Tomorrow's controversies, today?

[NoName]

Those of you who were politically aware 10 years ago may remember that there was a controversy over whether the Census Bureau, in preparing results of the 2000 Census, should use "unadjusted" numbers or "adjusted" ones.

The "adjusted" numbers were to be prepared by a process known as Dual System Estimation. Under that process, a few months after the main census, the Bureau would send enumerators to a sample of blocks and conduct a second census. While the results would be less accurate than the original census (because many people would have moved in the meantime), the hope was that combining the two sets of results would give more accurate answers that could then be extrapolated to the rest of the country.

Based on similar work done (but not reflected in the final numbers) in 1999, it was pretty clear that use of adjustment would not meaningfully change state shares of the population (and therefore was unlikely to affect apportionment, or the allocation of federal funds, which is based on the population of states and not substate regions). However, within a state, it was likely to conclude that there were relatively more people in Democratic areas, and thereby affect redistricting, probably giving Democrats an extra seat here and there in a state legislature, and maybe an extra seat or two in the House as well.

It is inherently hard to tell whether the adjustment works, since the results are very close and you have no way of knowing what the real population is. By an amazing coincidence, on this complex and unknowable scientific question, the Democrats came down sharply on the side of using adjustment and the Republicans came down on the side of not using it.

Evaluation of the data from the 2000 census convinced the Bureau that the adjustment probably wasn't working after all, and the unadjusted figures were used both for redistricting and for other purposes since then. There are no plans currently to use adjustment in the 2010 Census, and as far as I can tell nobody in Congress, not even Carolyn Maloney, a vocal proponent of adjustment ten years ago, has complained about this. It seems like it is too late now to put into place an adjustment mechanism now. But who knows...

I want to be the first to say Obama in this thread.

[The Drunken Actuary]

Originally Posted by NoName

Those of you who were politically aware 10 years ago may remember that there was a controversy over whether the Census Bureau, in preparing results of the 2000 Census, should use "unadjusted" numbers or "adjusted" ones.

The "adjusted" numbers were to be prepared by a process known as Dual System Estimation. Under that process, a few months after the main census, the Bureau would send enumerators to a sample of blocks and conduct a second census. While the results would be less accurate than the original census (because many people would have moved in the meantime), the hope was that combining the two sets of results would give more accurate answers that could then be extrapolated to the rest of the country.

Based on similar work done (but not reflected in the final numbers) in 1999, it was pretty clear that use of adjustment would not meaningfully change state shares of the population (and therefore was unlikely to affect apportionment, or the allocation of federal funds, which is based on the population of states and not substate regions). However, within a state, it was likely to conclude that there were relatively more people in Democratic areas, and thereby affect redistricting, probably giving Democrats an extra seat here and there in a state legislature, and maybe an extra seat or two in the House as well.

It is inherently hard to tell whether the adjustment works, since the results are very close and you have no way of knowing what the real population is. By an amazing coincidence, on this complex and unknowable scientific question, the Democrats came down sharply on the side of using adjustment and the Republicans came down on the side of not using it.

Evaluation of the data from the 2000 census convinced the Bureau that the adjustment probably wasn't working after all, and the unadjusted figures were used both for redistricting and for other purposes since then. There are no plans currently to use adjustment in the 2010 Census, and as far as I can tell nobody in Congress, not even Carolyn Maloney, a vocal proponent of adjustment ten years ago, has complained about this. It seems like it is too late now to put into place an adjustment mechanism now. But who knows...

Yeah, I remember that. Your point is???

[NoName]

We're old.

[Werewolf]

I wouldn't be opposed to a superior method of estimating the population, although only if applied in an unbiased manner. I would fully expect people to try to rig such a system, which is why I would oppose it if proposed today. And even if such a system were proposed, there is the little problem that a Constitutional amendment would be necessary, to get rid of that "actual enumeration" requirement in Article 1, Section 2.

[NoName]

My other point is this: Ten years ago, this was a fairly big deal. Democrats were using it as an example of how anti-science the Republicans were. Their fearmongering demagogues were out in full force, with Al Gore telling the NAACP, "The Republicans know theirs is the wrong agenda for African Americans. They don't even want to count you in the Census!" But now ten years later we know the Republicans' skepticism was plausible. It's not even controversial any more. A lot can happen in ten years.

[Griffinsky]

Originally Posted by NoName

My other point is this: Ten years ago, this was a fairly big deal. Democrats were using it as an example of how anti-science the Republicans were. Their fearmongering demagogues were out in full force, with Al Gore telling the NAACP, "The Republicans know theirs is the wrong agenda for African Americans. They don't even want to count you in the Census!" But now ten years later we know the Republicans' skepticism was plausible. It's not even controversial any more. A lot can happen in ten years.

Do you have a link or something? I remember the controversy but not the universal acceptance of the republican-preferred way.

Thanks!

[NoName]

For the absence of any controversy recently, use google; you will find a lot of complaints up to 2002 or so but few, if any, since then. It's hard to prove a negative, though. If you come across anything saying otherwise let me know and I will revise my remarks.

National Academy of Sciences panel's 2008 report on the census supports the Bureau's current goal of using coverage measurement techniques for evaluation purposes but not for an alternative count (pages 1 & 2 in link)
http://books.nap.edu/openbook.php?re...d=12524&page=1; see also the 2004 assessment http://books.nap.edu/openbook.php?re...d=10907&page=1

Census Bureau March 2001 recommendation not to use adjustment for redistricting: http://www.census.gov/dmd/www/EscapRep.html; October 2001 recommendation not to use it for non-redistricting purposes http://www.census.gov/dmd/www/EscapRep2.html; March 2003 decision to continue not using it http://www.census.gov/dmd/www/dipe.html

You're welcome!

[daaaave]

Originally Posted by NoName

However, within a state, it was likely to conclude that there were relatively more people in Democratic areas, and thereby affect redistricting, probably giving Democrats an extra seat here and there in a state legislature, and maybe an extra seat or two in the House as well.

Some of the people in the Berkeley statistics department looked at this more carefully sometime around 2004 or so, and actually found that had the census bureau used the proposed sampling methods that it would have resulted in a net gain for Republicans in the House. Sorry that I don't know details--I skipped the talk.

Edit: poking around, I suspect that the talk was given by David Freedman, and possibly based on this paper: http://www.stat.berkeley.edu/~census/612.pdf

[NoName]

I am not sure what to make of this.
http://www.cqpolitics.com/wmspage.cf...s-000003024858

The director of the Census Bureau will report directly to the White House and not the secretary of Commerce, according to a senior White House official.

The decision came after black and Hispanic leaders raised questions about Commerce Secretary nominee Judd Gregg ’s commitment to funding the census.

Gregg, New Hampshire’s senior senator, voted in committee and on the floor for a 1995 Republican budget that envisioned the elimination of the Commerce Department. Of even more concern to black and Hispanic leaders, Gregg battled President Clinton over a request for “emergency” funding for the 2000 census.

“Secretary of Commerce-designate Judd Gregg ’s record raises serious questions about his willingness to ensure that the 2010 census produces the most accurate possible count of the nation’’s population,”” the National Association of Latino Elected Officials said in a release on Tuesday, the day Gregg was named to the post.

I wonder whether President Obama will ask Congress to remove the Census Bureau from Commerce's jurisdiction.