Census Projections - Southwest States

[muon2]

jimrtex,
Your method is quite similar to the method I used in TX.  I usually start with counties over a quarter of a CD but less than one CD as seeds since I'd rather not split them, but they can be hard to group. Large counties near an edge are also natural seeds since the state line is a hard constraint on the district. I also group sets of large counties together such as my 5 CDs in the San Antonio - Austin area.

One difference we have in method is when dealing with large counties near a city center county. I generally try to group the outlying county(s) with a few other nearby counties first before grouping it with the city center county. For instance by Houston I found that Galveston and Brazoria (621 K together in 2010) could be grouped with Matagorda, Jackson and Calhoun to create a district within 200 persons of the ideal size. Similarly I then found Fort Bend (591 K in 2010) matched with Austin, Washington and Wharton came to within 500 persons of ideal. Had I not been able to get such a good match those counties would have been used with Harris splits. That was the case with Montgomery County.

[jimrtex]

jimrtex,
Your method is quite similar to the method I used in TX.  I usually start with counties over a quarter of a CD but less than one CD as seeds since I'd rather not split them, but they can be hard to group. Large counties near an edge are also natural seeds since the state line is a hard constraint on the district. I also group sets of large counties together such as my 5 CDs in the San Antonio - Austin area.

One difference we have in method is when dealing with large counties near a city center county. I generally try to group the outlying county(s) with a few other nearby counties first before grouping it with the city center county. For instance by Houston I found that Galveston and Brazoria (621 K together in 2010) could be grouped with Matagorda, Jackson and Calhoun to create a district within 200 persons of the ideal size. Similarly I then found Fort Bend (591 K in 2010) matched with Austin, Washington and Wharton came to within 500 persons of ideal. Had I not been able to get such a good match those counties would have been used with Harris splits. That was the case with Montgomery County.

The main difference is that my method will tend to create more districts completely outside the major urban cores (8 v. 6).  When you extend the districts outward from a large suburban county, the districts will be relatively unstable, with the suburban county (eg. Fort Bend being the dominant county, and the others attached more or less at random).  When a new redistricting happens, the less populous counties are introduced to who their new representative will be.

Am I correct that your projections for 2010 are:

p2010 = c2000 * ((e2005/c2000)**(1./5.25))**10  ?

That is you convert the estimated relative increase in population since the 2000 census into an annual rate (assuming compounding) and then compound that rate over 10 years.  The extra (0.25) represents the difference between the census date of April 1, and the Estimate date of July 1 (0.25 of a year).

If so, I have prepared the 2010 equivalents of my maps (but using the 2006 estimates just released, rather than the 2005 estimates).

[muon2]

jimrtex,
Your method is quite similar to the method I used in TX.  I usually start with counties over a quarter of a CD but less than one CD as seeds since I'd rather not split them, but they can be hard to group. Large counties near an edge are also natural seeds since the state line is a hard constraint on the district. I also group sets of large counties together such as my 5 CDs in the San Antonio - Austin area.

One difference we have in method is when dealing with large counties near a city center county. I generally try to group the outlying county(s) with a few other nearby counties first before grouping it with the city center county. For instance by Houston I found that Galveston and Brazoria (621 K together in 2010) could be grouped with Matagorda, Jackson and Calhoun to create a district within 200 persons of the ideal size. Similarly I then found Fort Bend (591 K in 2010) matched with Austin, Washington and Wharton came to within 500 persons of ideal. Had I not been able to get such a good match those counties would have been used with Harris splits. That was the case with Montgomery County.

The main difference is that my method will tend to create more districts completely outside the major urban cores (8 v. 6).  When you extend the districts outward from a large suburban county, the districts will be relatively unstable, with the suburban county (eg. Fort Bend being the dominant county, and the others attached more or less at random).  When a new redistricting happens, the less populous counties are introduced to who their new representative will be.

Am I correct that your projections for 2010 are:

p2010 = c2000 * ((e2005/c2000)**(1./5.25))**10  ?

That is you convert the estimated relative increase in population since the 2000 census into an annual rate (assuming compounding) and then compound that rate over 10 years.  The extra (0.25) represents the difference between the census date of April 1, and the Estimate date of July 1 (0.25 of a year).

If so, I have prepared the 2010 equivalents of my maps (but using the 2006 estimates just released, rather than the 2005 estimates).

Your formula matches mine. I'll start using the 2006 data on upcoming states. At some point I'll be appying that data to my posted results. I'm curious to see how well the estimates track going forward.

[jimrtex]

This compares 36-district maps for Texas.  The first is based on the 2006 census estimates.  The second is based on projecting the annual rate of increase for 2000-2006 forward to 2010.

As before, I started out with the major metropolitan cores (Houston+Southeast; DFW+Northeast; San Antonio-Austin-Central; Lower Rio Grande; El Paso).

Growth between 2006-2010 meant that Walker and Waller counties were not dropped from the Houston-Southeast area; Atascosa County from the San Antonio area; and southward shift of the Lower Rio Grande area.  The latter two changes permitted the Laredo-Border district to move eastward and southward, and out of the Hill Country.

