lot to persuade me that the system was both accessible and secure, two inconsistent goals.

It seems to me that the dissent had much the better of the argument in Dept. of Commerce v. House (putting aside the partisan nature of the majority). Failure to account for people missed by the census takers means that we've failed to "enumerate" all the people. That's so important to republican government that it should be a mandate.

# posted by

Mark Field : 10:46 AM

The Levinsons' post points out: "So preparations for the 2020 census are in limbo." There are many other matters "in limbo" as political dysfunction continues with the Trump Administration. What takes priority? How long can funding for the 2020 census be put off? The method for making the census will enter into this calculation. It's clear time is of the essence. The Levinsons' message is that Congress must act soon. Will Congress dither on political considerations of where the bar is set for this census in limbo? Political foresight is rarely 20-20.

# posted by

Shag from Brookline : 11:13 AM

The article notes that it was going to be test run etc. but I'm open to any concerns though again taxes are done online. As with absentee ballots, surely when they first were done, people were concerned about mail in this context too.

But, accessibility very well might go in the other direction, especially in populated areas with many places for accessibility. Libraries, e.g., are very accessible, including to the homeless. Also, computers are useful for language and disability issues. See, e.g., Shag's use of magnification.

# posted by

Joe : 11:36 AM

I'm certainly in favor of test runs. I'm just skeptical that they'll prove to be both accessible and secure. And I still think we'll need estimation of the uncounted afterwards -- after Trump, lots of people won't be able to trust being identified by the government, much less providing personal data.

# posted by

Mark Field : 12:08 PM

Joe, you can give them the key to the city, and if they're not here legally, they're not "Americans".

By current court precedent, they have to be counted, though.

Mark, feel free to estimate how much the Census fell short of counting everybody. Just understand that it's a count that the Constitution calls for, not an estimate.

# posted by

Brett : 12:41 PM

An estimate is a count. In fact, sending census takers door-to-door is itself just an estimate with the error on the side of undercounting -- nobody thinks they do reach or ever have reached every single person. The only questions are, Which direction do the error bars run? and What is the range of error?

They're pretty good these days at estimating the census undercount. I'm pretty confident a good estimate would make the totals more accurate than the "count".

# posted by

Mark Field : 12:53 PM

This comment has been removed by the author.

# posted by

Joe : 12:55 PM

"I'm pretty confident a good estimate would make the totals more accurate than the "count"."

So am I. I'm merely not at all confident that, were an estimate permitted, we'd get a good estimate.

# posted by

Brett : 12:58 PM

We are defining "American" different.

A person who came here when they were two without proper documentation, grew up, went to school, was as much a member of the community as everyone else in most respects etc. to many of us are "Americans." But, ymmv.

The word "enumerate" means "to count." There are various ways of counting. If a biologist wanted to "count" how many frogs were in a pond, they need not merely go one by one and count them. Anyway, current court precedent is duly noted. Like you do at times, we are disagreeing with the majority on a certain point.

(Mark Field's comment is telling. The traditional census taker to my understanding asked the head of the household for information, which as the article notes would include their count of slaves. They didn't require a head count of each specific person. I personally never had a census person do that to me. The specialized meaning given to "enumerate" by some is more wordplay than convincing.)

# posted by

Joe : 1:02 PM

I'm merely not at all confident that, were an estimate permitted, we'd get a good estimate.

As Mark notes, the door-to-door method is an estimate. Any system in the real world is going to involve some sort of estimate. So, you are left with the best approach possible to count. "Estimate" itself doesn't mean a lack of a "count."

# posted by

Joe : 1:13 PM

Being known to be inaccurate doesn't make a count into an estimate, Joe.

# posted by

Brett : 1:31 PM

How's this variation:

"Being known to be inaccurate doesn't make an estimate into a count, Brett."

The Preamble refers to a "more perfect Union." Perhaps a "more perfect census" should be the effort to lessen inaccuracies. When has Congress ever been "prefect"? Whatever method Congress endorses now, even for supposed political positioning, the resulting 2020 census may have a different political result. Consider it a bipartisan political gamble.

