6 More States for Free Redistricting Tool

Just a quick diary here to let you know I've enabled 6 more states (as of last weekend) for my free redistricting web application. They are Indiana, Louisiana, Massachusetts, Nevada, North Carolina and South Carolina. Next I plan to tackle PA and NY, which are time consuming because of the large amount of data. Thank you for your feedback on my previous diary a couple of weeks ago. Open Redistricting was one of the top feedback items. I've talked to Travis Crum, the person who proposed the Open Redistricting Project (via a Heather Gerken post on Balkanization -- my bad crediting Ms. Gerken) and we are exploring how to collaborate. So stay tuned. Getting political data was also quite popular, but I haven't figured out how to do this consistently. It may end up being part of Open Redistricting, where different individuals can contribute that data in some way. Another item from your feedback is to include a way to prepopulate with current CDs. That should not be too hard to do, so I will try to add that feature in a month or so. You may have already seen this, but andgarden at Swing State did a nice NJ redistricting map. Quick not on the 2008 population estimates. The Census Bureau gives those out by county. I take the additional population (or population loss) for each demographic group and distribute across the census blocks (really census block groups) in the county comensurate with the size of those demographic groups in each block. So it's as if the population grew by the same percentage in each census block. I also make sure that every additional person is the data is accounted for by taking any left over by rounding and adjusting the largest census block accordingly. Cross posted at Daily Kos. Thanks.

Free Redistricting Tool Now for 12 States

The free congressional redistricting web application I launched a few weeks ago now supports 12 states. These are Arizona, Florida, Georgia, Iowa, Michigan, Minnesota, Missouri, New Jersey, Ohio, Oregon, Texas and Washington. Also, the app now has 2008 population estimates. You can find the app here. I will continue to add new states (and make minor fixes) over the next month or so, in order to include all states that may gain or lose congressional seats. After that...what do you think? A number of people thru comments and emails have made suggestions on where to go with this tool. I don't have a ton of time to spend on it, but I want to keep it moving in a useful direction. Some of the suggestions include:

• Expand the number of districts so that the tool could be used for state legislatures; or better yet, support state legislative redistricting more fully by showing existing districts' shapes.
• Make this an open source project that could perhaps become an Open Redistricting Project as written about by Heather K. Gerken. The idea here is to enable citizen activists or commissions to do "shadow redistricting" and thus pressure the powers that be into staying in line.
• Have an option to color according to existing districts as a starting point.
• Add political data for districts. This may be difficult to do for some states.

And a couple of my own ideas:

• Figure out a way to get rid of the extra Windows app to make JPGs.
• Port from C#/Silverlight to Java/Flash and add in Google Maps.

What do you think? Thanks a lot.

Free Congressional Redistricting Tool Ready

I'm nervous about releasing this into the wild. There are bugs, but I think it's good enough now to let you all check it out. So, as promised a few days ago, here is my Redistricting App. With it you can create congressional districts the way you think they should be. It currently only supports Florida, Texas and Washington, but I'll get more states on line soon, especially those that stand to gain or lose seats after 2010. My last 2 diaries show examples of what you can do with the tool. The Redistricting App page gives a quick rundown of the features and a link to a more detailed Help page. Main Features include:

• Select state and number of congressional districts (CDs)
• Use mouse to sweep across map and assign block groups to CDs (block groups usually contain about 4 voting districts; using the larger size makes things more manageable)
• Pan/Zoom
• Save/Open your work in XML (saved in isolated storage)
• Use 2000 census numbers or updated census estimates
• Show/Unshow city, county and old CDs
• Create area maps of zoomed in areas
• Save state and area maps in XML (Separate Windows app can read this and produce JPGs; Silverlight won't allowing saving as JPGs directly; an app is available for download; XML schema also is available)

Caveats:

• Requires Silverlight 2.0 (runs on Windows and Mac)
• To get JPGs you have to run a separate Windows-only tool, due to limitations in Silverlight. Check the help page for more details.
• Loading data for a state from the server is slow, because the data is large, but after that the app is fast and does not go back to the server.

I've worked on this part time over the 5 or 6 months since the election. I've tried maps of 3 states, reworked a lot and come up with this. I certainly have fun playing Redistricter myself. Hopefully you will, too. And maybe this can be useful in helping you make choices in redistricting battles in the future. So, let me know what you think. There's an email address on the main page for more detailed feedback and suggestions. And I'll keep you posted when more states come on line.

If Washington State had 10 congressional districts...