Growth in the DFW area meant that the areas to the west and south of the Metroplex could be lopped off.  Along with Denton County reaching a population of about one CD, resulted in the outer 2 seats being split north/south rather than east/west.

2006


2010

[Cubby]

Growth in the DFW area meant that the areas to the west and south of the Metroplex could be lopped off.  Along with Denton County reaching a population of about one CD, resulted in the outer 2 seats being split north/south rather than east/west.

Denton is smaller than Collin. So will the 2 of them both be big enough for 1 congressional district to cover each county?

Hidalgo County should be big enough for its own CD too. (779,000 for 2010 based on this year's estimate of 700,634). I just hope they don't do that horrible gerrymandering again where they connect the Rio Grande Valley with San Antonio and Austin. That should have been unconstitutional.

[Undisguised Sockpuppet]

Jesus Christ just use proportional representation on a national already and cut it out with having districts.

[jimrtex]

Growth in the DFW area meant that the areas to the west and south of the Metroplex could be lopped off.  Along with Denton County reaching a population of about one CD, resulted in the outer 2 seats being split north/south rather than east/west.

Denton is smaller than Collin. So will the 2 of them both be big enough for 1 congressional district to cover each county?

Denton is projected to be very close (0.992), Collin way past (1.224).
If you look at the 2006 map, you'll see that Denton would also need 3 smaller counties to reach 1 CD.

The DFW region has 9.998 for 10 seats, so that we can avoid splits outside the 3 largest counties, and be within 150 of the ideal.

Dallas and Tarrant have 6.029, and with the southern district plus DFW 6.981, meanwhile Collin plus the northern district is 2.025.  So one of the Dallas seats takes about 18,000 from Collin County.  Since Collin is thus split into 3 districts, the little bit extra for the Denton district would probably come out of Tarrant.

You must be logged in to read this quote.

My estimates for 2010 are based on calculating the annual rate for 6.25 years (estimates are for July 1, the census date is April 1), and then projecting that forward for 10 years.

  rate = (e2006/c2000)^(1/6.25)
  p2010 = c2000 * rate^10

In the case of Hidalgo, this increases the projected population for 2010 to 793,000, which is the equivalent of 1.126 districts.  I've split it east/west with about 29% of the county in the coastal district that only extends as far north as Kingsville, rather than to Corpus Christi.  Most of the population is in a strip along the border, and it makes more sense to me to simply extend westward from Harlinger to Weslaco, Donna.

The USSC did not rule the McAllen-to-Austin district unconstitutional, but rather that it was a non-counter for purposes of the Voting Rights Act.  But the district court took the opporunity to eliminate not only that district, but the Laredo-to-San Antonio district (for the most part), and did not reinstitute the McAllen-to-San Antonio district of the Frostrocity.

Logically, the 2011 legislature would just pull the districts based in Webb and Hidalgo counties back towards the border (some of the other areas such as El Paso, Brownsville, and Corpus Christi are not that high growth).
But there will be an effort to create another Hispanic-majority district, which means combining 90%+ Hispanic areas along the border with 20-40% Hispanic areas further north, with fingers into concentration of Hispanic areas in cities such as Austin, San Antonio, Midland, Odessa, etc.

My map is clearly a violation of the VRA because it provides a relatively compact district based in Corpus Chisti and Victoria; and packed Hispanics into a narrow (a mere 100+ miles) strip extending from El Paso to Zapata.

[Undisguised Sockpuppet]

What would happen if we didn't count illegals and other noncitizens(immigrants, muslims, blacks, gays, jews) on the maps?

[jimrtex]

What would happen if we didn't count illegals and other noncitizens(immigrants, muslims, blacks, gays, jews) on the maps?

<Insert response-to-idiots.txt>
We thank you for your continuing interest.  Your views along with those of other persons will be given all the due consideration that they merit.
<End Insert response-to-idiots.txt>

[Cubby]

Hidalgo County should be big enough for its own CD too. (779,000 for 2010 based on this year's estimate of 700,634). I just hope they don't do that horrible gerrymandering again where they connect the Rio Grande Valley with San Antonio and Austin. That should have been unconstitutional.

My estimates for 2010 are based on calculating the annual rate for 6.25 years (estimates are for July 1, the census date is April 1), and then projecting that forward for 10 years.

  rate = (e2006/c2000)^(1/6.25)
  p2010 = c2000 * rate^10

Your formula looks like what I do, but your 2010 estimates are always higher than mine. I'll try and show my method for Hidalgo County but I'm not good at algebra.

I take the 2006 estimate (in this case 700,634) and subtract the 2000 census # from it (569,463) This gives me the numerical increase since 2000 (131,171) which I divide into the 2000 # to give me the percentage increase since then. (23.03....%) I divide this by 6.25 and multiply it by 10. This gives me 36.85% for Hidalgo. A 36.85% increase over the 2000 Census # gives me a 2010 estimate of 779,310.

If there's something wrong with this method please let me know.

