Dave's Redistricting App

[JohnnyLongtorso]

Not with the 2010 census numbers.

[DrScholl]

Jackpot.

[Dgov]

The Texas numbers are off.  I seriously Doubt that Houston and Dallas have changed that much in a decade--it looks like someone drew random digits on a color-by-numbers.  There's a ton of 90% Hispanic precincts next to 90% Black next to 90% White ones, so the numbering system is probably off there, and i seriously doubt University park is now Majority-Black while Western Ft. Worth is Majority WHite.

[JohnnyLongtorso]

Sounds like the same problem Illinois had, I'd give Dave a day or so to fix it.

[Sam Spade]

The Texas numbers are off.  I seriously Doubt that Houston and Dallas have changed that much in a decade--it looks like someone drew random digits on a color-by-numbers.  There's a ton of 90% Hispanic precincts next to 90% Black next to 90% White ones, so the numbering system is probably off there, and i seriously doubt University park is now Majority-Black while Western Ft. Worth is Majority WHite.

I'll have to examine.  You do get more situations like that in Houston and Dallas (especially Houston) than you might think, but University Park ain't black, of course.

Anyway, I'll get back to work now. 

[Verily]

Yeah, Texas racial numbers are definitely off since the partisan data was added (maybe the partisan ones, too, didn't check). It looks like the precinct data got randomized within each county, which means its only really obvious in the large, diverse counties. But it's absolutely wrong.

[jimrtex]

Yeah, Texas racial numbers are definitely off since the partisan data was added (maybe the partisan ones, too, didn't check). It looks like the precinct data got randomized within each county, which means its only really obvious in the large, diverse counties. But it's absolutely wrong.

I came across something like that when doing the 2006  gubernatorial race.  IIRC The precinct results from the county election sites wasn't matching the shapefiles from the legislative council because of renumbering of precincts, and other changes.

I think I corresponded with someone with the legislative council and they were doing something like normalizing the data to the 2000 VTDs.

It is possible that the political data is from the legislative council site, and the VTD shapefiles from the 2010 census bureau site?  Or maybe Dave is using the 2010 census racial data, and shapefiles from the legislative council?

The 2010 census data and 2010 census VTD shapefiles do match up, since that is what I used for the senate districts.

[Bacon King]

More states with partisan data: GA, OH, SC, TX, and WI, with the caveat: "Note: The statewide vote totals for Georgia and South Carolina will look very low. That's because those states did not report early and absentee voting by precinct, and there was a lot of early voting. Rather than attempt to distribute them, we've left the precincts as is. It still should give a pretty accurate representation of Dem/Rep strength."

Note that some Georgia counties (Gwinnett is the only one I know of, but I'm pretty sure there are more) do report absentees by precinct and they are included in the App numbers.

[JohnnyLongtorso]

Partisan data is also available for Idaho, which I know everyone was clamoring for.

[JohnnyLongtorso]

The Georgia data has been updated to reflect the absentee/early voting totals.

[Verily]

The Georgia data has been updated to reflect the absentee/early voting totals.

A few precincts, especially around Columbus, seem to be missing all of the data.

[Bacon King]

The Georgia data has been updated to reflect the absentee/early voting totals.

A few precincts, especially around Columbus, seem to be missing all of the data.

Which precincts? It could just be the uninhabited parts of Fort Benning.

[Verily]

The Georgia data has been updated to reflect the absentee/early voting totals.

A few precincts, especially around Columbus, seem to be missing all of the data.

Which precincts? It could just be the uninhabited parts of Fort Benning.

They have residents in the thousands. Check it out yourself.

[I spent the winter writing songs about getting better]

Partisan data is also available for Idaho, which I know everyone was clamoring for.

Actually it's pretty interesting to see what's the strongest Obama LD you can draw. A 66% Obama one is actually possible.

[MaxQue]

Version 2.2 is out.

It allows you to do congressional districts, State Senate, State Assembly and Counties Assemblies, all in the same file.

[JohnnyLongtorso]

Partisan data for FL, MI, NY, and NC is up:

"Florida and New York have only presidential data.

