Census Estimates for 2008 -> 2010 Apportionment

[cinyc]

The Census Bureau released its new estimates for the population of the states as of July 1, 2008. As in past years I have used that data to project the April 1, 2010 apportionment populations. This requires finding the population growth in the resident population for each state, then applying that to the apportionment population.

One special circumstance is to account for the effect of hurricane Katrina. LA saw an estimated drop of 250 K in the 10 months following the hurricane. If I used the normal methodology, that would project a continued decline through 2010. Instead, for LA I took the percentage growth from July 1, 2006 to July 1, 2008 then applied that to the estimate for July 1, 2008. To this I added the difference between the resident and apportionment populations in 2000 to reach a projected apportionment population for 2010.

Based on this projection, the following adjustments would be required to reapportion the seats in 2010:

AZ +2
FL +2
GA +1
IL -1
IA -1
LA -1
MA -1
MI -1
MN -1
MO -1
NV +1
NJ -1
NY -1
OH -2
PA -1
SC +1
TX +4
UT +1

Compared to last year's projection this is a shift of two seats from MN and OR to NY and SC. The last states awarded seats were FL 27 (431), NY 28 (432), CA 53 (433), SC 7 (434) and TX 36 (435). These seats are on the bubble and most at risk to fluctuations in growth in the next two years. TX 36 is particularly at risk since part of the population growth is due to Katrina relocation and dropped from seat 433 in last year's projection.

The next five seats would go to OR 6 (436), WA 10 (437), MN 8 (438), MO 9 (439), and NC 14 (440). Seat 436 is important if the new Congress passes a DC representation act like the one offered in Congress last year, since that seat could be real in 2010. OR and MN have been bouncing back and forth from this waiting list to the real list on the last couple of estimates, so they really could go either way in 2010.

It looks like the congress may pass it, so does that mean number increase to 436 or 437? And does it stay that way or go back to 435?

The bill to add a DC seat would increase the number of seats from the states to 436.

Where does RI 2 rank on the list?

[muon2]

The Census Bureau released its new estimates for the population of the states as of July 1, 2008. As in past years I have used that data to project the April 1, 2010 apportionment populations. This requires finding the population growth in the resident population for each state, then applying that to the apportionment population.

One special circumstance is to account for the effect of hurricane Katrina. LA saw an estimated drop of 250 K in the 10 months following the hurricane. If I used the normal methodology, that would project a continued decline through 2010. Instead, for LA I took the percentage growth from July 1, 2006 to July 1, 2008 then applied that to the estimate for July 1, 2008. To this I added the difference between the resident and apportionment populations in 2000 to reach a projected apportionment population for 2010.

Based on this projection, the following adjustments would be required to reapportion the seats in 2010:

AZ +2
FL +2
GA +1
IL -1
IA -1
LA -1
MA -1
MI -1
MN -1
MO -1
NV +1
NJ -1
NY -1
OH -2
PA -1
SC +1
TX +4
UT +1

Compared to last year's projection this is a shift of two seats from MN and OR to NY and SC. The last states awarded seats were FL 27 (431), NY 28 (432), CA 53 (433), SC 7 (434) and TX 36 (435). These seats are on the bubble and most at risk to fluctuations in growth in the next two years. TX 36 is particularly at risk since part of the population growth is due to Katrina relocation and dropped from seat 433 in last year's projection.

The next five seats would go to OR 6 (436), WA 10 (437), MN 8 (438), MO 9 (439), and NC 14 (440). Seat 436 is important if the new Congress passes a DC representation act like the one offered in Congress last year, since that seat could be real in 2010. OR and MN have been bouncing back and forth from this waiting list to the real list on the last couple of estimates, so they really could go either way in 2010.

It looks like the congress may pass it, so does that mean number increase to 436 or 437? And does it stay that way or go back to 435?

The bill to add a DC seat would increase the number of seats from the states to 436.

Where does RI 2 rank on the list?

My current projections place RI-2 at #419.

[Brittain33]

Florida population growth may have declined to zero.

http://www.heraldtribune.com/article/20090228/ARTICLE/902280347

What does this mean for reapportionment if it stays low or in the very low 6 figures through 2010? How about if Arizona has the same experience?

