Our friends at the Cook Political Report have released an updated Partisan Voting Index that now includes the results of the 2008 presidential election. They've also made a small change in the formula used to calculate PVIs, and I think it'll be instantly recognizable to SSPers:
To determine the national average for these latest ratings, we have taken the average Democratic share of the two-party presidential vote for 2004 and 2008, which is roughly 51.3 percent, and that of Republicans, which is roughly 48.7 percent. So, if John Kerry captured 55 percent of the vote in a district and Barack Obama carried 57 percent in the district four years later, the district would have a PVI score of roughly D+5. (Emphasis added.)
As we discussed at length, the old PVI formula compared district-level results for the past two presidential elections to nationwide results for only the most recent election. This choice sparked plenty of debate, and some folks even suggested we use our own "SVI" that would compare 2004 to 2004 and 2008 to 2008.
Fortunately, the debate has been resolved. As you can discern from the description above (the key part is in bold), Cook has decided to revise its methodology along the lines proposed by people here. Charlie Cook (an SSP reader, as is House editor David Wasserman) told me he wanted something that was "totally apples and apples," and I agree with the choice. Ultimately, this means that the new PVIs will be about two points bluer than under the old system - e.g., a district that would have been R+10 will now come in at R+8.
You can find the new PVIs by partisan rank in this PDF, as well as by member name and by state/district. There's also a giant-size map and a cool chart showing trends in the PVI over the last decade. (As you'd expect, the number of "competitive" districts, at least on the presidential level, has been shrinking.) Have fun!
DavidNYC's great post about House election winners from 2008 who underperformed their district's presidential numbers got me thinking. It left me wondering: what about the incumbents who outright lost? Were there a lot of incumbents who overperformed their district and still lost?
It turns out, yes, there were quite a few. (In fact, it's not a difficult question at all; you can just reverse-engineer the previous diary to find the overperformers. For instance, if Jim Himes underperformed by 16 points, Chris Shays necessarily overperformed by 16.) So, while doing this, what turned out to be interesting was who the truly pathetic figures were... the ones on both sides of the aisle who underperformed their districts' leans on their way down to their own ignominious defeats. By doing this, we can separate out the representatives who simply got swamped by a wave from those who lost purely on their own merits.
Let's start with our five Democratic casualties:
State
CD
Member
Party
Dem Margin
Obama Margin
Difference
TX
22
Lampson
(D)
-7
-17
+10
KS
02
Boyda
(D)
-4
-12
+8
LA
06
Cazayoux
(D)
-8
-16
+8
FL
16
Mahoney
(D)
-20
-5
-15
LA
02
Jefferson
(D)
-3
49
-52
Bill Jefferson takes the prize for futility in 2008, with a 52-point disparity. Tim Mahoney also had a run-in with his own petard, while Nick Lampson, Nancy Boyda, and Don Cazayoux overperformed their dark-red districts and still lost.
Now for the Republicans:
State
CD
Member
Party
GOPer Margin
McCain Margin
Difference
CT
04
Shays
(R)
-4
-20
+16
NV
03
Porter
(R)
-5
-12
+7
OH
01
Chabot
(R)
-5
-11
+6
MI
07
Walberg
(R)
-2
-6
+4
MI
09
Knollenberg
(R)
-9
-13
+4
FL
08
Keller
(R)
-4
-6
+2
PA
03
English
(R)
-2
0
-2
VA
02
Drake
(R)
-5
-2
-3
VA
05
Goode
(R)
0
3
-3
NY
29
Kuhl
(R)
-2
3
-5
NC
08
Hayes
(R)
-11
-6
-5
CO
04
Musgrave
(R)
-12
1
-13
FL
24
Feeney
(R)
-16
2
-18
ID
01
Sali
(R)
-1
26
-27
Not surprisingly, moderates Chris Shays and Jon Porter did what they could but simply got drowned by the blue tsunami in their districts. On the other hand, several unlikable wingnuts like Tim Walberg and Steve Chabot also overperformed, indicating that despite their out-of-whackness with their swing districts, they were primarily wave victims.
Instead, the majority of the GOPers who lost underperformed, some badly. Only two underperforming Republicans (Thelma Drake and Robin Hayes) were in districts that Obama narrowly won and thus have at least a partial excuse. The rest were in districts that McCain won, and have nothing to assign blame to other than their own loathsomeness. Bill Sali takes home top honors, managing to take a district that McCain won 62-36 and stil lose.
Now that we have complete presidential results by CD for the entire nation, we can take a look at how members of Congress fared compared to the top of the ticket in each district. The vast majority of Congressmen and women typically perform better than their party's presidential candidates. The reasons for this are plain: Most members don't face serious challenges, and individual Congresscritters can tailor their politics to suit their CDs far better than any presidential office-seeker, who (in theory, at least) has to appeal nation-wide.