With the likely passage of a bill creating 2 more congressional districts (1 for DC and the other into the mix), it’s possible Washington State could get a 10th CD. Democrats current hold a 6-3 advantage among Washington’s House members. The 6 Democratic seats have been pretty safe in recent years, although WA-03 (Brian Baird) used to be more of a toss-up. Two of the Republican seats take up all of Eastern Washington (east of the Cascade Mountains) and are pretty safe. The other, of course, is WA-08, which covers part of the Seattle-Tacoma suburbs and exurbs, where Dave Reichert (R) defeated Netroots heroine Darcy Burner twice by slim margins. So, with a tenth district thrown into the mix could Democrats gain a 7-3 or even 8-2 advantage? To explore this question, I built a 10-CD map of Washington State, using a software tool I’ve been working on (and you can get, too). First a note on the tool: Next week I’m releasing this easy-to-use web tool for free for anyone to play redistricting. It currently supports Florida, Texas and Washington. I’ll be bringing other states online in subsequent weeks. Check back Monday for a diary on the tool. It uses 2000 U.S. Census Bureau data plus 2007 population estimates (by county). Could Democrats gain an 8-2 advantage? Unlikely for a couple of reasons: Washington has a bipartisan commission that draws the maps. Typically this means incumbent preservation and a reasonably balanced approach to making changes. Also, WA-04 and WA-05 (safe R districts) would shrink in size, leaving a significant area east of the Cascades that would have to be added to other districts. (They used to include all of Okanogan, Chelan and Kittitas counties. Check out WA 2008 Election Results Map to get a feel for what parts of the state are more progressive or conservative.) If too much of eastern Washington were added to WA-03, for example, that could flip to an R. A possible scenario, especially if Dave Reichert (WA-08 R) wins reelection in 2010, is that WA-08 becomes a safer R seat in exchange for a new WA-10 that provides a D advantage, while all other incumbents are protected. This is the map that I’ve drawn. As you can see, I’ve preserved part of the WA-08 base in south King County (where Reichert lives) and eastern Pierce County. But I’ve taken away the “high-tech” suburban areas of Mercer Island, Bellevue, Issaquah, Sammamish and Redmond. WA-08 would still retain the I-90 corridor (Snoqualmie, North Bend) and then jump over the Cascades at Snoqualmie Pass (I-90) to Kittitas County. From there the district follows US 97 over Blewett Pass through Chelan County and up to Okanogan County. (The magenta line is the old CD.) The new WA-10 would become the “high-tech” district including those cities formerly in WA-08 plus Kirkland, Woodinville, Renton and parts of Bothell and SeaTac. (If Darcy tried again in this district, she would have a great chance!) In other areas, districts 1, 2 and 7 remain largely the same. WA-01 (Jay Inslee-D) would creep further into north Seattle and take a little from WA-02 (Rick Larsen-D), while giving up Kirkland and Woodinville to WA-10. WA-09 (Adam Smith-D) would become more compact, centered on Tacoma and would become an even safer D. WA-09 would take away Democratic support from WA-06 (Norm Dicks-D). WA-06 would pick up part of Kitsap county, part of Olympia (a strong D area), but also Chehalis and Centralia (Lewis County), which are more conservative. We could give some of Tacoma back to WA-06, but then the Lewis County areas would pretty much have to go to WA-03, making it more of a risk to Democrats. For WA-03, this plan would probably be a wash. It would lose conservative Chehalis and Centralia, but also lose perhaps a little more of Democratic Olympia than it gains. It loses Pacific, Wahkiakum and Skamania Counties, but they total only about 18,000 voters (2008 election) We'll have to wait until 2011 to see what really happens, but this is one idea of might transpire. Check back next week for information on the tool if you want to try your hand at redistricting Washington State (or Texas or Florida), and I promise I'll get the census data for other states that stand to gain or lose sometime in the near future.

