EDS Releases New Reapportionment Estimates

[jimrtex]

I'm really curious how a "private sector demographic firm" has access to "tentative Census data" absent a leak. Either Politico is misrepresenting a private group's estimates as linked to the official census or... well I don't have an "or," I think that's it. Not that this doesn't have some value.

http://www.palmbeachpost.com/news/state/florida-could-get-two-new-congressional-seats-939157.html

You must be logged in to read this quote.

EDS annually does reapportionment estimates based on the mid-year Census Bureau estimates released in December, projected forward to 2010.  April 2010, The most recent report shows estimates using different projection intervals, and includes spreadsheets.

http://www.electiondataservices.com/images/File/NR_Appor09wTables.pdf

So ESRI has done some more fine tuned population estimates, which EDS has calculated new estimates.

The changes from the estimates based on projecting Census Bureau estimates are:

FL and TX +1 (to 2 and 4), and NY and MO -1 to (-2 and -1)

Florida had earlier been projected to gain a 2nd seat, and then a slowdown, including net outward migration, had caused the estimate to barely drop below 2.  Texas has fluctuated between 3 and 4, and the 4th will be largely dependent on other states, since on a relative share of the USA population it is short of 35.5.

Missouri had been previously been projected to lose a seat, but had more recently been projected to barely hold on.

I suspect that if there was a margin of error attached to the estimates, that the probability of these last switches would be a lot closer to 50% than 100%.

[jimrtex]

Here is the actual EDS report, including the population estimates from ESRI.

http://www.sos.wa.gov/_assets/elections/NR_Appor2010ESRI_finalwTableMap.pdf

[jimrtex]

Here is the actual EDS report, including the population estimates from ESRI.

http://www.sos.wa.gov/_assets/elections/NR_Appor2010ESRI_finalwTableMap.pdf

How they got that data based off previous estimates seems a bit odd.

For example in Florida the estimated population difference between 09 and 2010 increased by 380,000, that is the 2nd largest yearly increase this decade (03-04 had an increase in 395,000 according to estimates) and between double and triple the increase of the past few years.

NY would be seeing an increase of just 2,000 from 09.  75,000, 35,000 and 66,000 have been the yearly increases the last three years

I think the Census Bureau estimates and the ESRI estimates are independent of each other.

The only connection is that EDS has practice in converting state population totals into apportionment estimates AND disseminating the results in a manner that newspapers, etc. can easily pick up and produce articles.

If the Census Bureau included estimates of error, one could produce probability estimates of apportionment.  It is really not correct to say that Missouri will lose a seat, when it would be better to say that it has an X% chance of losing a seat.

[jimrtex]

The 2009 ERSI estimates match up to the 2009 census estimates.

Do they? 

It looks like the extra seat for Florida and Texas is based on ESRI estimating a larger gain than the Census Bureau, over the decade, and the loss for New York is based on ESRI and the Census Bureau matching.  Similarly the Minnesota/Missouri switch is based on ESRI adding about 1% more to Minnesota.


State                ESRI 2010   Est. 2009  Change
Alabama              4,735,593   4,708,708   0.57%
Alaska                 695,751     698,473  -0.39%
Arizona              6,723,229   6,595,778   1.93%
Arkansas             2,923,603   2,889,450   1.18%
California          37,983,948  36,961,664   2.77%
Colorado             5,114,102   5,024,748   1.78%
Connecticut          3,535,787   3,518,288   0.50%
Delaware               893,724     885,122   0.97%
Florida             18,917,612  18,537,969   2.05%
Georgia             10,014,045   9,829,211   1.88%
Hawaii               1,309,580   1,295,178   1.11%
Idaho                1,581,697   1,545,801   2.32%
Illinois            13,089,726  12,910,409   1.39%
Indiana              6,479,832   6,423,113   0.88%
Iowa                 3,057,995   3,007,856   1.67%
Kansas               2,841,378   2,818,747   0.80%
Kentucky             4,339,471   4,314,113   0.59%
Louisiana            4,507,335   4,492,076   0.34%
Maine                1,338,645   1,318,301   1.54%
Maryland             5,730,892   5,699,478   0.55%
Massachusetts        6,555,736   6,593,587  -0.57%
Michigan            10,104,633   9,969,727   1.35%
Minnesota            5,334,772   5,266,214   1.30%
Mississippi          2,996,685   2,951,996   1.51%
Missouri             6,003,689   5,987,580   0.27%
Montana                983,932     974,989   0.92%
Nebraska             1,822,473   1,796,619   1.44%
Nevada               2,748,294   2,643,085   3.98%
New Hampshire        1,329,915   1,324,575   0.40%
New Jersey           8,822,373   8,707,739   1.32%
New Mexico           2,080,039   2,009,671   3.50%
New York            19,543,731  19,541,453   0.01%
North Carolina       9,552,054   9,380,884   1.82%
North Dakota           662,194     646,844   2.37%
Ohio                11,605,005  11,542,645   0.54%
Oklahoma             3,720,244   3,687,050   0.90%
Oregon               3,865,839   3,825,657   1.05%
Pennsylvania        12,574,407  12,604,767  -0.24%
Rhode Island         1,058,412   1,053,209   0.49%
South Carolina       4,649,749   4,561,242   1.94%
South Dakota           827,263     812,383   1.83%
Tennessee            6,366,430   6,296,254   1.11%
Texas               25,268,853  24,782,302   1.96%
Utah                 2,841,749   2,784,572   2.05%
Vermont                626,078     621,760   0.69%
Virginia             7,965,681   7,882,590   1.05%
Washington           6,756,150   6,664,195   1.38%
West Virginia        1,842,096   1,819,777   1.23%
Wisconsin            5,741,617   5,654,774   1.54%
Wyoming                548,154     544,270   0.71%

[jimrtex]

Do they? 

It looks like the extra seat for Florida and Texas is based on ESRI estimating a larger gain than the Census Bureau, over the decade, and the loss for New York is based on ESRI and the Census Bureau matching.  Similarly the Minnesota/Missouri switch is based on ESRI adding about 1% more to Minnesota.

The 2009 estimates for ERSI and the Census bureau were exactly the same, the ERSI estimated gain between 2009 and 2010 was the 2nd highest yearly gain of the decade, and triple the gain between 2008 and 2009.

Where are the 2009 ESRI estimates?  The ESRI estimates are based on small areas that the Census Bureau doesn't produce estimates for.  Do you really think that ESRI started with the 2009 Census Bureau estimates, and found a 2.8% increase for California?