Some Congressmembers, however, invariably run behind the ticket. First up are the laggard Republicans:
State
CD
Member
Party
GOPer Margin
McCain Margin
Difference
WY
AL
Lummis
(R)
10
32
-22
LA
4
Fleming
(R)
0
19
-19
KY
2
Guthrie
(R)
5
23
-18
AK
AL
Young
(R)
5
21
-16
LA
1
Scalise
(R)
31
47
-16
NC
10
McHenry
(R)
15
27
-12
OH
2
Schmidt
(R)
7
19
-12
SC
1
Brown
(R)
4
15
-11
CA
4
McClintock
(R)
0
10
-10
TX
22
Olson
(R)
7
17
-10
KS
2
Jenkins
(R)
4
12
-8
LA
6
Cassidy
(R)
8
16
-8
MO
9
Luetkemeyer
(R)
3
11
-8
NC
5
Foxx
(R)
17
23
-6
AL
3
Rogers
(R)
8
13
-5
MN
6
Bachmann
(R)
3
8
-5
AL
4
Aderholt
(R)
50
53
-3
AZ
3
Shadegg
(R)
12
15
-3
TX
7
Culberson
(R)
14
17
-3
GA
10
Broun
(R)
21
23
-2
AZ
2
Franks
(R)
22
23
-1
SC
2
Wilson
(R)
8
9
-1
UT
3
Chaffetz
(R)
37
38
-1
Most of the folks on this list are freshmen who are almost all certain to do much better in 2010. A handful of others are in extremely red districts to begin with, making any difference between their performance and McCain's mostly a matter of minor noise.
Some, however, stand out for reasons all their own, and their underperformance signals a weakness which could potentially be exploited (again). They include the ethically plagued Don Young, the hapless Patrick McHenry, the well-hated Jean Schmidt, the befuddled Henry Brown, the batshit Virginia Foxx, the caught-napping Mike Rogers, and the loose-lipped Michele Bachmann. While all sit in red districts of varying difficulty, each could be vulnerable (particularly Brown and Rogers, I feel).
Now for the Democratic list:
State
CD
Member
Party
Dem Margin
Obama Margin
Difference
CT
4
Himes
(D)
4
20
-16
MI
13
Kilpatrick
(D)
55
70
-15
IN
7
Carson
(D)
30
43
-13
ME
1
Pingree
(D)
10
23
-13
PA
11
Kanjorski
(D)
3
15
-12
CA
8
Pelosi
(D)
62
73
-11
NM
1
Heinrich
(D)
11
20
-9
OH
15
Kilroy
(D)
1
9
-8
NV
3
Titus
(D)
5
12
-7
CA
6
Woolsey
(D)
48
54
-6
IL
7
Davis
(D)
70
76
-6
NY
15
Rangel
(D)
81
87
-6
OH
1
Driehaus
(D)
5
11
-6
GA
13
Scott
(D)
38
43
-5
MI
7
Schauer
(D)
2
6
-4
MI
9
Peters
(D)
9
13
-4
IA
2
Loebsack
(D)
19
22
-3
IL
4
Gutierrez
(D)
69
72
-3
VA
11
Connolly
(D)
12
15
-3
CA
9
Lee
(D)
76
78
-2
CO
1
DeGette
(D)
48
50
-2
FL
8
Grayson
(D)
4
6
-2
FL
23
Hastings
(D)
64
66
-2
IL
1
Rush
(D)
72
74
-2
OH
10
Kucinich
(D)
18
20
-2
PA
2
Fattah
(D)
78
80
-2
WA
7
McDermott
(D)
67
69
-2
CA
35
Waters
(D)
69
70
-1
CO
2
Polis
(D)
29
30
-1
IL
2
Jackson
(D)
79
80
-1
MN
5
Ellison
(D)
49
50
-1
NJ
3
Adler
(D)
4
5
-1
OH
11
Fudge
(D)
70
71
-1
VA
8
Moran
(D)
38
39
-1
WI
8
Kagen
(D)
8
9
-1
Once again, the vast majority here are freshmen. There are also quite a few more folks in absurdly blue districts (much bluer than the most Republican districts are red). A few stand out as potentially more than just noise, though. Nancy Pelosi puts in a token appearance here - obviously you're going to take a few dings if you're the party leader, but not quite enough for us to be welcoming Congresswoman Sheehan as our new overlord.
Anyhow, I wouldn't be surprised if Artur Davis's long-manifest desire to run for higher office hurt him with the folks at home. (UPDATE: It was Danny Davis in IL-07, not Artur Davis in AL-07.) Meanwhile, Lynn Woolsey's outspoken, er, leadership of the Progressive Caucus might be chafing at her favorability ratings. Charlie Rangel, of course, endured tons of bad press thanks to his tax problems. And David Scott had to face down all kinds of bullshit from Deborah "The Defrauder" Honeycutt. None of these seats, of course, could ever turn red (is Anh Cao laughing at me as I say this?) - it's just that their current inhabitants had (and in some cases still have) varying "issues" in front of them.
As for more serious situations, it's no surprise to see Paul Kanjorski on this list. Carolyn Cheeks Kilpatrick also has a featured spot, thanks undoubtedly to the serious primary challenge she got last year and her bellicose defense of her disgraced son Kwame (the ex-mayor of Detroit). I wouldn't be surprised if she got primaried again. Otherwise, I don't see too many vulnerable Dem veterans on this list - but Kanjorski and Kilpatrick seriously need to consider retirement.
UPDATE: Himes's margin of victory was actually four points, not one - I had failed to include the votes he got on the Working Families Party line.
Swing State Project is pleased to announce a final and official tally of the presidential election results in all 435 congressional districts. As you might recall, when we last left off a few weeks ago, we were still six districts shy, with only Nassau County, NY and Tuscaloosa County, AL outstanding. With these last few counties relinquishing their data, we can wrap up those last few districts, making Swing State Project the first outlet anywhere, blogosphere or elsewhere, to provide a full public release of presidential election results by CD... not just of percentages, but a transparent display of the underlying precincts, if you're interested in delving that deep. If you want to bookmark the summary of the percentages for all districts for future reference, click here.