# posted by

Shag from Brookline : 2:07 PM

I can understand the desire to estimate instead of count, Shag; Not only is it potentially cheaper, but the process for converting the count, (There has to be at least something of a count to base an estimate on.) into the estimate is sufficiently opaque as to allow for almost any amount of shading of the results that you want. One model for extrapolating the count into a complete population will give you one result, another will give you another result, and all so complicated as to make sure that nobody will ever be morally certain that you've got your thumb on the scale.

This is quite obvious, and why we also count, rather than estimate, votes. Even though vote counts are subject to errors, too.

# posted by

Brett : 2:27 PM

Every "count" beyond a small number is an estimate. There are error bars no matter what.

# posted by

Mark Field : 2:41 PM

No, every count beyond a small number is likely inaccurate. An "estimate" involves something beyond knowing that your count won't be accurate. You have to take the count you've gotten, and alter it. Add items you didn't actually find, or subtract items you did find.

Until you do that, you're still just counting.

# posted by

Brett : 2:51 PM

Mark:

An incomplete count is an incomplete count, an estimate is a guess of what a complete count would be.

# posted by

Bart DePalma : 3:26 PM

And, just to be clear: Leaving aside that the Census IS constitutionally mandated to be a count, not estimate... I accept that using modern statistical methods, an estimate could likely be more accurate.

But it would be terrifyingly subject to being gamed, with no easy way to prove it, and with strong motives for shading the results to help one party or the other.

That alone would be reason enough to stick to just counting, even if it weren't constitutionally mandated.

# posted by

Brett : 3:30 PM

Brett:

A statistical estimate is not likely to be more accurate than an incomplete count. A good and objective poll has a margin of error of about +/- 3%.

The hazard of gamed polling is far worse. Democrat media polling during the 2016 campaign grossly overcounted Democrat leaning demographics and overstated Clinton's final margin over Trump by up to 10%.

# posted by

Bart DePalma : 3:43 PM

As Joe has pointed out, the Constitution does not demand a "count". It demands an "enumeration", which means "to ascertain the number of". One can do that by sampling and statistical methods as well as by "counting".

Under current SCOTUS doctrine, of course, "counting" is required. That decision was wrong.

# posted by

Mark Field : 4:01 PM

It's funny, I kept looking up defintions, and they kept coming up like this.

To count off one by one.

Yes, it's the process of ascertaining the number of something by counting it.

# posted by

Brett : 4:09 PM

My definition came from Merriam-Webster. OED gives the same.

# posted by

Mark Field : 4:19 PM

Brett, at 2:27 PM, with his closing:

"Even though vote counts are subject to errors, too."

seems to be channeling Trump's claims concerning the popular vote. Nice try, Hemlock!

By the Bybee [expletives deleted], has anyone else noted references to the phrase "popular vote losing President Trump" that should more accurately read "popular-vote losing President Trump." I've got to dash.

# posted by

Shag from Brookline : 4:57 PM

I did note the word is "enumeration" but not sure how much difference it makes.

Brett's dictionary:

1. To count off or name one by one; list: A spokesperson enumerated the strikers' demands.
2. To determine the number of; count.

So, 'count' is a basic aspect but "determine the number of" factors in as well. The word has somewhat more nuance. The first census act "authorized and required to cause the number of the inhabitants within their respective districts to be taken."

https://www.census.gov/history/pdf/1790_Census_Act.pdf

===

Regarding precedent, another one is Utah v. Evans (2002). The majority noted: "Utah argues that the words “actual Enumeration” require the Census Bureau to seek out each individual." But, " We do not believe the Constitution makes the distinction that Utah seeks to draw. The Constitution’s text does not specify any such limitation. Rather the text uses a general word, “enumeration,” that refers to a counting process without describing the count’s methodological details."

A sort of "imputation" system was upheld. Those interested can read the opinion.

"And the marshals who implemented that Act did not try to contact each individual personally, as they were required only to report the names of all heads of households." Basically, this resulted in some sort of "estimate" though at some point it might go too far.

# posted by

Joe : 5:12 PM

"My definition came from Merriam-Webster. OED gives the same."

Indeed, they do both mention counting, don't they...

# posted by

Brett : 8:35 PM

They both give "count" as a synonym. The actual definition is the one I gave. If "count" is a synonym, I think that cuts against your argument.