More on Florida Redistricting

Brownsox wrote Sunday about the Fair Districts Florida initiative that is gathering signatures now (A New Hope For Redistricting in Florida). I wrote about this, too, in my last post. So I worked up a congressional district map (with the current 25 districts), to see what fair districts might look like. Here's what I came up with: Click to see it full size. First, the methodology: I've written a web application (soon available for everyone on the web for free) that allows you to create the congressional districts as you want. It's built using Silverlight (Microsoft's Flash-like software). I've used the 2000 census data (population broken down by ethnicity) for census block groups, which are a little bigger than voting districts. For Florida, I simply started in the northwest and worked my way down. I tried my best to keep counties together and to keep smaller cities together. Also, I tried to keep larger cities in as few chunks as possible. And I used no partisan data (which I don't have anyway). This follows the Fair Districts Florida guidelines which say

(1) No apportionment plan or individual district shall be drawn with the intent to favor or disfavor a political party or an incumbent; and districts shall not be drawn with the intent or result of denying or abridging the equal opportunity of racial or language minorities to participate In the political process or to diminish their ability to elect representatives of their choice; and districts shall consist of contiguous territory. (2) Unless compliance with the standards in this subsection conflicts with the standards in subsection (1) or with federal law, districts shall be as nearly equal in population as is practicable; districts shall be compact; and districts shall, where feasible, utilize existing political and geographical boundaries.

The largest deviation any districts has from the average size is 699 people, which is a little more than .1%. Not bad, especially using the larger census block groups. Now, here's what the current districts look like (National Atlas PDF). There are so many differences. Since I do not have voting data for the census block groups, I can't do a more precise analysis on how this would change the congressional makeup, but here are a few comments on areas where my new map is very different and may open up more competetive opportunities: -- CD 3 is entirely in Duval County, the Jacksonville area, instead snaking all the way down to Orlando. Should still be a safe Dem seat. -- Tampa area: currently there are 2 Reps (old CDs 9 and 10) and 1 Dem (old CD 11, which is packed with both the heart of St, Pete and Tampa. In my more sensible districting, St. Pete and Clearwater are have 1 district (CD 13) and most of Tampa is in another (CD 12), both of which would likely be Dem. -- Around Orlando, Seminole County is within only 1 district. Then Orange County is split between 2 districts. Dems have 2 of the districts in that area already (old CDs 8 and 24); perhaps this would give us an opportunity to get another seat (new CDs 6, 8 and 9). -- Around Miami, Ft. Lauderdale and Palm Beach, the new map keeps counties and cities much more together. Reps have 3 districts (old CDs 18, 21 and 25). Mostly likely the new map does not help Dems with at least 2 of those (new 24 and 25), but could facilitate picking up 1 seat in the area. -- Note on Hendry County: it looks odd because the census block groups is very large in area. With the finer grain of voting districts, this could be smoothed out. In any case, though, there are not that many people in that county. Now, I have to admit that the language in the petition is vague enough that those in control will still be able to bend the process in their direction. So, we won't necessarily see maps as straight forward, and non-partisan, as this. But it will be a lot better than what we have now. I expect to have my app available within a month, with a lot more states, so you can try this on your own. That's all for now. Thanks.

Florida Redistricting Petitions

Two Florida constitutional amendment petitions that would restrict gerrymandering are gathering signatures until October 2009. One for congress; the other for the legislature. They need 676,811 signatures to get on the November 2010 ballot. If successful this would result in a significant redrawing of the districts in 2011. Florida will likely get 1 additional congressional seat, too. FairDistrictsFlorida.org is sponsoring the amendments. The Florida Supreme Court recently ruled against those attempting to block the petitions. As I wrote in my previous post, Florida is one of the few large states where the distribution of seats between Dems and Reps is more than 1 seat away from what we might expect. Look at the district maps to see how seriously gerrymandered they are. So, Floridians, sign the petitions, collect signatures, do what you can. And if you know anyone in Florida...this is a great step in the right direction. NationalAtlas.gov has great maps of the congressional districts. Here's the PDF for Florida (800KB), which you can pan and zoom. Some of the districts are incredible (check out FL-16 (the Foley-Mahoney-Rooney district) or FL-23 (which Alcee Hastings (D) won with 82%)). So what would the constitutional amendment regarding congressional districts do? The ballot summary is

Congressional districts or districting plans may not be drawn to favor or disfavor an incumbent or political party. Districts shall not be drawn to deny racial or language minorities the equal opportunity to participate in the political process and elect representatives of their choice. Districts must be contiguous. Unless otherwise required, districts must be compact, as equal in population as feasible, and where feasible must make use of existing city, county and geographical boundaries.

And the main text is actually short and very readable.

(1) No apportionment plan or individual district shall be drawn with the intent to favor or disfavor a political party or an incumbent; and districts shall not be drawn with the intent or result of denying or abridging the equal opportunity of racial or language minorities to participate In the political process or to diminish their ability to elect representatives of their choice; and districts shall consist of contiguous territory. (2) Unless compliance with the standards in this subsection conflicts with the standards in subsection (1) or with federal law, districts shall be as nearly equal in population as is practicable; districts shall be compact; and districts shall, where feasible, utilize existing political and geographical boundaries.