If you're looking for additional detail about previously-discussed districts, see waves one, two, three, four, five, and six. For a truly ridiculous level of detail, each state's database is accessible through our master database.
As you can see, AL-06 was one of our roughest districts, in about a three-way tie with AL-04 and TX-13 for worst Obama performance. And while Obama won Nassau County on Long Island by a decent margin, he didn't improve on Kerry's numbers by much. In fact, the 2008 numbers in both NY-03 (the Republican part of Long Island, relatively speaking) and NY-05 (a mix of Nassau County's wealthiest areas and working-class white and Asian parts of Queens) matched the 2004 numbers exactly. The more diverse NY-02 and NY-04 saw larger improvements.
As with the last few waves, our commitment to accuracy compels us to issue a few more minor adjustments as we've refined our databases and/or gotten newer numbers. And, for one last time, thanks to jeffmd, Democratic Luntz, californianintexas, Benawu, Benjamin Johnstone-Anderson, and all the other SSP contributors who helped out anonymously... you all put the "crowd" in crowdsourcing.
The waves keep getting smaller and smaller, as we wend our way closer to the conclusion of our massive presidential results-by-congressional district crowdsourcing project. For those of you who are counting, that leaves only six districts that we need to complete (AL-06 and AL-07, NY-02, NY-03, NY-04, and NY-05) in order to be not just the first outlet to make all this information public, but just plain the first outlet, period.
The geography nerds among you might be thinking, hey, that looks like we're only two counties short of completion: Tuscaloosa County, AL, and Nassau County, NY. (You're almost right: we also need Coosa County, AL, but it has only 12,000 people so I'm making a "close enough" call on AL-03 until we actually wrangle some data out of them.) Our ground forces in Alabama are already on the case of Tuscaloosa and Coosa Counties, but, to expedite matters, we need to switch on the SSP Batsignal over Gotham: we need an NYC-area correspondent to make the trek out to Mineola and have a date with the Nassau County Board of Elections' copy machine. If you're available to take this mission, please e-mail our intrepid publisher, DavidNYC (see the right column) and he'll tell you what we need.
If you want to see a summary of the whole list of districts, click here. Waves one, two, three, four, and five provide additional detail, and for a truly ridiculous level of detail, each state's database is accessible through our master database.
Points of interest in this wave include AL-04, which, to our surprise, plummets past the West Texas districts to grab the dubious distinction of Obama's worst performance (at 22%). This district used to send a Democrat to Congress until 1996, and even Gore got 37% here... but this is Alabama's whitest and most rural district, where the southern end of the Appalachians and Birmingham exurbs meet.
Aside from some stagnation in NY-27 (the blue-collar white parts of Buffalo), everything else here is good news: huge swings in both Denver and its conservative suburbs, and even bigger swings in Indiana, where we not just flipped IN-02 (South Bend) but won it pretty convincingly.
As with our previous wave, our resident numbers guru jeffmd has been refining our figures as new data continues to trickle in, so we have another corrections table with 16 revised districts over the flip. Again, nothing major, but we know that many SSP readers are fans of utter and complete accuracy.
The presidential-results-by-congressional-district crowdsourcing project at Swing State Project just keeps rolling along, and we're really getting close to total completion. We're adding 31 more districts today, having scored precinct-level data from some of the largest counties still outstanding (most notably, Queens County, NY, and Wayne County, MI). That leaves only 18 districts with problem counties left to go! (Nassau and Erie Counties, we're lookin' at you...)
As always, big thanks to all SSP readers who've contributed to this project, with extra thanks for this batch to Democratic Luntz and jeffmd, who rocks an Excel pivot table like nobody's business. If you want to see a handy list of all districts in one place, look here. If you want a fuller picture, waves one, two, three, and four are here. And if you want an absolutely crushing level of detail, just click on our master database and then on a particular state to see each district in all its precinct-level glory.
So what are the highlights and lowlights for this installment? As we've seen earlier, California just went from dark blue to even bluer, and that seeped all the way down to some of its reliably red districts (CA-24 in Ventura County went narrowly for Obama... which hopefully will convince Elton Gallegly of the many botched retirement attempts to actually get off the pot this time... and we even came close in CA-19, which stitches together the Sierras and the whiter parts of the Central Valley). A lot of that movement may have to do with California's huge Latino population, fed up with the GOP's increasing reliance on immigrant-bashing; parallel movement is seen in Texas, where two mostly-Latino districts (TX-15 and TX-27) also show wide swings in the Democratic direction.
Also, as we've seen in other districts, Indiana had some of the biggest Democratic swings in the nation, simply by virtue of the Democrats showing up and competing there for once. Check out IN-06. Remember, this is the district represented by Mike Pence, arch-wingnut who just got promoted from leading the RSC to the #3 position in the whole GOP caucus... and now he's in a district that McCain won by just 6 points.
On the bad side of the ledger, we're seeing continued declines in some of the blue-collar white-ethnic districts in the NYC area. These districts suffered some of the biggest declines in that nation from Gore to Kerry, and I thought that might be a temporary 9/11 effect since those districts were some of the ones hardest hit. However, we've continued to lose ground in NY-09 (the old-school parts of Brookyln and Queens), and are stagnant in NJ-04 (Ocean and Monmouth Counties, where people from NY-09 go to retire). Not that it matters too much; these districts are outweighed by the overall blue trends in these already-blue states. And in NY-09 they still managed to kick out state senator Serphin Maltese to finally flip control of the New York senate; Obama's performance may have to do more with 2008-specific racism/latent PUMAism than an overall trend.
