Groves rules out use of sampling in 2010 census:
President Barack Obama's pick to lead the Census Bureau on Friday ruled out the use of statistical sampling in the 2010 head count, seeking to allay GOP concerns that he might be swayed to put politics over science. Robert M. Groves, a veteran survey researcher from the University of Michigan, also testified during his confirmation hearing that he remains worried about fixing a persistent undercount of hard-to-reach populations . . . Census officials have already acknowledged that tens of millions of residents in dense urban areas -- about 14 percent of the U.S. population -- are at high risk of being missed because of language problems and an economic crisis that has displaced homeowners.
Adjusting the census would probably help the Democrats, both in the direct sense increasing the number of Democratic-leaning congressional districts, and in the indirect sense of increasing the political influence, and perceived political inference, of inner cities and ethnic minorities.
My comments:
I have a great respect for Bob Groves, and I would trust his decisions on what to do with the Census more than I would trust my own.
Bob's statement that "there is simply no time to prepare for it" seems eminently reasonable to me, especially given the cost constraints under which the census operates. On the statistical merits of the issue, I'm pretty sure that adjusted numbers would be better than unadjusted numbers. The census people know what they're doing, and there are known problems of nonresponse, and, for anything where I care about the damn answer, I'd use their adjusted estimates over the raw numbers.
As a social scientist, I hope the census bureau could release two sets of numbers, one unadjusted for political reasons and one adjusted for those of us who want the most accurate inferences possible.
That said, I'm ignoring a possible indirect effect of adjusting the numbers: If people know that the census will do adjustment, maybe they'll be less likely to participate in the enumeration in the first place. It's hard to measure such an effect and, hey, it might be important. I don't know.
I'm not thinking so much of individuals deciding whether to respond to the census, but rather of the decisions of local jurisdictions, where various spending formulas depend on population. For example, if it's known that the census won't be adjusted, then I'd expect the government of New York City to put a lot of effort into convincing people to participate. If it is known that the census will be adjusted, then there'd be a lot less motivation for localities to do what it takes to boost participation.
Conditional on the data already being collected, you'd definitely want to make statistical adjustments; it's a tougher call to decide on this ahead of time. Also, if you know for sure you won't be adjusting, this will affect the effort you put into collecting the data in different places. So if you're not going to adjust, you might as well make that decision right away.
To expand on this slightly, I think any debates over census adjustments are fundamentally political debates, not statistical disagreements. The scientific consensus on adjustment is pretty easy (although people can argue about the details of implementation, as noted by Lawrence in comments below). It's the political consensus that's difficult, as there are clear winners and losers. With a lack of political consensus, all you need is a little bit of dust and confusion in the air to give a sense of a lack of scientific consensus, which then gets piped back in to justify inaction in the political process.
P.S. Thanks to the commenters below for discussion of the legal aspect of this issue, of which I was unaware.
46 comments
There's not just a political debate over making adjustments. There's also a legal one. (And yeah, I know it's often easy to think the two are indistinguishable, but I think that's a bit too cynical of an attitude to take.)
Namely, the Census is specifically provided for in the Constitution. It's not a matter of statute, and it's not clear to me that statistical adjustments would be admissible, since the Constitution calls for an "enumeration", i.e. a count.
From a scientific standpoint, I agree there is little debate.
Whispers wrote: It's not a matter of statute, and it's not clear to me that statistical adjustments would be admissible, since the Constitution calls for an "enumeration", i.e. a count.
O.W. Holmes wrote: A word is not a crystal, transparent and unchanged; it is the skin of a living thought and may vary greatly in color and content according to the circumstances and time in which it is used.
Notwithstanding Holmes, whose exposition of the Common Law has been rejected by the born-again Originalists on SCOTUS (for whom a word has only one meaning -- that ascribed to it in 1791), right now "enumeration" means a nose count.
Perhaps once we get 2-3 SCOTUS nominees from Obama who want to drag the Constitution back into the 20th Century (although we are in the 21st), perhaps then we can get a proper census count.
Or maybe we could just amend the constitution to reflect precisely what we want it to say rather than resorting to sophistry and "might makes right" judicial politics?
@Andrew: Your brief discussion leaves out mention of the two Supreme Court cases from the Clinton era, which I'm sure Groves also is aware of.
The first case (resolved in 1998, I believe), prior to the 2000 census, by a Court that was probably roughly at the same median voting point as the current one, and in a majority opinion by Sandra Day O'Connor, established that it was Congress's right to establish the method of enumeration, and reinforced the specific language of hte constitution that referred to the census being an "actual count" as oppose to an estimate. Based on that precedent, a GOP minority (or some other group) could bring a case to see an order from the SCOTUS to stop the explicit use of sampling in the 2010 census.