The USSC did not rule the McAllen-to-Austin district unconstitutional, but rather that it was a non-counter for purposes of the Voting Rights Act.  But the district court took the opporunity to eliminate not only that district, but the Laredo-to-San Antonio district (for the most part), and did not reinstitute the McAllen-to-San Antonio district of the Frostrocity.

So its okay to gerrymander for partisan reasons but not racial ones. Well that shows this country's priorities.

Besides being a Democrat in Texas, what did Martin Frost do to annoy you?

[jimrtex]

Hidalgo County should be big enough for its own CD too. (779,000 for 2010 based on this year's estimate of 700,634). I just hope they don't do that horrible gerrymandering again where they connect the Rio Grande Valley with San Antonio and Austin. That should have been unconstitutional.

My estimates for 2010 are based on calculating the annual rate for 6.25 years (estimates are for July 1, the census date is April 1), and then projecting that forward for 10 years.

  rate = (e2006/c2000)^(1/6.25)
  p2010 = c2000 * rate^10

Your formula looks like what I do, but your 2010 estimates are always higher than mine. I'll try and show my method for Hidalgo County but I'm not good at algebra.

I take the 2006 estimate (in this case 700,634) and subtract the 2000 census # from it (569,463) This gives me the numerical increase since 2000 (131,171) which I divide into the 2000 # to give me the percentage increase since then. (23.03....%) I divide this by 6.25 and multiply it by 10. This gives me 36.85% for Hidalgo. A 36.85% increase over the 2000 Census # gives me a 2010 estimate of 779,310.

If there's something wrong with this method please let me know.

Your method assumes that the assumes that annual numeric increase method is constant.

The other method (which is what Muon has used for projecting 2010 populations) is to assume a constant annual percentage rate of increase.

1.2308^(1/6.25) is 1.0338 or a rate of increase of 3.38%.

1.0338^10 is 1.3941 or a 39.41% over 10 years.

A compounded rate is propably better for growth due to deaths and births (with more people there will be more deaths, and more people giving birth).

It may or may not be correct with regard to migration.  You have to have a certain base population in order to support immigrants, including jobs and roads.  But they have to come from somewhere else, which may not be able to supply people in ever increasing numbers.

You must be logged in to read this quote.

So its okay to gerrymander for partisan reasons but not racial ones. Well that shows this country's priorities.
[/quote]Redistricting is a political issue.

You must be logged in to read this quote.

[Cubby]

Hidalgo County should be big enough for its own CD too. (779,000 for 2010 based on this year's estimate of 700,634). I just hope they don't do that horrible gerrymandering again where they connect the Rio Grande Valley with San Antonio and Austin. That should have been unconstitutional.

My estimates for 2010 are based on calculating the annual rate for 6.25 years (estimates are for July 1, the census date is April 1), and then projecting that forward for 10 years.

  rate = (e2006/c2000)^(1/6.25)
  p2010 = c2000 * rate^10

Your formula looks like what I do, but your 2010 estimates are always higher than mine. I'll try and show my method for Hidalgo County but I'm not good at algebra.

I take the 2006 estimate (in this case 700,634) and subtract the 2000 census # from it (569,463) This gives me the numerical increase since 2000 (131,171) which I divide into the 2000 # to give me the percentage increase since then. (23.03....%) I divide this by 6.25 and multiply it by 10. This gives me 36.85% for Hidalgo. A 36.85% increase over the 2000 Census # gives me a 2010 estimate of 779,310.

If there's something wrong with this method please let me know.

Your method assumes that the assumes that annual numeric increase method is constant.

The other method (which is what Muon has used for projecting 2010 populations) is to assume a constant annual percentage rate of increase.

1.2308^(1/6.25) is 1.0338 or a rate of increase of 3.38%.

1.0338^10 is 1.3941 or a 39.41% over 10 years.

A compounded rate is propably better for growth due to deaths and births (with more people there will be more deaths, and more people giving birth).

It may or may not be correct with regard to migration.  You have to have a certain base population in order to support immigrants, including jobs and roads.  But they have to come from somewhere else, which may not be able to supply people in ever increasing numbers.

What does "^" mean?

I understand wanting to have a constant rate of percentage increase, I'm just not sure how you do it. The above formula is too complex for me to understand.

You must be logged in to read this quote.

(and other maps)
[/quote]

Yikes! Thats terrible.

[True Federalist (진정한 연방 주의자)]

I like fractal geometry, but that's plain ridiculous.

[jimrtex]

1.2308^(1/6.25) is 1.0338 or a rate of increase of 3.38%.

1.0338^10 is 1.3941 or a 39.41% over 10 years.

What does "^" mean?

It is an exponentiation operator, the first number raised to the power of the second number.  I'll rewrite with superscripts.

1.2308^1/6.25 is 1.0338 or a rate of increase of 3.38%.

1.0338¹⁰ is 1.3941 or 39.41% over 10 years.

If you use a spreadsheet it would be:

= C2000 * POWER( POWER(E2006/C2000),1./6.25), 10. )

Where C2000 is the 2000 Census population, and E2006 is the 2006 estimate.