Also troubling is what's going on in eastern Ohio, where we lost ground in OH-06 and OH-17. It's not hard to explain -- OH-06 is considered the Appalachian part of Ohio, while OH-17 is centered on Youngstown, a place similar to Pittsburgh's collar counties where the once-strong union base is dying off or drifting away as the manufacturing sector evaporates. This is more worrisome since Ohio is a swing state where every vote counts, but as this part of the state is hollowing out while the Columbus and Cincinnati areas are starting to move into our column, it's not a killer.
Finally, I'm making good on my promise of some updates, based on further refining of early-vote or split-precinct data, or finding more data from small counties where we'd previously made a "close enough" judgment. As you'll see, the numbers haven't moved that much, with a few exceptions (perhaps most significantly in IL-18, which we originally thought Obama had won by a few thousand votes but turns out he lost by a few hundred; see also improvements in FL-06 and FL-07, SC-01 and SC-06). This will only be of interest to people who are real sticklers for accuracy or who are keeping their own spreadsheets on this subject. (Of course, since we're talking about Swing State Project here, that probably describes most of our readership!) The updates tables is over the flip...
Since we now know the presidential results in most congressional districts for 2008, we now possess all the tools we need to calculate the new PVIs.
Charlie Cook's official results will be out in a few months, but unlike in 2004 (where results from 2000 had to be fitted to the new congressional districts) there's no reason we shouldn't jump the gun and have our own figures ready. And there's every reason to want to know who's representing their districts and who's a lot more vulnerable than they used to be.
I'm therefore going to attempt to calculate the new PVIs for the states listed in the diary title. They were picked because we have all the results from districts in those states, and because they represent around 20% of America's congressioanl districts and I'm too lazy right now to do more than that in a sitting.
My methodology conforms to that given in David NYC's comment to DGM's diary. My figures are taken from this spreadsheet. My figures are approximate and you should probably ignore everything beyond the decimal point, because I'm using data rounded to the nearest whole number for the district-by-district results.
When we at SSP first hatched the idea of compiling some numbers for presidential election results for congressional districts, we were thinking we'd be lucky to get to 60 or maybe 100 districts. After all, we couldn't track down precinct-level data for hundreds of counties, sort out what precinct goes into what district, and pick apart large metro counties with thousands and thousands of precincts... could we?
Well, with the aid of SSP's crack cadre of some of the brightest and most tenacious elections geeks out there -- in particular the relentless number cruncher jeffmd and master BoE cajoler Democratic Luntz -- we're closing in on completing all 435 districts. With another 54 added to the pile today, we're near the 90% mark, with only 51 remaining incomplete. If you want to see all district percentages so far, the link is here; you can also check out the diaries where we released the numbers in more detail here, here, and here.
Some points of interest to check out in this batch: look at PA-06, with some of the steepest improvement in all of Pennsylvania. Any question why Jim Gerlach may be planning to cash it in and run for governor in 2010? It might be because his district just shot past PA-07 and PA-08 to become the bluest all-suburban district in the Philly area.
We have data for most of upstate New York (except for Erie County, where Buffalo is), and it's striking that Obama improved on Kerry at a much greater clip upstate than in the NYC metro area. One thing that might give us some optimism heading into the NY-20 special election is the nearly 6-point improvement, as well as the fact that the Dem candidate actually won the district in the first time since, well, probably Barry Goldwater. But this is pretty typical across upstate NY, as we also flipped NY-23 and NY-24, moved NY-25 from swing to pretty safe D, and almost even won in New York's reddest district of NY-29. Compare this with, say, the whiter urban districts, like NY-08 or especially NY-13 (Staten Island and white ethnic parts of Brooklyn), where Obama lost narrowly while barely improving on Kerry's numbers, and thus nearly overtaking NY-29 as New York's reddest district.
The biggest improvements here, as in previous installments are in the Mountain West. This is plain to see in Colorado, not just in the 2nd (where the improvement over 2000 is gigantic, although that may have to do with the huge Nader effect among Boulder's granola-munching crowd) but also in the 4th, where Obama lost by less than a point where Gore lost by 20. And although we didn't come even close in Utah, some of the biggest percentage gains were there. Look for UT-03 to lose its worst-PVI-in-the-nation status, as Obama made up 9 points there on Kerry.
Is there any bad news to report here? Well, we came oh-so-close to flipping OH-14 in Cleveland's suburbs (fewer than 1,000 votes), while not moving the numbers much there. And we lost ground in AL-05, the Appalachian portion of Alabama, and PA-04, which, like PA-12, is in the collar counties around Pittsburgh where the Rust Belt fades into the Appalachians.
Probably least appetizing are the numbers out of Oklahoma, but even it provides some interesting insights into the changes from the old Democratic coalition to the current Democratic coalition. Most of the state stayed in neutral over the decade, but compare OK-02 (rural NE Oklahoma around Muskogee) vs. OK-05 (Oklahoma City). We're getting absolutely hammered in the 2nd, a traditionally Yellow Doggish area that Gore almost won. On the other hand, we shot up in the 5th, the most cosmopolitan part of the state.