This case, in which the GOP sued and beat the Clinton Department of Commerce, had to do with the use of sampling in the non-response follow-up stage of the census data collection: could the census enumerators rely on sampling to decide which residences to actually visit, and infer the missing people based on such samples.
What Groves is really saying -- it seems to me -- is that there isn't time to fight a legal battle over sampling. He can't be saying there is no time to implement a sampling procedure. After all the methodology itself was developed and ready to go 10 years ago. And it was signed and certified by a special panel of the National Academy of Sciences/National Research Council.
Then there was another SCOTUS case, heard in 2001, I believe, that had to do with the use of imputation to infer missing data (and maybe even missing whole families) during the data processing phase of the census. Imagine Justice Clarence Thomas (who was on the losing end of this argument) railing against "hot deck imputation" (a SAS procedure)!
This was in a lawsuit brought by the State of Utah which thought it had lost a Congressional seat to North Carolina because of the use of procedure. Utah lost the case.
This means, however, that it remains possible to use missing data imputation -- including, presumably, probabilistic models, to fill in the blanks when the census forms omit key variables such as race, ethnicity, education, or age. It even may allow -- this was Utah's claim -- the imputation of whole people, by inputing a family or children from related data about the household, or even imputing it from other data such as the type of housing and the household size of neighboring housing units.
So there is still a role for statistics in estimating the population, in the data analysis if not the data collection stage of the census.
Furthermore, there is nothing forbidding the use of a post-enumeration survey (PES), which is a standard way of checking the size of the under- or over-count, even if the PES information can't be used to "adjust" the census afterwards.
The PES in 2000 helped to show that despite the lack of sampling during the field stage of the census, the undercount was smaller in 2000 than it had been in prior censuses. So perhaps by allocating sufficient funds and effort to that stage of the enumeration the need for sampling was obviated (at least as a cost-saving measure).
"concerns that he might be swayed to put politics over science??"
Seems he has decided to do just that by rejecting sampling.
@Andrew: I do agree with you that the argument over the use of sampling is "political." But when the political actors -- including the Supreme Court -- get into technical issues like this they can do some pretty laughable things.
The second case that I referred to actually involved the Court deciding whether the use of hot deck imputation was "sampling" (which it has previously ruled could not be used in the census, if Congress had so determined). Of course it does involve statistical inference, but the court majority only seemed to be concerned with sampling and therefore let the cow out of the barn, so to speak, by letting the census bureau process the data by using standard statistical procedures.
So when you discuss sampling in your statistics courses, you can also cite the Supreme Court for a definition of what it is or is not.
Great post, Juris. Sounds like this is your area of law.
@coachjdc: I thought the same thing. The AP article by Hope Yen makes it sound as if a strict count is "sound science" and any attempt by the White House to encourage statistical sampling is "political tampering." Yen basically repeats Republican talking points verbatim, and in the process, gets things completely backwards. You'd think the AP would be above such sloppy reporting.
Juris - as I understand the rulings, while the Supreme Court did discuss the meaning of 'enumeration,' that definition was not really at issue and therefore not precedent-setting. What the Supreme Court did clearly decide was that Congress has the authority to determine how the enumeration takes place, and I believe that Congress would be within its authority to authorize statistical adjustments to compensate for undercount.
Would there be a legal challenge? Absolutely, and with the current makeup of the court it would likely be sucessful; but I'm not at all sure that the Court's previous decisions amount to the last word on this issue.
To expand on this slightly, I think any debates over census adjustments are fundamentally political debates, not statistical disagreements. The scientific consensus on adjustment is pretty easy (although people can argue about the details of implementation, as noted by Lawrence in comments below). It's the political consensus that's difficult, as there are clear winners and losers. With a lack of political consensus, all you need is a little bit of dust and confusion in the air to give a sense of a lack of scientific consensus, which then gets piped back in to justify inaction in the political process.For example, see http://www.advocate.com/news_detail_ektid84295.asp
@Richard: You wrote "I believe that Congress would be within its authority to authorize statistical adjustments to compensate for undercount."That is correct, what the Court implicitly affirmed is that if the Congress determined that sampling could be used or statistical adjustments could be used, then that would be permissible. In the Congress vs. Clinton case (1998), Congress had previously amended the census law to expressly rule out the use of sampling, and it was the Census Bureau's plan to use sampling in violation of Congressional will that the Court overruled.