# posted by

Mark Field : 8:41 PM

And the definition of "synonym" is "a word or phrase that means exactly or nearly the same as another word or phrase in the same language, for example shut is a synonym of close."

So, yeah, we're talking about counting when we say "enumeration".

But, I think this is one of those arguments where you're too invested in being able to replace the census with an estimate to care about the details.

# posted by

Brett : 6:22 AM

"Technically, of course, undocumented immigrants are incapable of being Americans who are overlooked."

Technically, the Constitution doesn't say to enumerate Americans, but rather the "Number of free Persons, including those bound to Service for a Term of Years, and excluding Indians not taxed, three fifths of all other Persons."

Brett, one line of your argument seems to come down to: an estimate would be open to being gamed. Why wouldn't a count be the same? You'd think if anything would get your paranoid conspiracy antennae twitching it'd be the idea of all those inner Blue city populations being counted by federally paid and directed locals.

# posted by

Mista Whiskas : 9:12 AM

The fact that multiple definitions give enumeration both 'to count off' and 'to ascertain the number of' as main definitions combined with the 'in such Manner as they shall by Law direct' makes me conclude that either avenue is constitutionally open to Congress.

# posted by

Mista Whiskas : 9:14 AM

Mr. W: The fact that multiple definitions give enumeration both 'to count off' and 'to ascertain the number of' as main definitions combined with the 'in such Manner as they shall by Law direct' makes me conclude that either avenue is constitutionally open to Congress.

Count and "ascertain the number of" mean the same thing and neither are analogous with statistical sampling.

The former counts everyone they can find in a city.

The latter polls a tiny percentage of the city and then makes an estimate as to the total population. The optimum mathematical accuracy of an average poll has a random error of +/- 3%. Given that polling participation is wildly uneven between groups (and far worse than census participation), pollsters have to arbitrarily weight different demographics in a guesstimate of the population. This is where the mischief occurs.

# posted by

Bart DePalma : 9:36 AM

I don't see any reason to elevate the synonym as definitive rather than the actual definition. In any case, as I said above, a "count" is merely an estimate anyway. We should be concerned about the accuracy of the population estimates. If that can be done by "counting", fine. If it can't, and some other method is more accurate, then the other method should be preferred.

It is, btw, silly to claim that census takers actually "counted" people. As Joe pointed out, they didn't go out and count the number of slaves. They asked the slaveowner "how many slaves do you have?" and wrote down his answer. The same was true for children and their ages. There's a good chance it was true for spouses.

Neither by word choice nor by practice has "enumeration" been interpreted to mean "line them up in a row and start counting".

# posted by

Mark Field : 9:40 AM

A "synonym" is a word that is similar to another word without being that word. The English language is a complex one where different words that mean similar things are used to get certain shadings.

Utah v. Evans notes that "actual enumeration" is used, which was used to separate something like sampling (more a rougher estimate; not allowed) with the "hot deck imputation" (to help determine a fully "actual" number of inhabitants; allowed even though it along the margins was an estimate of some sort).

But, this wordplay really doesn't change the basic bottom line. Prof. Levinson is correct to be concerned about there being a full enumeration, both for purposes of apportionment and for purposes of all the other things (other than direct taxes, which never was much of a thing) population numbers are used for.

There are many ways to do that, and just going down the line, one by one, and counting over 300 million people is not going to be the way it is done. Some sort of "estimation" will be involved no matter how we "enumerate" or "count" if you like. And, computers will play some role, like eventually mail did over census takers going to the head of the household and asking them information.

# posted by

Joe : 9:53 AM

Our dynamic dyslexic duo Bert and Brat reveal the influence of Sesame Street of their youth on "count" enumeration for census purposes:

https://www.youtube.com/watch?v=2AoxCkySv34\\

# posted by

Shag from Brookline : 10:17 AM

Joe's last comment correctly brings us back to the purpose of the enumeration clause: it's to allocate the number of representatives fairly among the states. A census should "ascertain the number of people" in each state in the most accurate way possible in order to achieve this purpose. If that means "counting", then by all means "count". If it means "counting" as many as possible and sampling to account for the undercount, then that should be preferred. The purpose of the clause -- "the reason of the law", in Blackstone's phrase -- should control the way we interpret it.

# posted by

Mark Field : 10:43 AM

""ascertain the number of" mean the same thing and neither are analogous with statistical sampling."