So what's left to do? Our main task is, at this point, getting data from counties who have been unresponsive or are charging an arm and a leg for it. If you're interested in helping out, check out this diary for a primer; here's our database of elections boards to contact. And, as always, here's our master crowdsourcing database... although, as you might notice, most of those blanks have been filled in! Thanks to you guys, of course.
One final caveat: these numbers are subject to change slightly, as we refine the data. In fact, in a few days I'll be posting a list of several dozen updated districts. None of these changes should amount to more than a fraction of a percentage point, but caution is warranted where a fraction of a percent would make a lot of difference in how the district is perceived (for instance, PA-03, where a very small revision could make all the difference in terms of McCain's 17-vote margin in the district).
A number of people who've expressed interest in helping gather election data so that we can compile presidential results by CD have asked for a more detailed guide. So here's what I suggest:
2) Find a county where the three right-hand columns (F, G & H) are all blank. (If there's information in any of those, it means someone has already requested data from that county, or at least investigated it.)
3) Call the phone number listed in column D. (If there's no phone number, please look it up and paste it into column D.) When I call, this is what I like to say:
Hi. I'm a researcher looking for detailed election results from the 2008 election. Whom might I speak with about this?
Once I have the right person, this is the request I make:
I'm interested in precinct-level results for the Presidential and Congressional races in your county for the 2008 election. Are you able to send that to me?
That's really you need to ask for - this request is very basic and should be readily understood. If you encounter any confusion, report back here in comments and we'll try to figure out what the misunderstanding is.
4) At this point, the response you might get will vary. Some election officials will email you on the spot, some will only mail you hardcopy versions, and some might even insist on mailing you a CD. Still others might ask you to fill out a particular request form, or fax them a signed letter, or file a Freedom of Information request.
Just ask what you need to do and you should be given straightforward instructions. If you follow these, you should get the data you're looking for without a problem. Note: If the county you talk to can only send hardcopies but you don't have access to a scanner, let us know in comments.
5) Side note: Some counties - and this really cheeses me off, but there's not much we can do - may require a payment for the data. If that happens, I recommend you do NOT pay for the data. Rather, find out how much the data would cost. Then open up this spreadsheet again and type your username into column F and the cost into column G. We'll look into making purchases later.
6) Once you get the data, please upload your files to Scribd. (You'll need to create an account there first.) Then, post the URL(s) in column H. That way we'll know we have the data, and we'll know where to find it.
That's really all there is to it. If you have any questions at all, please feel free to ask in comments.
The good news: We've scrounged up precinct-level election results for about twenty counties that were on our list in order to complete the presidential results by CD. The bad news: We still need data from another ninety.
The bottom line is that we've downloaded from every website that lets you download, and we've emailed every county that lets you email. The remaining counties either don't have websites or email address, or just haven't responded to emails. So we need to start making phone calls.
I think the netroots - and really, we're just talking about a single small blog here - could make a big impact by releasing a complete set of data. Before we started, I never thought that doing so would be possible, but now I believe it's in our grasp. Finishing this would demonstrate that a dedicated band of volunteers can tackle a project most would assume would require a bunch of professionals and a lot of money.
It would also demonstrate that when it comes to data analysis, the Internet really has ushered in a new era of openness, transparency and accessibility. Indeed, our work has already been favorably cited in places like the Guardian and Roll Call, and in local newspapers as well. We're breaking barriers, people!
Alright, enough with the attempts at rousing exhortations. There are still phone calls to make - the full list is here. If you have some free time during the day and can make a few calls, this short list of "high value" municipalities is a good place to start:
Jurisdiction
CDs Covered
Would Let Us Complete
New York City, NY
13
12
Wayne, MI
4
4
Santa Clara, CA
4
3
Ventura, CA
2
2
Fountain, IN
2
2
Fall River town, MA
2
2
St. Louis City, MO
2
2
Josephine, OR
2
2
Cass, TX
2
2
Getting precinct-level data for these counties/cities/towns (especially those toward the top of the list) is key, but all are important. If you want to try another route, start with your home state. If your home state is not on the list, then just pick some counties at random.
Remember that when you call, you need to ask for precinct-level results for both the presidential race and any United States House races within the county in question. (Without the latter data set, we can't figure out which precincts are assigned to which CD.)
Also note that if you see a notation in the "Data Requested" column, or a price listed in the "Cost" column, that means we've already made contact with that county, so there's no need to call them. (Mostly we're waiting to figure out if we can find a sugar daddy to pay for the data from the counties which charge - grr! The nerve of them!)
If you do make a phone call and request the data, please make a note of it in the proper column. If the county emails it to you, great - just upload it to Scribd or Google Docs and post the URL in the spreadsheet. If they offer to mail it to you, please make a note of your request date (along with your name or user name) so that we can follow up if need be. And if they quote you a price, hold off on ordering - just note the price in the proper column.
Again, the full list of counties we need data for is right here. Let's do this thing!
SSP's Presidential Results by CD project continues to chip away at the outstanding districts left up on our big board, and I'm pleased to announce that today, thanks to the tireless efforts of SSP hero jeffmd and the lawyerly phone charm of DavidNYC, we now have complete data for New York's newly-open 20th District.
A traditionally GOP district, Bush carried the 20th by a 54-46 margin in 2004, and also smacked Al Gore by a 51-44 spread four years earlier.