However, in that ruling a lot of the argument on both sides had to do with the meaning of the Article I term "actual enumeration," and whether any method other than an "actual enumeration" of "whole persons" was permitted.
In principle it would have been possible for Congress to amend the Census Law to permit sampling; but that, too, would likely have been impossible to achieve in the current climate, and not in time to prepare for the next census, for which the "critical date" is April 1, 2010. And absent such an amendment, if the Obama Commerce Department decided to use sampling in a form that was not permitted under existing law, the Congressional GOP or any other party with standing could sue for a court order to unhold the 1998 decision. I think that order would easily be forthcoming.
Whatever effect statistical sampling may or may not have on a particular census, it will surely be less than that engendered by purely political shenanigans, such as the recent mid-decade House district reshuffle in Texas that resulted in the GOP picking up five more seats in Congress.
There is currently a push underway in California, as elsewhere, to give reapportionment over to a board consisting of retired judges, an equal number of Democrats and Republicans, in order to eliminate the biases inherent in reapportionment done by a legislature dominated by one or the other party.
I vehemently oppose such measures, unless they are done on a nationwide basis. As long as Texas is under the thumb of the GOP, redistricting there will be to that party’s advantage, and even though I am not a Democrat (or a Republican for that matter) it would be foolish to allow the Republicans an edge in one state that could not be countered somewhere else.
This of course is rife with unfairness, but so is the Electoral College, and we all know how far efforts get trying to reform that outdated contraption.
Maybe I'm just persnickettedy, but couldn't we obviate -most- of the undercount by spending a reasonable amount of money on the Census and not nickle-and-diming it like we always do? I'd love to see all those who get wound up about "actual enumeration" and the sacred intent of the Founders put their money where their mouths are and pony up for the cash needed to do it right.
Honestly, is there any task more important in a democracy than knowing who your citizens are?
Bernard:
You unintentionally hit upon the biggest problem with the census:
It's not a census of citizens, but of residents. That's right, ALL residents, including illegal aliens. And if you were an illegal alien, would you want to be counted by a government agency? See what the problem is?
Under any fair reading of the Constitution, Congress is free to adopt any fair method of counting the residents of the country. The phrase "actual enumeration" in the Constitution was entirely intended to distinguish it from the original arbitrary allocation of House seats. Of course, our "original intent" justices won't buy that "original intent," any more than they accepted the constitutional methods that were provided for settling the 2000 election.
@John. I agree with you. Interestingly, (going from memory here -- last read this a few years ago) in her majority opinion O'Connor used a phrase that the "plain language" of Article I gave Congress the prerogative "and therefore we do not need to reach the Constitutional issues raised."
Now that was a curious statement. But what she apparently meant was that since Article I was clear on this issue, the court did not have to adcress the "equal protection" argument that some in the minority did accept. Namely, since the undercurrent of the "politics" of the census was whether everybody ought to be counted -- with the GOP in particular NOT wanting to count all of the hardest-to-count people, namely, African Americans and Latinos -- those who supported sampling were seeking to get the most complete and accurate count, while those who opposed it would prefer (or accept) an undercount in which it was mainly African Americans and Latinos who were missed from the count.
Thus, there were TWO constitutional principles involved in the case: the Article I Congressional prerogative to determine the method of enumeration, and the 14th Amendment principle of equal protection (and the derivative concept of fair representation).
There was no reason to regard the Article I principle as somehow superior to the 14th Amendment principle: they are both integral parts of the Constitution today. An "originalist" can't arbitrarily choose to ignore the 14th amendment just because it wasn't in the original Constitution.
Along the lines of Juris's point. If the government decides on enumeration over sampling, it is clear that certain populations will be undercounted; therefore disenfranchised, and likely violating the 14th amendment.
So it seems to me the only way the government can uphold the constitution if they want to directly head count, is by giving the census adequate funding.
Let us not forget, that the Republicans are arguing very hard to ensure a system is in place that guarantees the poor and disempowered are ignored by our government services.
In about 1999 I tried to explain the statistical census debate to a Finnish visitor. He didn't understand at all.
He said "Why don't they just look it up on their computer and see how many people there are?" In Finland, every person has an ID and it is listed and all government agencies have access to the data.
I think that's good. Here's a case where liberal civil rights privates advocates are cutting into the other bigger idea
that would benefit democrats, that every person should be counted.
It's true that population registries are common in Europe. The lack of such registries in the U.S. has little to do with liberalism, however. It has to do with a general concern across the political spectrum with the privacy of personal information, and a desire not to have the government in everyone's business.