A key point here is that the legislature is still responsible for producing a plan; no commision. But they will be much more constrained than before. From my perspective, there is still a lot of flexibility and therefore opportunity for the legislators to play favorites. However, there will be a lot less opportunity than there is today. If this passes and passes any legal challenges, I'm sure we'll see challenges to whatever plan the legislature comes up with in 2011.

What's the Expected Number of Dems and Reps If Redistricting Were "Fair?"

I've generally thought that the current state of redistricting skews the balance between Democrats and Republicans in the House of Representatives. But how much and in whose favor? And how do we measure?

To get an "expected" value, I took the 2008 vote totals (from The Green Papers) for each House race and totaled them by state. Given those totals, how many Democrats and how many Republicans would be elected from each state if statewide proportional voting were used? And then how many do we actually have? Also, I looked at, for each state, the average victory margin for Democrats and Republicans.

State	# CDs	Expected # Dems	Expected # Reps	Actual # Dems	Actual # Reps	Avg Dem Victory Margin	Avg Rep Victory Margin
CA	53	32.8	20.2	34	19	0.55	0.22
FL	25	12.1	12.9	10	15	0.41	0.22
GA	13	6.6	6.4	6	7	0.45	0.33
IL	19	11.7	7.3	12	7	0.54	0.20
MA	10	8.8	1.2	10	0	0.63	--
MI	15	8.2	6.8	8	7	0.40	0.23
NC	13	7.2	5.8	8	5	0.34	0.25
NJ	13	7.4	5.6	8	5	0.41	0.20
NY	29	19.6	9.4	26	3	0.48	0.25
OH	18	9.4	8.6	10	8	0.29	0.22
PA	19	10.6	8.4	12	7	0.31	0.21
TX	32	13.3	18.7	12	20	0.44	0.42
VA	11	5.9	5.1	6	5	0.34	0.21

The table shows all states with at least 10 House seats. What's interesting is that the actuals for all of the states except Florida and New York are within 1 of the expected, using the "proportional voting" measure.

Now, about the methodology, this is admittedly a crude measure, because people often vote for individuals and may vote a different party if the individuals change. Also, a number of states had races with no opposing Democrat or Republican: California had 6 Dems and 2 Reps who had no opposition from the other major party; Texas had 2 Dems and 6 Reps unopposed; Florida, Georgia, New York and Virginia each had 2 Dems unopposed; Illinois, Michigan, New Jersey and Pennsylvania each had 1 Dem unopposed; and Massachusetts had 6 out 10 Democrats unopposed by Republicans!!! These races can obviously skew the results. (Note: Florida reported no vote totals for the 2 unopposed races, so I counted them as 100,000 votes for the winning Dem.)

I've included in the table the average margin of victory for Dems and Reps in each state. To compensate for races where a party faced no opposition from the other major party, I capped the margin at 75 points. Most states races with candidates from both parties did not show margins greater than 75 points. However, New York was an exception, where CDs 10, 11, 12, 15 and 16 all had greater margins (88, 87, 80, 84 and 93 points, respectively). The interesting thing to note is that in every one of these states, the Democrat's average margin of victory was greater than the Republican's. This would indicate that, overall, gerrymandering is still favoring Republicans.

Looking at individual states, it jumps out to me that Florida is most egregious in favoring either party, and the favored party is the Republicans. By the total vote count, we would expect a split of 12D-13R, but we have 10D-15R. In addition the Democrats average margin of victory is 41 points, while the Republicans is 21 points. As mentioned above, 2 Democrats ran unopposed, so we counted their margins as 75 points, but the other 8 Democrats had Republican opponents. Of those 8, 4 won by 40 points or more (40, 43, 55 and 64, respectively). The others won by 4, 9, 16 and 24 points, respectively. For the Republicans, 11 of 15 races were decided by 25 points or less, and only 1 of those was less than 10 points (FL-25), so most of their wins were in that comfortable, but not wasteful 10-25 point range. Efficient. (Here's the district map.)