[Devilman88]

Very nice work. It will be fun to see how much each state's population has change over the past ten years.

[Verily]

Florida population growth may have declined to zero.

http://www.heraldtribune.com/article/20090228/ARTICLE/902280347

What does this mean for reapportionment if it stays low or in the very low 6 figures through 2010? How about if Arizona has the same experience?

FL-27 is very much on the bubble; if Florida's growth has truly stopped, the state will get only one new seat, not two. That means somewhere, probably Oregon or Minnesota, would get the extra seat.

[I spent the winter writing songs about getting better]

Oh, how much would I love it for us to keep all our seats by denying Florida one!

[jimrtex]

I took a look at the effect of some different apportionment methods.

If we take the projected 2010 50-State population, and divide by 435 we get a national average district population of 709,491.  So let's say that we divide each State's population by 709,491 to determine how many representative the State receives.  Each State gets a whole representative for each whole 1/435 of the population.  We can't apportion fractional representatives, so this is quite reasonable.

But then we end up with only 412 representatives.  After it is pointed out that Alaska, North Dakota, Vermont, and Wyoming would be left unrepresented, they are apportioned the one representative that they are entitled to under the Constitution.  But this still leaves us with 416 representatives, 19 short of our target.

Someone suggests if we reduced our quota somewhat, we could apportion 435 representatives among the States - one representative per every N persons in a State.  In effect, if we reduce our price a bit from 709,491l the increased demand for seats will meet the supply.  And with a bit of experimentation, we find that with a quota of 673,770, that New Mexico will get its 3rd seat, or 435 overall.  Alaska is now able to pay full fare, with only 3 States falling a bit short.

This system is D'Hondt, with the adjustment to guarantee every State representation.  The States which would gain (compared to the current system) are: California (+2 to 55); Florida (+1 to 28); Illinois (+1 to 19, replacing the seat it would lose from 2000 to 2010); New York (+1 to 29, replacing 1 of the two it was forecast to lose); North Carolina (+1 to 14); Ohio (+1 to 17, replacing 1 of the two it would have lost); and Texas (+1 to 35).

The States which would lose one seat are Alabama to 6; Hawaii, Maine, New Hampshire, and Rhode Island to 1; Nebraska and West Virgina to 2; and South Carolina to 6 (losing its forecast additional seat).

The legislation passes out of the House on  support from the States that would gain from it, plus others like Pennsylvania and Georgia who might benefit in the future.  Based on States that would at least marginally benefit, one might expect a 314 to 121 vote.
The proposed apportionment goes nowhere in the Senate based on the opposition of the senators from 32 States that would always lose under the proposed apportionment scheme.

Their argument is that under D'Hondt, some persons in every State (but one) will go unrepresented because of the fraction that is discarded.  They propose that the State population be divided by the average district size (709,491) as in the original plan.  But that any State be apportioned a representative for each whole number or fraction in the quotient.  All States would be assured representation since they all have at least a fraction of the quota.

They note that in practice, that one representative would not be assigned the possibly small fraction of the population that did not qualify for full representation.  But rather the extra representative would share some of their duties, so that each representative in a State would represent an equal number of persons.

The calculation is done, and it is found that 462 representatives would be apportioned (50 more than under the initial plan).  This is not unexpected, since every State had a fraction of its representation discarded.  So like under the initial plan, the quota is adjusted upward.  States with a relatively small fraction might be expected to lose representation.  But the remaining whole number of representatives could take up the slack.

So again with a bit of work, we find that a quota of 747,991 would mean that California would
give California 50.0+ representatives, with a 51st representative apportioned for the minuscule fraction.

Winners under the senate plan would be Delaware (2), Idaho (3), Iowa (5, retained), Minnesota (8, retained), Montana (2), Oregon (6), and South Dakota (1) all States that would had rather largish CDs under the original plan.