Libertarians, Liberals, and old-style Conservatives all have questioned the idea of a "national ID card," for example. But there is a group of more authoritarian conservatives (well represented among the G. W. Bush leadership) who supported government intrusion into people's private communications and other affairs.
It's all sampling
Whatever courts may have ruled on the question of whether the Founders meant by using the word "enumeration" to exclude extrapolation from the actually counted sample, it seems to me obvious that the actual census conducted every ten years has always used extrapolation. The census-takers have never, to my knowledge, even pursued the goal of physically counting every person they enumerate. They talk or write to some notional head of household and let him or her extrapolate numbers, without requiring anybody in the household to actuallly present themselves for ocular inspection or laying on of hands.
It would seem that contemproary practice, which is more dependent on mailings than a door-to-door survey, gets even further away from a physical headcount, even if that is what you imagine an "enumeration" to require. You get your sampling, your filter, into the process very intrusively, and at the front end, in the form of dependence on your mailing list, even if that is a list of addresses that is as complete as you can make it.
The only real question is whether we are going to extrapolate from the raw data with a set of corrections and adjustments designed to get us the best estimate of the true population, or whether we are going to leave in place, uncorrected, the distortions built in to the filter, the address lists, that we interpose between our "enumeration" and the true count. The answer is obvious, and any judge who rules otherwise needs to be stricken, via the impeachment process, from the actual enumeration of sitting federal judges, on the grounds that they are letting their attachment to the party favored by distorted population counts pervert their judgments.
After we're finished that, we should amend the Constitution to remove the temptation to future bad rulings, which in this case arises from a provision of that document that is functionally obsolete. We have a much more direct way of counting, actually enumerating, the population count that is germane to Congressional representation. How many people in a given state actually voted in the last Congressional election should determine numbers of representatives and districting. This method wasn't available to the Founders, because they envisioned widely divergent voter qualifications among the several states. But in an era when the entire adult population is everywhere eligible to vote, this method would not only present a cheap and easy enumeration, but would also have the salutary effect of encouraging states to get as many people as possible to vote in every federal election. States that chose to go on making voting unnecessarily difficult in order to see fewer poor, blacks and Hispanics voting, would have to accept the consequence of diminished representation in Congress.
How many people in a given state actually voted in the last Congressional election should determine numbers of representatives and districting. This method wasn't available to the Founders, because they envisioned widely divergent voter qualifications among the several states. But in an era when the entire adult population is everywhere eligible to vote, this method would not only present a cheap and easy enumeration, but would also have the salutary effect of encouraging states to get as many people as possible to vote in every federal election. States that chose to go on making voting unnecessarily difficult in order to see fewer poor, blacks and Hispanics voting, would have to accept the consequence of diminished representation in Congress.
This is the best damn idea I have heard in years.The idea of making politicians who practice voter suppression pay a heavy political price brings tears of joy to my eyes.Absolutely fantastic.
"This is the best damn idea I have heard in years.The idea of making politicians who practice voter suppression pay a heavy political price brings tears of joy to my eyes.Absolutely fantastic."I have to strongly disagree. Glen Tompkins made a rather eloquent explication of the situation, and had me persuaded up until this suggestion. Which is terrible.
#1 People are disenfranchised in voting in the same ways that they are disenfranchised in the census. It is no coincidence that the people who are against census adjustments are largely the same people who were attacking voting registration organizations during the last election. Rather than these two incentives running counter to each other, we might find that they tragically reinforce.
#2 The voting rates vary wildly between elections. We might find people voting for a rep, find out that their new "census" leaves them without the very representative they voted for. Even assuming that we worked a special rule into the law to deal with this, irregular variation in the census figures would be terrible for allocation of government resources. An agency might drop a large fixed cost to set up a program only to find their work destroyed due to a census justified budget cut.
#3 The idea of using voting roles for census, suggests that only voters are entitled to government services. Census determines not only apportionment of representation but distribution of many other services that are unrelated. People that are unable to vote, should nonetheless still be recognized by the government. We still need to know where to build our prisons and hospitals.
#4 The census is very very important for scientific, statistical and epidemiological research. Voting roles are largely useless for these purposes.
#5 People can fight just as much(if not more) about voting registration and actual voting numbers. This is hardly a suggestion that would do anything to dispel the current controversy. Our many recent and ongoing election disputes are clear evidence of this.
So Glen Tompkins: Eloquent persuasive argument, but terrible policy recommendation.
I wrote above
~ a national id could make the census
more accurate as well as serving
other good government efficiencies.
Juris responded:
~ the opponents of a national id are
privacy advocates that span the
political spectrum.