New York is interesting because by the total vote count measure, it would seem to favor Democrats pretty heavily. Yet, the average margin of victory for Democrats was 48 points, to the Republicans 25 points. As I noted above, in 5 seats, all in the NYC area, Democrats beat Republicans by more than 75 points. In 2 other seats, no Republican ran (which we counted as 75 point wins). Democrats won the 2 closest races (NY-29 by 2 points and NY-24 by 4 points). Both of those districts have traditionally been Republican, as has NY-20, which Kirsten Gillibrand won by 24 points. So Democrats won 3 traditionally Republican seats fair and square (i.e. by running great candidates). It's hard to see how districts could be drawn upstate to be more fair and still favor more Republicans. (See the map here.) One could make districts 26 thru 29 more compact, but Democrats won those NY-27 and NY-28 by 53 and 56 points, respectively, some the modifications would more likely make NY-29 safer for Democrats and give them a better shot at NY-26. I would also argue that NY-25, NY-22 and NY-19 are districts Republican should have a legitimate shot at, but they have not done well there in the last 2 cycles.

Downstate it's hard to find a way to give Republicans more seats without serious gerrymandering. Of course, they only just this cycle lost NY-13, but they lost it by 29 points in an open seat race. They have some significant voter concentrations out on Long Island, but they lost NY-1 and NY-2 by 17 and 34 points respectively. They won NY-3 (Peter King) by 28 points, but when King moves on, we would expect a much closer race (i.e. not necessarily Republican votes to spare). Except for NY-1 and NY-3 on Long Island and NY-19, which stretches from north Westchester County up to Poughkeepsie, Republicans could not muster 100,000 votes or a closer margin than 28 points in any downstate district. Thus, "fairer" districts would not legitimately give them more seats. Republicans got about 2 million House votes in New York, to the Democrats 4.2 million, But upstate Republican votes are fairly spread out among the districts. They lost the 2 closest races and have competed poorly in traditionally Republican districts update. The best conclusion for New York is that gerrymandering is not costing them seats. They simply have failed to compete effectively (or one might say, have turned voters off) in the last couple of cycles.

California: given the total vote count, it seems pretty much in line with what's expected. But look at the average margin of victory (Dems 55 points, Reps 22 points) and it looks like Democrats should expect more seats. Overall the state is very good at incumbent preservation. Out of the 53 seats only 7 had victory margins of 15 points or less, and of those Republicans won 6 of them. Two of those 7 are in the northeast (CA-3 and CA-4), 1 near the Bay area (CA-11), and the rest in southern California (districts 26, 44, 46 and 50). (See the map here.) To me the Republicans should have nothing to complain about. Perhaps they could take CA-47 if it were carved different, but CA-46 next door is carved for them. Down in San Diego, CA-53 is strongly Democratic, but making it more compact could make CA-50 more competitive. Also, more aggressive Democratic district-making would threaten Republican's CA-52 by swapping some areas of it with CA-51. California districting does not look like it's unfair to Republicans.

And what about Texas? Looking at the total vote count, the split is close to what you would expect. Interestingly, the average victory margin for Republicans is pretty close to that for Democrats (44 points vs. 42 points). Like California, there are few competitive seats. Only 6 out of 32 hard victory margins of 15 points or less. Democrats won 2 of those (TX-17 by 8 points, TX-23 by 14 points). The Republican wins were in districts 7, 10, 22 and 24 by 14, 11, 7 and 15 points, respectively. Even though Democrats lost 5 seats after the 2003 Tom Delay honorary gerrymandering, it's hard to see how Democrats can make significant gains through more compact redistricting, although 1 or 2 seats would be possible. Of course, as noted previously, Texas will probably get 4 more seats, and the population increase is 2/3 Hispoanic. And one last note: what's the deal with TX-29? The total vote count was barely over 100,000 whereas most other districts had close to 200,000 voters. (Here's the map.)

Looking at the rest of the large states, the number of Democrats vs. Republicans is fairly in line with expectations. In Pennsylvania, Ohio and North Carolina, Democrats should not expect more from fairer districts. In all 3 states there are slightly more Dems than the total vote count would predict and the difference in victory margin between Democrats and Republicans is less than 10 points. In Georgia, Michigan, New Jersey, Virginia and Illinois, Democrats possibly could get an extra district in each thru fairer districts, but I have not studied the maps in those states. In those states the differences in victory margin are 12, 17, 21, 13 and 34 points, respectively.

In summary, in the largest states we see lots of safe districts for both parties; no surprise there. Overall the balance between Democrats and Republicans is not far from what we would expect if we had proportional voting. But looking at the difference in victory margins, Democrats could expect to get a handful of additional seats if districts were fairer.

Thanks for reading.