That's absurd. For example, biologists engaged in population estimates for starlings in a region are not trying to ascertain the number of starlings?

# posted by

Mista Whiskas : 10:48 AM

Mark: A census should "ascertain the number of people" in each state in the most accurate way possible in order to achieve this purpose. If that means "counting", then by all means "count". If it means "counting" as many as possible and sampling to account for the undercount, then that should be preferred. The purpose of the clause -- "the reason of the law", in Blackstone's phrase -- should control the way we interpret it.

Mathematically, sampling can never be as accurate as counting because its participation rate is lower and its extrapolation brings with it a margin of error.

This is the reason why we decide elections by counting ballots rather than using horse race polling.

The only reason Democrats favor sampling is to game the methodology.

# posted by

Bart DePalma : 10:56 AM

Mr. W:

Do you understand difference between counting all the starlings in a tree and counting the starlings on one branch and then multiplying by the number if branches to guess the number of starlings in a tree?

The former you know for a fact, the latter is a guess.

# posted by

Bart DePalma : 11:22 AM

And yet both are plainly attempts to 'ascertain the number' of starlings' in the tree.

In fact, if you can only see one branch clearly, then the latter is going to be a better ascertainment (the counter would know 'for a fact' only what was on the one branch where she could count and would not in anyway 'for a fact' know how many starlings there actually were in the tree).

# posted by

Mista Whiskas : 11:25 AM

This comment has been removed by the author.

# posted by

Bart DePalma : 11:29 AM

Mr. W:

To ascertain means to determine a fact with certainty. See the word root "certain." To ascertain does not mean to guess.

# posted by

Bart DePalma : 11:30 AM

Picture SPAM as "The Count" not on starlings but bats, as on the Sesame Street link I provided. What if it is determined after the census that certain people were not actually counted? Would the census be invalid for uncertainty? Trump's base might claim 3 million in remote rural areas were not counted, all of whom have assault rifles.

# posted by

Shag from Brookline : 11:43 AM

Anyways, nice to see Prof. Levinson's wife involved in the linked blog and she has politely engaged with the comments so far.

In the book "Wrestling with Diversity," their daughters also took part. One just tweeted regarding the disenfranchised inhabitants of the Virgin Islands. They are counted too if not for representative purposes. Finally, recently re-skimmed "Written in Stone," Prof. Levinson's book from some time back on public monuments. Timely yet again.

Have a nice Constitution Day on Sunday.

https://www.nytimes.com/2017/09/12/opinion/irma-virgin-islands-damage.html

# posted by

Joe : 11:46 AM

Shag: What if it is determined after the census that certain people were not actually counted?

Certain people?

Article I, 1, § 2, cl. 3 requires the government conduct an enumeration and 14A, § 2 requires the government to conduct this enumeration "counting the whole number of persons in each State."

In a nutshell, the government has to perform a head count, not an estimate, of all the persons in the United States.

If the government is refusing to count "certain people," the enumeration violates the foregoing provisions of the Constitution. However, any requirement to conduct a headcount has to assume some small group of people will refuse to participate. The former is under the control of the government, the latter is not.

# posted by

Bart DePalma : 11:58 AM

"To ascertain means to determine a fact with certainty."

And in my last comment I gave you an example of a situation where a traditional count would not give you that.

According to your reasoning, if I came across a tree with twenty branches, and the view to 19 of them were obstructed and I saw (and traditionally counted) five birds on the one clearly viewed but could hear birds, but not see them, in the other branches, I'd have to report there five birds in the tree. That's not an accurate determination of the whole number of birds in the tree.

# posted by

Mista Whiskas : 1:17 PM

At 11:43 AM I asked:

" What if it is determined after the census that certain people were not actually counted?"

This question was not suggesting that "certain people" were intentionally not counted or that the government refused to count them. But that seems to be how SPAM read that question. So, if certain people through inadvertence and not intentionally were not actually counted, would the reported census count be invalid for uncertainty per SPAM's constitutional analysis?

SPAM is concerned that Democrats would game the system. But the method determined by Congress for the 2020 census, a Congress controlled by Republicans, could address avoiding gaming the system by either party. With over 300 million Persons in America, what are the chances that ALL Persons would actually be head counted? Why some Persons may be in outhouses on daily constitutionals when the census takers visit; perhaps outhouses should be included as part of the head count just in case. [No comments, please, concerning two- and three-holers.]