I agree with Juris. But here is a case where liberals, by aligning themselves with conservatives, are fighting what I think are greater causes. Amongst other things, the census would go much better if we always counted every person (as in with a national id), not just every 10 years.
I think the privacy advocates on this issue are a mixture of disingenuous (as in conservatives who actually don't want everyone counted), delusional (as in people who think there is a right to anonymity and that they have it in any useful way now), and paranoid (as in thinking that any information anyone has will most be used in abusive ways).
Statistical RiggingAndrew, another possible "indirect effect" of adjusting the numbers is statistical rigging of the count. What if the "adjustment" makes a redistricting favor Democrats over Republicans, or vice versa?
What do we do with outliers? With 438 congressional districts, there will be some. Do we use Least square error regression, or Least median errror regression (LMS) or LOESS regressions? Should it be linear at all? Perhaps a parabolic regression, or more likely a gamma regression will fit better.
I disagree completely with your statement there is any "scientific consensus on adjustment." As a scientist, I have seen completely different conclusions drawn from gamma regressions, linear regressions, exponential decay regressions and parabolic regressions ALL ON THE SAME DATA SET. Even on simple inflation rates we get different answers on the trend depending on how we measure the error in a simple linear regressive fitted line. Shall we measure least squared error, or least median error (which discards outliers)?
The debate over such methods would be esoteric and opaque to people, and in the end would allow politicians to simply make up the answer they want and skew the "adjustments" in that direction.
Counting is pretty transparent, and could be made more transparent by releasing more details of the counting process. It might be nice of the Census folks to include some other raw data, which we as scientists could use to make adjustments to our own tastes. In our work we deal with larger statistical truths, so for example our conclusions are not invalidated if a probability is off by 1% or 2%.
The simpler and more transparent the governmental process, the less room there is for shenanigans. If you really do give a damn about the answers, this is what you want, along with whatever simple and transparent data we can get that would help us make the adjustments ourselves.
Actually, there is not a need to amend the Constitution to implement a variation on Glen Tompkins idea because that idea is already in the Fourteenth Amendment. Clause 2 of the Fourteenth Amendment mandates that if a state abridges the right to vote in federal and state elections (but not local and county elections) the congressional representation of that state shall be adjusted accordingly.
Of course, Congress has never had the guts to call a state on this issue.
I am a liberal against a national ID program. I fall into Andy's third category. We have seen many times in recent years hospital records, police records, and state department records of famous people being looked at illegally. While most people would not abuse the system many will even for innocent reasons like looking up their friends. Once people have access to private data it will be misused and abused. It has happened in the recent past and it will happen in the future.
It should also be noted that while most European countries have population regisries, they also have rather strict regimes for controlling who has access to personal data. The registries themselves have a limited amount of information about individuals and households, and do not include financial transactions and the like.
@Pragmatus
Whatever effect statistical sampling may or may not have on a particular census, it will surely be less than that engendered by purely political shenanigans, such as the recent mid-decade House district reshuffle in Texas that resulted in the GOP picking up five more seats in Congress.Which continued a decades long tradition of shenanigans established by the Democrats, who managed to maintain a congressional majority (long after most votes went to Republicans) through gerrymandering. When they finally lost power, there was tremendous potential to gerrymander the other way.
There is currently a push underway in California, as elsewhere, to give reapportionment over to a board consisting of retired judges, an equal number of Democrats and Republicans, in order to eliminate the biases inherent in reapportionment done by a legislature dominated by one or the other party.
I vehemently oppose such measures, unless they are done on a nationwide basis. As long as Texas is under the thumb of the GOP, redistricting there will be to that party’s advantage, and even though I am not a Democrat (or a Republican for that matter) it would be foolish to allow the Republicans an edge in one state that could not be countered somewhere else.In my understanding, the overriding principle in California redistricting is incumbency/preservation of the power of the ruling individuals, rather than party advantage as it is in Texas. The Democrats who control the process make districts as uniform as possible so they don't have to compete. The Republicans collude in this since they get safe seats too, and more seats than they would if the Democrats were trying to make every district have a slight majority of Democrats.
In other words, the Republican advantage in Texas is not countered in California at present. The introduction of neutrality through judicial panels curbs both political tendencies: favoring incumbency (ruling individuals) and favoring the majority party (ruling group). To the extent the first tendency is expressed at the expense of the second, as in California, neutral district drawing would not upset the national balance of power.