That's one more district that we can take down from our big board, but we still have plenty more to go. If you enjoy the work that this project has produced, please consider lending a hand by calling county offices to inquire about obtaining precinct data. We can do this thing, but we need more people to participate in lightening the load. More details are available here.
Thanks to the heroic efforts of this community, we were able to post a third wave of presidential results by Congressional district yesterday. And as you can see from our perma-post, we now have numbers for an impressive 332 districts. That leaves us with just over 100 districts to go, and as you'd expect, these are some of the thorniest.
The real problem is access to data. Most of the more recently-posted numbers have been drawn from precinct-level data. This sort of information is usually only available at the county level, rather than from Secretaries of State. Many counties make this data easily accessible on the web (some in more usable forms than others), but some don't offer it online at all - and those are the counties we need to tackle. Some offer it, but charge for it. And some don't even appear to have websites.
The bottom line is that there are about 100 counties whose precinct-level data we need in order to finish this project - and we're only gonna solve this problem by throwing as much manpower (or dare I say, people power?) at it as possible. To that end, we've created another crowdsourcing spreadsheet for folks to start tackling this.
The first order of business is finding out contact information for the county Boards of Election. Many won't have email addresses, and even for those that do, emails may go unanswered. So that means we're going to have to start making phone calls. I can't stress this enough: Please be super-polite when making these calls. These are hardworking folks who ensure our elections are run properly and probably don't do this for much money.
Anyhow, if you do call up a county BoE and request the data, please put the date of your request & your name (username is fine) in column E. This will help us avoid inundating the BoEs with multiple requests. Some might be able to email you the data, in which case you can just directly upload it. (For spreadsheets, please upload them to Google Docs. For PDFs, please use Scribd - PDFs on Google Docs can't be shared. Spreadsheets which exceed Google's 1MB limit can also be uploaded to Scribd.) When you do, please put the URL in the right-most column.
Some counties might insist on mailing you a disk. It's even possible that some will only want to send you a hard copy. If this turns out to be the case and you don't have access to a scanner, please make a note on the spreadsheet so that someone else who does have a scanner can make the request instead.
Finally, some (maybe a lot) of these counties will try to charge us for the data. I think that's a load of bollocks, seeing as the information has already been gathered (and paid for with tax dollars) - obviously the cost of distributing it is zero. But this is something we just have to live with. Anyhow, if you do encounter a county which charges a fee, PLEASE DO NOT ORDER THE DATA AND PAY FOR IT YOURSELF. I'd hate for us to make duplicate orders and wind up wasting money. Please just make a note of the cost in column F and we'll revisit this soon in a co-ordinated fashion.
A complete list of counties we need (plus a few cities and towns) is below the fold, and of course in our crowdsourcing spreadsheet. Oh, and please share your tips/experiences in comments. Let's get to work!
The elves were busy while I was taking Christmas off, and now that I've picked the crowdsourcing project back up, we've made another big jump, taking us to the point of having presidential election results for 3/4s of all congressional districts.
Results from the first wave are here, and results from the second wave are here. If you want to see all results in one place, they're permalinked here. Also, please check out our master database; although we've made a lot of headway, there's still plenty to do if you have access to precinct-level data (however, the remaining states are the ones that tend to be most coy about releasing precinct-level data, so those remaining districts may never see daylight until Polidata somehow solves those enigmas).
A few words about some of the states. Many of you have already seen the California numbers, which californianintexas published in her excellent diary; for those of you who haven't, here they are on the front page again. There are unfortunately some California districts missing; a number of large counties (Santa Clara, San Joaquin, Ventura, and Fresno especially) haven't provided precinct-by-precinct data, so districts incorporating parts of those counties can't be completed.
The missing precinct-level data problem explains missing districts in certain other states, too. (In some cases, there was missing data for smaller counties, but I made a judgment call that the counties in question were small enough that they wouldn't affect the overall percentage much, so they're included.) In Indiana, we're still missing data for Allen and Elkhart Counties, so that rules out IN-02, IN-03, and IN-06. (I already did the 8th in the first wave.) The partial totals for the left-out districts are still available in the Indiana database (the same is true for OH, PA, and TX as well), if you click the link. They may well be very close to the actual percentages, but there's just no way of knowing.
In Ohio, large counties we're missing include Mahoning, Trumbull, and Medina, so we're short OH-06, OH-13, OH-14, OH-16, and OH-17. Pennsylvania is missing Montgomery, Butler, and Cumberland Counties among others, so there we're also missing the PA-03, PA-04, PA-05, PA-06, PA-07, PA-09, PA-10, PA-13, and PA-19. (MontCo also occupies a tiny bit of PA-02, PA-08, and PA-15, but it's such a small percentage of those districts I decided to let it slide.)
In Texas, Cameron County is missing, so that leaves out TX-15 and TX-27. (I also did the 1st and 31st in the first wave.) Finally, there's the matter of New York, where only a few counties bother to report by precinct. Luckily, two of them are Westchester and Rockland, so at least we can do NY-18 and NY-19 there.
There was also one missing county in Oregon, which kept me from including OR-02 and OR-04 in the first wave. I found enough information about Josephine County to decide how to allocate its votes (66.8% of the county's voters voted for a candidate in the OR-02 congressional race, while 33.1% voted in OR-04, so I just applied those percentages to the presidential race).