Losers would be California (-2 to 51, Florida (-1 to 26, net gain of 1 from 2000), New York (-1, net loss of 3), Pennsylvania (-1, net loss of 2), and Texas (-2, net gain of 2).
The proposal will go no further in the House, than the original plan did in the Senate.  So someone suggests a compromise.  Instead of requiring a whole quota for even the last representative in a state, as in the house plan; or a small fraction for the last representative under the senate plan; why not apportion an extra representative to each State where the fraction was greater than (or equal to 1/2).

This sounds fair.  If a state is entitled to at least 1/2 of an additional representative it will gain a full representative.  The calculation is done, and it is found that using the national average of 709,491 that 432 representatives are apportioned (a surprising 30 states have a fraction less than 0.5).  As before we tweak our quota a bit, and find that a quota of 704,977 will give Washington 9.5 representatives, which rounds to 10, and a total of 435 representative.  The 3 final seats go to South Carolina, Texas, and Washington.

Again the winners and losers are determined.  There is little difference from current law.  Washington gains its 10th seat, and Rhode Island loses its 2nd.  This scheme is the same as St. Lague (since all States have a population of at least 1/2 the national average district population, there is no need for further adjustments).

This plan might gain overwhelming approval in the House since it is favorable to States with 6 representatives, and might even win approval in the Senate, though the slim margin would not be able to win a cloture vote.
Rhode Island finds that it would be entitled to 1.482 representatives if the national average district population was used, and 1.491 representatives if the adjusted quota was used.  But after thinking about it a while, they decide that the important thing is not that Rhode Island is slightly closer to 1/435 of the national population than it is 2/435; but rather that a single representative would have to represent 1,051,312 persons, or 341,821 over the national average, while each of two representatives would represent 525,656 persons or only 183,835 less than the national average.  It is not the number of representatives but rather the number of persons represented by each that should be equalized as much as possible.

They would call this difference a representation [/i]error[/i], and argue that a State should be awarded an nth representative when that would reduce the representation error.

That is when  P/(n-1) - Q > Q - P/n

where Q is the average national district population, and P is the State population.  P/n is the State average district population for n districts.

Rearranging:  P (1/(n-1) + 1/n) > 2Q

And:  P (2n-1)/(n*(n-1)) > 2Q

And    P/Q > 2*(n*(n-1))/(2n-1)

The divisors under this scheme are 4/3 12/5 24/7 40/9 ...

If a State had 4/3 of national average population, and had a single representative he would would represent 1/3 more than the average representative, while if it had 2 representatives, he would represent 1/3 less than the average representative.  This would minimize the variation in district sizes throughout the country.

After adjusting the quota to 710,091, Oregon is awarded its 6th and the USA's 435 seat.  Its population is 60/11 (5.45 of the quota).  Each of 5 representatives would represent 774,644 people, or 64,553 more than the quota.   Each of 6 representatives would represent 645,537 64,554 less than the quota.

The winners under this plan are Montana (its 2nd representative restored), and Oregon (6th).  South Carolina would lose its 7th and Texas one of its 4 new representatives leaving it with 435.

Following the 1990 Census, Montana argued that Congress should use this apportionment method.  That is, just as States are compelled to have districts as equal as practicable within a State, that Congress is compelled to have districts as equal as practicable among the States.  The Supreme Court ruled that Congress was within its discretion to determine the details of the apportionment method.

This plan finds favor among only the smallest of States and about 1/2 the States.  It is nominally, the system used in England for apportioning constituencies among counties.
Finally someone proposes a further compromise.

Rather than divisors of

  n  or  n-1

or

  n - 1/2  or (2n-1)/((n-1)*n)

Why not use the geometric mean of either pair:

  sqrt(n * (n-1))

This minimizes the relative difference among either the number of representatives per person or the number of persons per representative.

Montana would not get a 2nd representative, because that would mean that it would have 44.5% too many representatives; while a single representative only represents 38.4% too many constituents.

With a final quota of 707,617, South Carolina is award its 7th and the countries 435 seat.  South Carolina's population is 6.481 or sqrt(7*6) times the quota.  With 6 representatives, each would represent 764,314 persons or 1.080 times the quota.  With 7 representatives, each would represent 655,126.  The quota is 1.080 times as large.