@Andy
I think the privacy advocates on this issue are a mixture of disingenuous (as in conservatives who actually don't want everyone counted), delusional (as in people who think there is a right to anonymity and that they have it in any useful way now), and paranoid (as in thinking that any information anyone has will most be used in abusive ways). How about disingenuous liberals who don't want a tool against illegal immigration as a category?
Brian...
I would bet that if the GOP controlled the process in California, within six months they would jigger things in such as way that the next election would return (surprise, surprise) an extra four or five seats for them.
By making your statements about abuses by the Democratic Party in Texas, I guess in order to assert that two wrongs make a right, it is clear that you evidently missed my last point, which was to state that the process currently in place is not fair, but also that it's overturn is impossible.
I'm another liberal who would apparently fall into Andy Ruina's "paranoid" category. It's not that I think any information anyone has will mostly be used abusively—that's a caricature of the position I and others actually take, which is simply that some abuse must reasonably be expected, that it will likely occur fairly soon after the new power or information becomes available, and that when it does happen it will do enough harm to outweigh the benefits of a national ID system.
These objections might be defeated if the system were set up with strong built-in defenses against expected abuse. I am open to that possibility. But in recent years, we have seen too many new powers granted to government under the fairy-tale assumption that no one will ever seek to use them for illegitimate purposes. Law enforcement and antiterrorism organizations, in particular, are still seen by too many people as incorruptible guardian angels. In this climate, I just don't think there's enough political will for the oversight a national ID system would require.
Getting back to the main topic, I do think the census is too important to be so underfunded, and I'm not convinced that we can't find a way to have a reasonably accurate census based on counting of real people rather than statistical techniques.
Some thoughts on this debate:
The data that the Census Bureau collects is held for 72 years before being released to the public. The latest public census forms are the 1930 census. That concept is difficult for most to understand, though, as they keep hearing of data breaches and hacking into computer systems.
The original 1790 census through (at least) the 1930 census had census takers actually go to each residence in their area, and sit with the head of household to get the information for the census.
In 1790, the census takers got the following information:
- County
- City
- Page
- Heads of Household (always male unless the husband had died)
- Free white males of 16 years & upwards, including heads of families (this was a numerical count, not a count by names of individuals)
- Free white males under 16 years (again, a numerical count)
- Free white females including heads of families (numerical count)
- All other free persons (numerical count)
- Slaves (numerical count)
By the 1820 census, the actual name of the head of household was still being collected, and only numerical counts were made of all others, but the categories were more refined, such as 'Free white males' divided by those:
- under ten years
- of ten and under sixteen
- between sixteen and eighteen
- of sixteen and under twenty-six
- of twenty six and under forty-five
- of forty five and upwards
Foreigners not naturalized
Numbers of persons engaged in:
- Agriculture
- Commerce
- Manufactures
Slaves were counted, males and females separately, by
- those under 14
- those of fourteen and under twenty-six
- of twenty-six and under forty-five
- of forty-five and upwards
Additional columns were on the Census lists to list a numerical count of 'Free Colored Persons', with the breakdown mirroring the age ranges of Slaves.
It was not until the 1850 Census that a listing of all the names of members of the household were included in the census records. Also included was the actual age of the person, place of birth of each person, occupation, whether 'deaf and dumb, blind, insane, idiotic, pauper or convict' (one category). A separate page was available for those states and locales where slaves were present.
The above are illustrations that the purpose of the Census has changed over the years, from just the Constitutional 'enumeration' for representational purposes, to today's additional uses for distributing various other benefices, such as urban planning, distribution of federal funds, etc.
The use of 'census workers' who went from residence to residence was found to be very expensive, as by the 1930s, each family visited would take up to 1/2 hour or more, plus travel time from one residence to the next, plus those residences where no one was home or available at the initial visit, requiring a re-visit, etc.
Eventually, the use of mail-in forms was used, based on the address information of the USPS, plus pre-census verification. That system works quite well when the mail is delivered directly to a person's residence. But it doesn't work for the homeless (undercount); and those who have a residence, but accept delivery of mail through a Post Office box only (undercount), or at both the residence AND a PO box (possible overcount).
With the postal address system, if the USPS can't deliver a census form and returns it, a census worker travels to the address to confirm if there is an actual residential abode at the site. If the form is not returned by the USPS (as undeliverable) and is not returned by the residents at the address, current procedure is to have a census worker go to the address and collect the information.
There is an additional problem with the 'USPS' method - namely the conversion of residences from/to single family to multifamily, or changes in the number of dwelling units of the building. For instance, the address at which I reside was a dingle family residence until the early 1950s, at which time it was converted into four dwelling units. During the late 1980s, it was reconverted back into a single family dwelling unit. Twenty years after the reconversion, I still sometimes get mail delivered to this address with a sub-unit indicated (Apt. 1, Unit 2, etc.).