In Georgia, as with many of the other southern states, early votes aren't broken down, so what jeffmd did, as before, was to use both 'hard' and 'soft' totals, where soft totals included early votes allocated proportionately. I'm including the soft totals (otherwise, we wouldn't have even won GA-02 and GA-12, where victory clearly depended heavily on black turnout).
So what are some of the highlights in this data set? Check out some of the traditionally Republican districts in California (where in many, not coincidentally, we came very close to surprising long-term incumbents) like CA-03, CA-26, CA-44, and even GOP strongholds like CA-25 and CA-48: all won by Obama.
Some of the biggest gains were in Indiana, especially in the Indianapolis area, where both the city itself (IN-07) and its right-wing suburbs (IN-05) zoomed to the left. Amazing what you can accomplish when you actually try to contest a formerly uncontested state.
One area where the GOP might take heart is western Pennsylvania, where there's apparently the one district in the nation that flipped from going for Kerry to going narrowly for McCain: John Murtha's PA-12. Also, the Philly burbs didn't move as much as one might expect (the needle barely budged in PA-08 in Bucks County); where the biggest progress occurred in Pennsylvania was out in places like Lancaster and Harrisburg (see PA-16 and PA-17).
Texas is a very complicated tapestry: in many rural parts of the state, there was no real improvement from 2004, despite the loss of the favorite son effect. For example, expect TX-13 to replace UT-03 as the district with the worst PVI once they recalculate. And look at TX-08, where both growing right-wing exurbs and declining Dem fortunes in the Beaumont area were a double-whammy. Contrast that, though, with not just hugely improved percentages in the minority districts, but also a lot of progress in the suburban districts that we've discussed a lot recently where the minority growth is accelerating: TX-10, TX-22 (where the growth wasn't enough to save Nick Lampson, sadly), TX-32, and especially TX-24 in the area around DFW airport.
And, as always, if more results trickle into the master database, I'll be sure and post them to the front page. So keep on number-crunching!
I got most of the results for California's congressional districts right here. Breakdowns by county I have entered into a Google Spreadsheet. I'd greatly appreciate if anybody could find the results for the districts which I couldn't find any data.
Last week we released our first wave of results, for over 100 congressional districts. Today, as promised, here's our second wave, with the results for another 95 districts.
Despite the huge avalanche of data, we're still only halfway done. Please take a look at our master database to see what states remain. Even if you don't have time to tabulate data yourself but if you've sniffed out some precinct-level data sources anywhere, please let us know in the database! We need you, to put the "crowd" in "crowdsourcing." (A permalink to all our results so far is available here.)
This round was even more fun than the previous batch, because it involves a number of states where the turnaround was huge (either because of the favorite son effect, as in Illinois, or because the Obama campaign actually decided to compete there for once, like Nevada and North Carolina). Want to see some truly staggering progress? Check out IL-08, once the core of Chicago's deep red suburbs (and home to Rep. Phil Crane), and even the site of a 56-44 edge for Bush in 2004. This year? Obama won 57-42.
As we get into the more complicated states here (the last wave picked all the lowest hanging fruit), there are some caveats to be mindful of, which may affect the data reliability a small amount. In Florida, for instance, we're missing precinct-level data for one county, which affects two districts (the 3rd and 6th); unfortunately, it's a pretty big county (Alachua, home of Gainesville and Univ. of Florida). In Illinois, all districts are affected by the curse of split precincts, which don't seem to make much of a big difference, while some of the downstate districts are affected by some missing precinct-level data from some smaller counties; because of their small size, it doesn't seem to affect the district's overall percentages much, though.
In North Carolina, our spreadsheet whiz there broke the totals down into 'hard' and 'soft' totals, with 'hard' reflecting only known numbers and 'soft' allocating early votes and absentees proportionately (which apparently just sit in an indistinguished lump in NC). I chose to proceed using the 'soft' totals, which are much larger, but a stickler might choose to focus on 'hard' totals instead. South Carolina also has a high number of 'indivisibles' in its spreadsheet, which appear to be a lot of split precincts. The high number of indivisibles appears to exert some pressure on some of the percentages in South Carolina, perhaps in SC-01, where there appears to be little leftward movement since 2004.
So, take the numbers with a grain of salt, and certainly don't expect these numbers to be a 100% match for Polidata's numbers, forthcoming this spring. And of course, please contribute to this project however you see fit!
We're pleased to announce that we've created a perma-post for the 2008 presidential results by congressional district. (We've included the 2004 and 2000 results as well.) You can use this link, and you'll also find a permalink in the "Resources" box on the right-hand sidebar of the site.
As our crowdsourcing project continues, we'll keep adding new numbers. In fact, you should expect another round of numbers this week.
Some other site news stuff: Thank you to everyone who has taken the Blogads reader survey so far! We've gathered about 75 responses so far, which is terrific. We'd love to hit at least 100. This may not be a scientific study, but nonetheless, everyone on this site knows the value of a big N! So please take the survey today. Once it's complete, I promise that we'll share some of the results.
According to the Census Bureau, thirty of Florida's 67 counties are split between multiple Congressional districts. (The remainder are all located entirely within single CDs.) Unfortunately for us, the state of Florida does not publish results by CD, which means that calculating pres-by-CD for FL requires digging down into precinct-level results.
Frustratingly, the Florida Division of elections doesn't publish that information either. However, there is good news: Florida's county-level election departments almost universally do. I've gone through all thirty split-county websites to try to track down this information. I was successful in all but two counties - let us know if you have any luck with Alachua and Jefferson. (UPDATE: Jefferson County sent me a spreadsheet in response to my email request to them, which you can find here.)