# posted by

Shag from Brookline : 2:17 PM

Mr. W: According to your reasoning, if I came across a tree with twenty branches, and the view to 19 of them were obstructed and I saw (and traditionally counted) five birds on the one clearly viewed but could hear birds, but not see them, in the other branches, I'd have to report there five birds in the tree. That's not an accurate determination of the whole number of birds in the tree.

Your hypo is also not in any way analogous to our national census. It is not that difficult to find people living in a modern society and only a tiny fraction of the citizenry is actively hiding from the government.

Shag: So, if certain people through inadvertence and not intentionally were not actually counted, would the reported census count be invalid for uncertainty per SPAM's constitutional analysis?

Inadvertence is a minuscule problem. Census takers make repeated visits to residences which do not answer and actively search for the homeless.

People actively seeking to avoid the census do not violate the Constitution because our basic law does not require anything of the people and does not require the government to compel people to perform civil duties like voting or participate in the census.

SPAM is concerned that Democrats would game the system.

No, I am certain Democrats in the bureaucracy would game any statistical sampling to gain a political advantage like they do registering more voters than exist in Democrat counties in CA, participation data for Obamacare, and playing games with raw temperature measurements to create fictional atmospheric warming. "Big data" is the source of the vast majority of modern propaganda posing as "science."

# posted by

Bart DePalma : 3:29 PM

It's stupid/immoral to assume people would only be missed because the people are hiding. It's interesting that conservatives can spend so much time talking about the ineptitude of the federal government but then think that the only way it could miss people in a *count of the entire US population* (one of the most difficult social science projects one can imagine) is if the people willfully hide from them.

"only a tiny fraction"

A fraction almost certainly greater in number than, say, people whose votes are 'cancelled out' by illegally cast votes. Somehow the latter is an unmatched affront to our democracy but the former 'eh, just a fraction, who cares' for conservatives.

It's almost like there's no actual principle other than momentary perceived partisan advantage at work....

"civil duties like voting or participate in the census."

The Census is not a civic duty like voting, it's a charge to the federal government. The Constitution commands it to make a count, and an accurate one, so that representation can be made according to the actual numbers of the population.

It can choose the manner in which it fulfills this charge, and it need not engage in coercive measures to do so, but it has an affirmative duty to make as accurate of a census as it can.

# posted by

Mista Whiskas : 3:58 PM

Is SPAM (aka "Little Datum") looking for a job in the EPA? It would seem so with SPAM's:

" ... and playing games with raw temperature measurements to create fictional atmospheric warming. 'Big data' is the source of the vast majority of modern propaganda posing as 'science.'"

SPAM made a wise choice in leaving FL "Big Law" for the little mountaintop clime in CO which might be a landing spot for Noah's Ark II.

# posted by

Shag from Brookline : 4:31 PM

Mr. W: It's stupid/immoral to assume people would only be missed because the people are hiding.

What part of "only a tiny fraction" did you understand as meaning none?

The census makes redundant checks on residences. If the residents are not there for multiple checks, they speak to landlords and neighbors.

The estimated total error rates for the census, including intentional and inadvertent non-participation, are minuscule. Inadvertent non-participation is a fraction of miniscule.

https://www.census.gov/newsroom/releases/archives/2010_census/cb12-95.html

# posted by

Bart DePalma : 5:50 PM

When someone makes the statement that "counting will always be more accurate than estimation" it merely exposes an ignorance or bias.

There are many situations in science, especially where large numbers are involved, where it has been shown that statistical estimation produces more accurate results concerning the correct numbers than physically counting individuals.

Obviously Bart didn't pay attention to the case of a flock of starlings, which are constantly in motion. Which, if you think about it, is really an excellent analog for the situation of attempting to count the number of people in a large city.

I would suggest that Bart buy himself about a ton of wheat, and then, with as many people as he can hire, count the individual grains in his acquisition. Twice. Let's see if he gets the same result twice.

# posted by

C2H5OH : 6:21 PM

C2H5OH said...There are many situations in science, especially where large numbers are involved, where it has been shown that statistical estimation produces more accurate results concerning the correct numbers than physically counting individuals.