As to Utah, it is my opinion that the Mormon Church was the prime reason the state got 'screwed' out of the additional Congressional district. The church tells those members 'on mission' to use their current residence as their 'permanent residence'. When they are out of state, the church doesn't consider them a resident of the state, but of where ever they are at that time living, be it a different state, or in a foreign country. If the church policy were to consider them a resident of Utah, but only in temporary residence outside of Utah, not permanent, then the Utah address could be used, and the state of Utah would have a somewhat larger number of residents. In 2000, it probably would have been enough to push Utah ahead of North Carolina for the additional Congressional seat.
Mike in Maryland
My Blogger ID is http://www.blogger.com/profile/02848893412251095965
A correction to my post (above):
The word 'dingle' should read 'single'. (It was a typo, not some other, nefarious, reason for the incorrect word being used.)
Mike in Maryland
My Blogger ID is http://www.blogger.com/profile/02848893412251095965
Actually, have to correct a mis-statement about Texas history posted above.
In 1980, the Democrats were still a sizable majority in Texas. The lines drawn in 1981 mostly reflected that voting breakdown.
In 1991 and 2001, the Texas Legislature were unable to draw lines so they ended up being drawn by the courts. The courts (while having to deal with new districts) tended to believe that, since they weren't a political branch, they should make as few changes to the district lines as possible.
@Pragmatus
"I would bet that if the GOP controlled the process in California, within six months they would jigger things in such as way that the next election would return (surprise, surprise) an extra four or five seats for them."
It's not a left/right issue. It's a political culture issue, such that the Texas Democrats have recently chosen to use their power fight the Republicans, the Texas Republicans have used their power to fight the Democrats, and the California Democrats have used their power to fight the challengers and keep their cushy jobs. Both are forms of corruption. I don't know how why you are so sure the California GOP would be corrupt like the Texas parties rather than corrupt like the California Dems; I agree that as politicians they would almost certainly abuse their power in one way or the other (to promote themselves as individuals or their party).
"By making your statements about abuses by the Democratic Party in Texas, I guess in order to assert that two wrongs make a right, it is clear that you evidently missed my last point, which was to state that the process currently in place is not fair, but also that it's overturn is impossible."
Perhaps I wasn't clear; I never tried to say that two wrongs made a right. My point was that sometimes one wrong makes another wrong easier to commit, so we should not commit wrongs. The implication will be that if we support the Democrats, and another state has a politicized/corrupt system advantaging Republicans, we should think very, very, *very* carefully about doing the same in our state to "counter" that advantage. In other words, *your* argument flirts dangerously with the assertion that two wrongs make a right, and no fair reader will find in my above comments that I implied that at all.
You missed *my* point: that you can't intelligently reflexively oppose replacing political gerrymandering with a neutral process by appealing to the tendency of politicians to advantage their party. This is because sometimes that tendency is overwhelmed by a countervailing one: politicians act to keep their seats ideologically pure and thus electorally safe. This does not advantage either party, since each seat is as uniform as possible, resulting in a fair apportionment of seats to parties by population. Neutral, rational division of districts would preserve this apportionment.
The unfortunate effect of one party districts in California is to marginalize moderates everywhere and reduce the importance of elections. It is therefore a good idea to replace the politicized district drawing there with a more neutral system, and this will not disadvantage either party (only extreme elements of both). I personally would even go further and support it when it would disadvantage a party, but that's not what we are discussing and may be a product of my lack of identification with either party.
You only said that replacing the Electoral College is politically impossible. I agree. As I outlined above, increasing fairness in reapportionment is possible without this through neutral reapportionment, at least in those states with a balance of GOP and Democratic districts proportional to their population (those where every seat is a safe seat, and probably also swing states where neither party could gerrymander itself extra seats).
I am opposed to a national identification card, and I think that pointing to the EU to counter concerns about abuse is laughable, but the US already has a de facto compulsory national ID number - the SSN - which is only more vulnerable to abuse because it isn't properly set up to serve that function.
As for the census, I think I'd prefer that the official census data reflect the data collected in the census. With the Democratic party standing to benefit from a more complete census, and currently in control of both the process and the budget, I see no reason that a count can't be made reasonably accurate.
There will of course be imperfections, but the sort of counting difficulties raised in this thread are largely apparent to the naked eye, so to speak. It seems to me that burying another set of assumptions and technical manipulation in the official results only makes it that much more difficult to parse or adjust them after the fact.