Anyhow, I've posted links for every set of precinct data I could find in this new spreadsheet. The linked files are in a variety of formats. Some are nicely parsed Excel or CSV files, some are clean TXT or HTML files, and some are thorny PDFs. (One non-split county, Glades, even posted a handwritten chart!) The most sophisticated sites will give you a choice of formats.
No matter how you slice it (and at the precinct level, you're slicing it awful thin), there's a lot of data to work with. Miami-Dade's file, for instance, is over 2,700 pages. As Crisitunity observed, we've picked off most of the low-hanging fruit already, so bigger projects are what remains. But as Crisitunity also noted, with a million nerds pecking away at a million keyboards, we can accomplish just about anything we set our nerdly minds to.
So if you want to start tackling the big prize of Florida, check out the spreadsheet and have at it! And if you have any thoughts or strategies for how best to pursue this project, please share them in comments.
Enough things have happened (such as states certifying their results) that we're ready to roll out our first wave of results from Swing State Project's big crowdsourcing project of compiling presidential results by congressional district. Usually, knowing presidential results by CD requires waiting for Polidata to compile this data and make it public in March... but the power of an infinite number of nerds typing on an infinite number of spreadsheets makes it possible for us to short-circuit the process. (There's still tons of stuff left for enterprising nerds to do, especially if you have access to precinct-level data. Check our database in progress.)
Without further Apu, here's the first wave, representing nearly one quarter of all congressional districts. Explanation of many of the technicalities follows below the chart (and a simple spreadsheet of just the 2008 numbers is available here):
The easy ones to do were the at-large states, and states where the SoS has already reported by congressional district (Connecticut, Idaho, Kentucky, Minnesota, and Virginia). Also easy were states where district lines precisely follow county lines (Arkansas, Iowa, and West Virginia).
We also have a number of excellent spreadsheets in our portfolio where people were able to locate precinct, ward, or town data. (A huge thank you to everyone who has contributed, and one more reminder that there are still many more states to do, although they get progressively harder from here on out.) These include Louisiana, Maine, Massachusetts, Michigan, Mississippi, New Hampshire, New Mexico, Oregon, Rhode Island, and Wisconsin.
You may notice that not all the districts from MA, MI, and OR are included. That's because in each of these states, there's one pesky jursidiction that hasn't reported at the precinct level yet: Fall River in Massachusetts, Wayne County in Michigan, and Josephine County in Oregon. If you find this data anywhere, please let us know! (A few other minor requests for our anonymous spreadsheet wizards: if the persons who did MA and MS have "other" data, could you add those to the databases? And whoever did WI, could you provide the "Wisconsin long" form that shows precinct-level data in split counties? Thanks in advance!)
You're probably wondering about those ones where there's a total and then a range of numbers in parentheses. These districts (AL-01, CO-03, CO-05, IN-08, MO-08, TX-01, and TX-31) are ones where there were county splits but I felt confident proceeding even without precinct data, because there was only one split county and it represented such a small percent of the total that even if I allocated the votes within the county completely wrong it still wouldn't affect the total percentages by more than a fraction of one percent. In these cases, I'm presenting both range values (of the maximum and minimum possible) and a point estimate (calculated by allocating half of those counties' votes for each candidate to the district in question, and half to the other district).
As we get more states done, we'll roll more of them out. We're expecting California and Nebraska to report by CD soon (which will give us another 56 CDs right there), but for almost all the other states, we're missing precinct-level data. If you like this resource and have access to useful information, but don't have the time or stamina to spreadsheet it all, please just let us know in the comments or the master database, and I'm sure someone else will pounce on it.
Right now we're in a bit of a holding pattern with the presidential results by CD project. Most states haven't yet certified final vote tallies, and some haven't even finished counting. But the good news is that we have some preliminary numbers for the following states:
Connecticut
Iowa
Kentucky
Minnesota
New Hampshire
Rhode Island
Virginia
West Virginia
You can find these by clicking on the links in the "Calculations" column (column E) in the collaborative spreadsheet.
On the flipside, we still need links to official data sources for the following states:
Alabama
Florida
Indiana
Louisiana
Mississippi
New Jersey
Oklahoma
South Carolina
Tennessee
Utah
If you know the proper links to official results for any of these states, please enter them as a TinyURL on the spreadsheet. Even if official 2008 results haven't yet been released (and as I say, in most states they have not), links to where you expect the official data to show up at some point would be greatly appreciated.
As always, please share any thoughts about this project in comments. Also, a couple of helpful resources from the US Census Bureau:
Congressional District Data Tables: Contains tables of information about CDs broken up in a variety of ways, such as which counties are in which CD and whether or not they are split (example: Alabama).
Just a quick update on our project to crowdsource presidential results by congressional district. The following states are all in need of links to official data sources and "plans of attack":
Alabama
Colorado
Florida
Illinois
Indiana
Louisiana
Mississippi
New Jersey
Oklahoma
South Carolina
Tennessee
Texas
Utah
Washington
The data sources don't have to have actual numbers available yet - in fact, since most states won't certify their vote totals for a while yet, that is only to be expected. And as for "plan of attack," I'm talking about how we expect to crunch the numbers - do the states publish results by CD? Do they divide up (relatively) neatly by towns or counties? Or will we have to rely on raw precinct data?
