Google Ads


Site Stats

Crowdsourcing Project: Presidential Results by CD

by: DavidNYC

Thu Nov 06, 2008 at 8:42 PM EST


(Bumped.)

A favorite reference for election junkies like those of us who inhabit this site is, of course, presidential results by congressional district. Unfortunately, most states don't publish this data,  but it's reliably churned out by a firm called Polidata every four years. It's a difficult task, though, because it usually involves crunching data on a precinct level (and also figuring out what the hell to do with absentee ballots), so Polidata typically releases its findings some time in March after a presidential races.

But the good news is that, working together, we can come up with some preliminary data for at least some states. There are at least three types of states where we can get this data relatively easily:

1) States with just one at-large district (duh), like Montana.

2) States which actually publish presidential results by CD, such as Virginia.

3) States where district lines follow county lines or, in New England, township lines (or at least follow them closely), like Iowa.

Where we need help (at this stage) is in figuring out which states fall into the second and third categories. I know California also releases results by CD, but I believe a few other states do as well. And Arkansas and West Virginia follow county lines, but some other states probably do, too.

Also, let's use this thread for brainstorming about other ways we might try to figure out presidential results by CD (short of acquiring detailed precinct-by-precinct data). Please share your thoughts in comments.

UPDATE: Jeffmd observes that some states offer easy-to-use precinct data, so where available, that might be an option as well. If you're aware of any states which provide this information, please let us know that as well.

UPDATE 2: I've created a public Google Docs spreadsheet that we can use to compile a list of data sources. Please feel free to input any helpful information you're aware of. Note that we're not looking for the actual numbers just yet - rather, we want to know where we can find the numbers (and what format - ie, CD, county, precinct, etc. - we should be looking for). And if you are adding a URL, please use TinyURL.

DavidNYC :: Crowdsourcing Project: Presidential Results by CD
Tags: , , (All Tags)
Print Friendly View Send As Email

States by type:
County Lines (completely): Arkansas, Iowa, West Virginia.

I know Maryland, North Carolina, and New York published by CD for the primary, but not sure about now. Maine and Nebraska publish by CD out of necessity.

Minnesota, Wisconsin, and Ohio publish very easy-to-use precinct data (I'd be fine with looking through those myself, this would be a few hours per state, tops).

In Illinois, Chicago, Suburban Cook, and Lake publish precinct data as well, so you can get full data for CDs 1, 3, 4, 5, 7, 9, and 10. Again, maybe half an hour per district.

New Hampshire follows county lines, roughly, and since no townships are split, can also be calculated without too much difficulty.  


Just as an addendum...
I think building a detailed district-by-district collection of the presidential vote would be a powerful tool, an open source Almanac of American politics, almost.

This is work intensive, but I think it'd be a good idea and easily accomplished through crowdsourcing.


[ Parent ]
CQPolitics
has such a thing in their election map, with results from 2004 and 2000, as well as congressional results from 2000-2006 all handy, just open the map and click on the districts you want to know about.  

Call no man happy until he is dead-Aeschylus

[ Parent ]
I'm talking specific numbers.
As in vote counts, which CQ doesn't have.

[ Parent ]
they give numbers for congressional
races, and percentages in the presidential races are good enough for me.  

Call no man happy until he is dead-Aeschylus

[ Parent ]
Polidata makes the raw CD numbers available
See here:

http://www.polidata.us/books/

Click on any state, then scroll down to "PRESIDENTIAL & CONGRESSIONAL ELECTION RESULTS BY DISTRICT" and select any of the PDFs linked there.


[ Parent ]
I agree
I think we can accomplish a lot. When the nerds at SSP put our minds together, we can do anything!

[ Parent ]
Minnesota by CD
I've already done Minnesota based on the data posted at their SOS web site as of yesterday.  I've put together voting for President and senate by precinct and included CD and county data for each precinct.  Now that I have the data imported, it's a trivial command to create summaries at the CD level.  I created a script in Stata to automate the process for updating.

[ Parent ]
I was just going to PivotTable that thing...
since it nicely lists county and CD for each precinct, as soon as it was certified.

But Stata works too. :)


[ Parent ]
hmm
I was actually looking at New Jersey, and the majority of the CDs seem to entirely contain certain counties, with few exceptions. While not perfect, you can almost guess and calculate the PVI there.

New Jersey...
when it does upload results (several months after the election), also publishes township-by-township results - further reducing the inaccuracy, especially given the "artful" way the 6th, 7th, and 12th are drawn.

[ Parent ]
NJ
None of the 13 CDs contain just etire conties and many counties are split into muliple CDs. Data by municipality is better as few districts seem to take only parts of towns but both Newark and Trenton, at a minimum, are split.

My own Essex County, for example, is split into at least three CDs with the largest chunk being in the 10th but the 8th and the 11th get multiple towns.  Bergen is in the 5th and 9th.  Passaic in the 8th and 5th. And so on.  Hudon's splt.  Sussex, even tiny Sussex, is split.  Warren is whole.  Some of the south Jersey counties are whole.  But the east-west nature of many districts sacrews that up a lot.  The two amjority minority districts in north Jersey (10,13) also add to the mix with the "hispanic district" crossing multiple lines.

The lines were drawn to protect the existing incumbents with the big gain over the previous map being making Rush Holt safer.  