Speaking of counting difficulties, there are necessarily people who should be counted but are unable to respond in person; it is one thing to allow a representative who avers personal knowledge to respond for them, quite another to guess at their existence and/or likely response. And the use of mailing lists, as far as I can see, is one technique which in no way filters out any data or excludes the use of other techniques (to the contrary, it should make them more efficient). These are both rather unpersuasive equivocations, in my opinion.
One question about the legality: If the Constitutional language doesn't require an actual enumeration, what phrase other than "actual enumeration shall be made" would better convey such a requirement?
One of the major questions now confronting some jurisdictions is how to count prison inmates. Do you count them as 'residents' of the jurisdiction in which they are imprisoned? Do you count them as residents of their last legal, non-prison, residence?
This is an especially tricky question for those prisoners who formerly lived in one state (say Oregon), committed a crime in another state and was convicted and sentenced in that state (say California), but are serving prison time in still another (say Arizona). Two of those three states might feel they are getting 'cheated' out of the 'census stat' if the prisoner isn't counted in their particular jurisdiction.
I'm not sure how this might be sorted out, especially when a prisoner has a reasonable expectation of living well beyond their expected release date. Let's say a generally healthy 35 year old, sentenced to a term of 10-15 years - even without early release on 'good behavior', they would be released with an actuarial expectation of living at least 20-30 years beyond their release date.
Mike in Maryland
My Blogger ID is http://www.blogger.com/profile/02848893412251095965
Brian...
I'm afraid the situation in California is exactly the same as in Texas. In Texas districts are rigged to benefit the GOP, in CA they are rigged to benefit the Democrats.
It is not possible IMO to ever purge this practice in every state nationwide, therefore in my view we are stuck with it, and any state that opts out of partisan redistricting is doing its dominant political party a disfavor. I further believe that the folks who run both parties are well aware of this, thus providing additional motivation not to change the situation.
In my earliest remark the example of the Electoral College was provided as a comparative.
Re: Redistricting in California: this was Prop 11, it was approved by voters in the Nov 4, 2008 election, and is now in effect.
Prop 11 details: http://www.smartvoter.org/2008/11/04/ca/state/prop/11/
Zosima,
I would not propose abandoning the other ways the federal govt has to prevent states from obstructing voting rights. I simply point out that making every eligible voter who isn't registered and doesn't vote cut away from a state's Congressional representation, would create a self-interested motive in the direction of universal voting, a carrot to reinforce the sticks already in place, sticks that, compared to other countries, are doing an objectively lousy job.
If we had such a system in place, and tmess2 reminds us that it wouldn't take another amendment, since we already have the license to do this in the 14th, we would quickly see some states doing what other countries do to get near-100% voter participation. These staes would pass laws revoking your driver's license, for example, if you failed to vote. They would also presumably make it easy to vote, a la Oregon. As these states got near 100%, their increased representation would start an irresistable competition among the states, and soon all would have adopted whatever scheme gets us closest to 100% voting, and seeking every possible means to the maximum count possible.
Yes, data gathered by the Census Bureau is used for many things other than the purpose mandated by the Constitution, the determination of the number Congressmen allotted to each state, and the size of the tracts from which their districts must be constructed. Yes, most of these other purposes would not be well-served by this polling count. But Congress has already seen fit, because these other purposes are of such public value, to fund data collection both as part of but in addition to the count used for electoral calculations, and even completely separate efforts. Of course Congress would continue to fund such data collection, only now there would not be this political football attached to the process, and we would see no objections to using valid statisitical methods to arrive at the best numbers possible for these other purposes.
On the statistical merits of the issue, I'm pretty sure that adjusted numbers would be better than unadjusted numbers. The census people know what they're doing,In 2000 the Census career staff decided not to adjust the head count numbers, a decision endorsed by the National Academy of Sciences.
That's because the capture-recapture method that would have been used for the adjustment is a lot more complicated and controversial than conventional statistical sampling. The idea is to conduct a second, independent, "census" using a large sample. Then, the count is adjusted based on how many people got counted twice, how many people were only counted the first time, and so on. There's a lot of things that can go wrong in this process, and the statisticians (not the politicians) decided that it wouldn't improve the accuracy of the count.
@Pragmatus
"I'm afraid the situation in California is exactly the same as in Texas. In Texas districts are rigged to benefit the GOP, in CA they are rigged to benefit the Democrats."
A lot of articles online imply otherwise. See http://www.caltax.org/member/digest/oct2001/10.2001.Quinn-BipartisanRedistricting.08.htm and http://rangevoting.org/GerryExamples.html.
Post a Comment