There are no winners or losers, since this is the system used under current law.

[○∙◄☻¥tπ[╪AV┼cVê└]

Where does one get the apportionment (not the resident) census estimates, anyways?

[Padfoot]

Where does one get the apportionment (not the resident) census estimates, anyways?

What's the difference?

[○∙◄☻¥tπ[╪AV┼cVê└]

Where does one get the apportionment (not the resident) census estimates, anyways?

What's the difference?

The apportionment has additional military and other federal overseas population.

[bgwah]

I really want WA-10...

[CultureKing]

I really want WA-10...

Agreed. Any new system should have in it that Washington is guarenteed at least 10 seats.

[jimrtex]

Where does one get the apportionment (not the resident) census estimates, anyways?

The Census Bureau does not issue apportionment estimates.  More specifically, it does not make estimates of that portion of the overseas population that it has decided to include in the apportionment population - nor any allocation among the States.

What I do, is ignore it.  It generally only comes into play in a few instances.  There is not a gross disproportionately among home states and military personnel deployed overseas, and the population is small (about 0.2%).  The lowest overseas share is about 0.1% in MA, CT, NJ, MN, RI.  And the highest are about 0.3% to 0.4% in HI, MT, SC, AK, AL.

Western states with small population, and southern states tend to have higher shares.  There may be an effect from larger military bases, or states without income taxes.  There are also probably age related effects.  People in the military tend to be disproportionately young relative to the overall population.

Massachusetts was the State that sued to stop the practice of counting the overseas population (I believe Washington was the beneficiary).  In 2010, it could play a role in Minnesota and South Carolina.

For example, Minnesota could go from 7.500 to 7.494 (I arbitrarily picked 7.500, but used a 0.075% deduction based on Minnesota's relative share of overseas population of 0.126% vs the USA average of 0.205%.  This gives an order of magnitude of the effect.  Similarly, South Carolina could go from 6.500 to 6.508.   The differences are less than 1/100 of a representative.

What Muon does for his projections is to project the 2010 resident populations, and then adjust based on the same proportions of overseas population as in 2000.  So Minnesota would get a 0.126% increase, while South Carolina would get a 0.324% boost.

The overseas population is not used for redistricting, but rather only for apportionment of representatives.

[jimrtex]

I really want WA-10...

If the DC bill would also change the apportionment formula so that Washington would be more likely to get a 10th seat in 2010, then New York rather than Utah would get the 436th seat.

[Holmes]

I don't think it merits its own thread, so I'll post it here. Pew has new interactive maps that show immigration over the years by region and state. Some are obvious but the state info is pretty interesting.

[Brittain33]

Anything else interesting yet?

Several Florida counties went from fast-growing to stagnant, as expected.

[cinyc]

Anything else interesting yet?

Several Florida counties went from fast-growing to stagnant, as expected.

St. Bernard Parish in Lousiana was the fastest-growing.  Orleans Parish was the third-fastest. The repopulation of New Orleans continues - though St. Bernard is only about half as big as it was in 2000 and Orleans about 2/3rds as big.

The Travis County, Texas (Austin) estimated population was 998,543 as of July 1, 2008.  It should be over 1,000,000 by now.  Fulton County, Georgia joined the over 1,000,000-person county club as of July 1, 2008.

Of the top 25 metros, the Atlanta Metro passed the Washington Metro to move into the 8th spot on the largest metro area list.  The Phoenix Metro moved ahead of the San Francisco-Oakland metro into 12th.  Sacramento moved into the 25th slot, bumping Cleveland off the top 25 list.  The Detroit and Pittsburgh Metro areas lost population (no surprise there).

Michigan (-0.5%) and Rhode Island (-0.2%) lost population from 2007.   Maine, Ohio and Vermont were fairly stagnant, gaining 0.1%.  Pennsylvania was next worst at 0.2%.  The national growth rate was 0.9%.

[Brittain33]

Anything else interesting yet?

Several Florida counties went from fast-growing to stagnant, as expected.

Ugh. Sorry for posting this to the wrong thread.