Missouri Redistricting, 2010

[Verily]

Same format as the Iowa and New York redistrictings I've done. Unlike in Iowa, I needed to split counties, but only two were split. County populations are extrapolated to 2010 based on 2007-2008 growth rates, compounding each year.

The seat lost was Republican, either Luetkemeyer or Graves. (They get thrown into a district together.) But Carnahan's district is now fairly vulnerable on the new map, making up for that considerably. And of course Ike Skelton's district is still very Republican but would only flip if he retires.

Also, I didn't bother with even pretending to do a black-plurality district. It's just impossible. Lacy Clay's district was already only barely plurality black, and the loss of a district statewide combined with the fact that Carnahan and Akin's (pre-redistricting) districts are already bleached (9.1% and 2.2% black, respectively) make retaining a black plurality district unfeasible. Clay would presumably easily get renominated in the district I drew for him anyway; it's still at least 40% black.

Without further ado...

[Verily]

Hmmm... Apparently I've drawn Luetkemeyer into Emerson's district, actually. He lives in Miller County. I suppose it would be possible to put him back in the green district with some finagling, but why bother if either he or Graves is going to have to lose anyway?

Most of his former district is now in Graves's district, although a big chunk is in Akin's, too, and some is in Skelton's. And obviously his tiny home county is in Emerson's.

(Miller County is the purple county just south of the meeting of the green, magenta and purple districts.)

[muon2]

I figure I ought to add mine here as well. As usual I went for the minimum number of county fragments while minimizing the population deviation. In this case there is only one county split into two fragments that is inevitable in St Louis County. This particular map gives a maximum deviation of 0.17%.

The unfortunate shape of the KC district results from this minimization of population deviation. A better shape within the 0.5% tolerance would be to group Lafayette, Ray, Caldwell and Carroll with Jackson. I would note that an Iowa style computation of compactness would actually prefer my map below (L shapes being OK in IA.)

[Purple State]

Kind of a random question, but how do you guys figure all of this out?

[muon2]

Kind of a random question, but how do you guys figure all of this out?

I have a spreadsheet programmed with the projection algorithm. I then use the spreadsheet and a map to reach a basic design. Then I look for swaps by hand that would optimize the outcome.

[Verily]

I do basically the same, except my strategy is a little different from muon's. I prefer to achieve community unity sometimes at the expense of a couple extra county splits or slightly more uneven populations.

Also, muon and I differ on how we calculate the 2010 population. I use the estimated growth rate for 2007-2008, which I think better represents current population trends, then extrapolate the 2008 estimated population to 2010 by compounding that formula. Muon uses the estimated growth rate for 2000-2008, breaks that down to an annual growth rate, and uses that to extrapolate from 2008 to 2010, again compounded. He has the advantage of having a more stable number.

My formula might be more affected by single-shot events in the 2007-2008 period which resulted in anomalous declines or increases in population that won't take place in other years. I think this is a relatively small risk since the only counties it affects are very small ones in any case. Muon's formula has the problem of reflecting growth patterns from the first half of the decade which might no longer be affecting current patterns; this is more notable in some places (like Nevada) than in others.

I am, however, using muon's calculations for the 2010 reapportionment.

[Purple State]

Kind of a random question, but how do you guys figure all of this out?

I have a spreadsheet programmed with the projection algorithm. I then use the spreadsheet and a map to reach a basic design. Then I look for swaps by hand that would optimize the outcome.

Is that something you setup? I always find it cool that you can just create these maps. How long does each one take on average? (obviously smaller states take less time)

[Verily]

Kind of a random question, but how do you guys figure all of this out?

I have a spreadsheet programmed with the projection algorithm. I then use the spreadsheet and a map to reach a basic design. Then I look for swaps by hand that would optimize the outcome.

Is that something you setup? I always find it cool that you can just create these maps. How long does each one take on average? (obviously smaller states take less time)

How long it takes depends less on the size of the states and more on the size of the counties. States with lots of small-population counties are really easy, but states with a few large population counties are tough. (Some just aren't even worth doing this for because their counties are too large to avoid splitting nearly every county multiple ways--think New Jersey, Arizona, etc.)

[muon2]

Kind of a random question, but how do you guys figure all of this out?

I have a spreadsheet programmed with the projection algorithm. I then use the spreadsheet and a map to reach a basic design. Then I look for swaps by hand that would optimize the outcome.

Is that something you setup? I always find it cool that you can just create these maps. How long does each one take on average? (obviously smaller states take less time)

How long it takes depends less on the size of the states and more on the size of the counties. States with lots of small-population counties are really easy, but states with a few large population counties are tough. (Some just aren't even worth doing this for because their counties are too large to avoid splitting nearly every county multiple ways--think New Jersey, Arizona, etc.)

New Jersey isn't so bad. The secret there is to use the town (or other municipality) as a fundamental unit. See my post earlier this year. As the post indicates it took a few hours to get that map.

Arizona is more problematic since Maricopa county is not fully divided into subdivision that have estimates. However, the incorporated places do have estimates and that can be used to make a reasonable 2010 division.