Such as?

Obviously Bart didn't pay attention to the case of a flock of starlings, which are constantly in motion. Which, if you think about it, is really an excellent analog for the situation of attempting to count the number of people in a large city.

A flock of birds flying in and out of a tree in a tree are not at all analogous to individuals or small social groups people living in fixed residences.

I would suggest that Bart buy himself about a ton of wheat, and then, with as many people as he can hire, count the individual grains in his acquisition. Twice. Let's see if he gets the same result twice.

If necessary, such a task would be easy.

Somehow banks and their employees have managed trillions of dollars in transactions over centuries by counting the transacted currency with only a minuscule error rate. No banker in her right mind would even consider using statistical sampling to determine the amount of currency in its possession. They count it down to the penny.

# posted by

Bart DePalma : 7:09 PM

I apologize to Sandy for not having read his article. Let me offer the excuse that it is my intention only to discuss the mathematical, not the legal, issue here.

Bart, put up or shut up. You have no concept of what it means to count 300,000,000 plus items of any kind. As for the birds/people analogy, a simple mathematical analysis will demonstrate to anyone not suffering from innumeracy that the rate of motion relative to the time required to do the count is what matters. And that, over the course of weeks during which a count is to be done -- in a large city, there is going to be a not-insignificant amount of movement.

Statistically, the error of a poll or statistical sample can be shown to obey certain mathematical laws which are readily proven. The Central Limit Theorem, for example. 3 percent is an arbitrary number chosen by pollsters -- if higher accuracy were desired for some reason, they would simply have to increase the sample size to obtain it.

Philosophically, there is no means, whether estimation-based or enumeration-based that will obtain a perfect count, so the real question should be: how accurate does the census need to be? If it needs to be perfect, then give up now and make whatever choice of method you wish, but do not pretend it is for anything but political reasons.

Once you decide to settle on an accuracy, it can be achieved (and checked) by mathematical statistical methods. No legal issues required there.

Thank you for your patience, and, although I will almost never comment, Sandy -- I truly appreciate your articles, and will watch the future of this one (as it deals, ultimately, with issues related to mathematics, or at least should.)

# posted by

C2H5OH : 7:59 PM

SPAM's 7:09 PM closing paragraph on banks the their employees being down to the penny brings back memories of the Bush/Cheney Great Recession of 2007/8 that involved many banks and their employees. I wonder how that worked "down to the penny" for the many who suffered losses as a result of the Bush/Cheney Great Recession of 2007/8?

Perhaps SPAM's nit-pickling skills would permit SPAM to easily count the individual grains in a ton of wheat, which is fungible (my second favorite word). Of course it would be easy to separate SPAM from the more worthwhile chaff, something Mr. W accomplishes so readily.

# posted by

Shag from Brookline : 8:03 PM

Oh, and Bart? Counting money in large amounts is made possible by passing the cash through machines. Are you proposing to gather up everyone and pass them through a counter? Seems that would be a bit of a legal issue, but others will deal with that.

I will merely point out the infeasibility of something like that in any real circumstance with the number of people we're talking about.

Also -- money, unlike people, comes in fixed, easily separated and identified denominations. Please try not to be so ridiculous in your "examples".

# posted by

C2H5OH : 8:09 PM

Er, Bart, do you get how the Census in your link determined over and under counts?

They used a survey sample...The very thing you decry, you now cite.

More importantly, how to explain that Bart thinks the Census "Democrats in the bureaucracy" can't be trusted to engage in modern statistical modeling, but they can be trusted to traditionally 'count' and also to use modern statistical modeling to conclude that their traditional 'count' wasn't off by much! It's almost like he picks and chooses when they can be trusted...based on whether it's to his momentary, perceived, partisan advantage...

# posted by

Mista Whiskas : 8:11 PM

I cited a couple cases.

These things matter along the edges in part because in practice it can result in an extra representative going to one state over another, especially since the current practice is to use a set 435 denominator in the House of Representatives.

As Mark Field has argued in the past, a republican system very well might warrant an expansion of the number of representatives.

It also matters for funding purposes. Every little bit might help, especially when estimating something like homeless population.

# posted by

Joe : 8:11 PM