Michigan and North Carolina have some average data. Also, the vote totals are slightly low, I think due to missing absentees, but the numbers are not that significant. In Michigan, at least 1 precinct is missing data.

The average for Michigan is the three 2006 statewide elections: governor, attorney general and SoS. Granholm (D) won the governorship that year, but Republicans won the other two, skewing the average toward Republicans.

The average for North Carolina is broader: all statewide elections (incl Pres) from 2004-2010."

[I spent the winter writing songs about getting better]

A good partisan measure of the area is probably just to average the Obama figures and the "average" numbers.

[JohnnyLongtorso]

Partisan data has been added for IN, NM, OK, and PA:

You must be logged in to read this quote.

Not that exciting, except for PA.

[Bacon King]

Heh, I managed to draw a 51.9% Obama district in Oklahoma. Zig-zags across the state and back taking in every remotely Democratic precinct.

[minionofmidas]

Heh, I managed to draw a 51.9% Obama district in Oklahoma. Zig-zags across the state and back taking in every remotely Democratic precinct.

Color by election is a wonderful tool:



53.4% Obama. 52.0% White.

[davebradlee]

I've uploaded DRA Version 2.2.4, which has a new feature for California, which allows you to work with a subset of the state at a time. Because California is so large, even on a fast machine panning and zooming can be pretty slow. By working with a subset those operations are noticably faster. (And it's not the old feature where it automatically puts a bunch of districts around the center of the screen! :-))

Also, I uploaded Block Groups for California. These were much needed because some voting districts had more than 100,000 people in them -- way too big to reasonably work with.

The new feature for California only, and only using 2010 vote districts or block groups, lets you select between a predefined set of overlapping regions. There's a box in the lower right. This is the way you select regions of the state. For example, if Central is selected, block groups (or voting districts in the center of the state are shown, but if you were to pan up north, block groups for the northern counties would not be shown.

Other new features:
-- additional check boxes under Tools have been moved to More Options and the new features are found there.
-- You can toggle Color By Election between Presidential and Average.
-- You can turn off showing vote district and block group tool tips.

I also fixed a bug where if you saved a DRF using non-US regional settings (comma and period flipped for numbers) you could not load it. That should work now, but will have to be loaded using the save settings as when you saved.

Enjoy.

Dave's Redistricting is a project of ProgressiveCongress.org. You can support the project with a tax-deductible contribution thru the link on the app. Thanks.

[I spent the winter writing songs about getting better]

I've uploaded DRA Version 2.2.4, which has a new feature for California, which allows you to work with a subset of the state at a time. Because California is so large, even on a fast machine panning and zooming can be pretty slow. By working with a subset those operations are noticably faster. (And it's not the old feature where it automatically puts a bunch of districts around the center of the screen! :-))

That's the main reason I haven't made any California maps. Thanks!

[JohnnyLongtorso]

Partisan data for NJ and TN:

"The TN presidential vote counts are about half of what they should be, I think due to early and absentees (although the notes from the team was not specific on this). The overall percentage is about right. In any case, the average is there as well and should provide a useful guideline. Average includes presidential, governor, senate and house races from 2004-2008.

The NJ vote counts are just a little low. Some vote districts changed between 2008 and 2010 and the team had to distribute vote counts across some vote districts. The average includes presidential, governor and senate races from 2002-2009."

[Verily]

The NJ partisan data seems to be missing all of downtown Jersey City, the Ironbound in Newark, part of Trenton, and a couple of stray precincts elsewhere (including most of the city of Salem). Overall probably amounts to a tenth or two in favor of the Republicans, but not a big difference.

[Verily]

Actually, there's something weird in Lakewood, too. There are precincts with 3,000 people with only 10-20 votes in them. Now, this is an Orthodox area, so there will be weird patterns (and all of those precincts are about 90% McCain, which wouldn't surprise me), but as far as I am aware the Lakewood Orthodox have ordinary voter turnout, not ridiculously low turnout, although maybe I am wrong.