[ Parent ]
Southern States
I could be wrong, but I think the districts in KY, TN and GA are fairly contiguous with just a few split counties.

28, male, NC-13 formerly NC-01, 04, 05, 07, 11

Probably not Georgia
The problem you're going to have is that most of the counties that are split (ie. Fulton, Douglas, Cobb, Muscogee, Chatham, Richmond, Gwinnett, Carroll, DeKalb) are the bigger counties, so there are more precincts to sort out.

Follow the elections in Georgia at the 2010 Georgia Race Tracker.

[ Parent ]
Virginia does President by district
https://www.voterinfo.sbe.virg...

I hope that link works!

28, male, NC-13 formerly NC-01, 04, 05, 07, 11


one more thing
NC offers precinct data:

http://results.enr.clarityelec...

Click on "Select County" at the top
Once you've selected a county, click "Reports" and drop-down list of precincts is available.

Problem is, we don't break out our early votes by precinct, so you might only get absentee-by-mail and election day totals.

28, male, NC-13 formerly NC-01, 04, 05, 07, 11


Georgia has precinct level data
But it won't be available until after the vote is certified.

Follow the elections in Georgia at the 2010 Georgia Race Tracker.

wikipedia.org
often has sources for presidential numbers, in every california district over the past four elections actually.

Call no man happy until he is dead-Aeschylus

Can do it in Texas
The SoS of Texas apparently does not report results by CD. But their county results break out the subtotals for each of the CDs that include a part of the county. And many of the districts are comprised of many whole counties and fractions of a couple more. An old-fashioned adding machine could do it. But you'd need patience -- one district has 44 counties to add up, 42 whole, two in part! Even Ciro's district has 20 counties.

one good way to find out which
precincts and in which CDs would be to pull it off of votebuilder.com

Good project
I've been trying to build the basis for a study of who overperformed and underperformed relative to Obama and each other amongst Senate and congressional candidates. I also want to check 2004, 2006 and 2008 performances against each other, to see where the presidential vote buggered us in Congressional races this time.

Once I'm home and have the time to waste, I'll see what I can dig up here.


Are we including media sources or just official ones?
The Boston Globe has the complete Massachusetts town by town data for both 2004 and 2008, however, it may be more likely to contain errors than official returns, which at least on the SoC website, do not break down results in any useful way.

28, Unenrolled, MA-08

New PVI's for Virginia
VA-01 R+7
VA-02 R+5
VA-03 D+20
VA-04 R+4
VA-05 R+5
VA-06 R+12
VA-07 R+8
VA-08 D+16
VA-09 R+11
VA-10 R+2
VA-11 D+3

The Pre-2008 PVI's for VA
VA-01 R+9
VA-02 R+6
VA-03 D+18
VA-04 R+5
VA-05 R+6
VA-06 R+11
VA-07 R+11
VA-08 D+14
VA-09 R+7
VA-10 R+5
VA-11 R+1    

[ Parent ]
yep
When the results get certified I will plug the Pres by cd numbers.

2010 Race Tracker Wiki

I calculated some results and have more here
FL-24(Kosmas, D):  McCain 50%, Obama 49%.  This is good news for Kosmas in a district that was drawn by Feeney himself to be "safe GOP".  Obama came within 3,000 votes here.

NC-09(Myrick, R):  McCain 54%, Obama 45%.  This one shows an amazing trend towards Democrats. Bush won here 63%-36% twice.

GA-13(Scott, D):  Obama 70%, McCain 29%.  Why the did the DCCC waste money here?

MI-01(Stupak, D):  Obama 50%, McCain 48%.  

KS-03(Moore, D):  Obama 51%, McCain 48%.  Moore should be safe here from now on.

MO-09(Lukkeymer, R):  McCain 55%, Obama 44%.  A pretty tough district for any Democrat these days.

LA-04(Open):  McCain 59%, Obama 40%.  We should really try to win this one to compensate for the loss of LA-06.

LA-06(Cassidy, R):  McCain 57%, Obama 41%.  Michael Jackson should be thrown right out on his ass by the Democrats in the state legislature for the crap he pulled that cost us this seat.  He should be made to wish that he had never done what he did.

SC-05(Spratt, D):  McCain 53%, Obama 45%.  Pretty good performance here for Obama.

TX-23(Rodriguez, D):  Obama 52%, McCain 47%.  Nice swing here.  Rodriquez should be fine here in the future.

AL-02(Bright, D):  McCain 62%, Obama 37%.  One of my favorite results from election night.

IN-01(Viclosky, D):  Obama 61%, McCain 38%.  Here lies Obama's statewide victory margin.

IN-06(Pence, R):  McCain 52%, Obama 47%.  We should consider targetting this one next time.  That is an awfully small McCain margin for such a right wing Republican leader.  Imagine if Henry Waxman held a district where Kerry won by only 52%-47% in 2004.  



I think it would be really excellent
If you could post your stuff in a Google spreadsheet and "show your work" (as your fifth-grade teacher might say) so that people can follow along with your methodology.

[ Parent ]
Nevada
Nevada should be an easy state because it is a lot like a single district state. All you need to worry about is Clark County, which is split by all three districts in the state.  

I'm going to be doing NM
all three districts sometime this week. I already have county-level data (look for some badass maps on Monday), and most of it is pretty contiguous. But five counties are mixed, and Bernalillo County (the largest in the state) has all three counties. So that will suck, especially with 400+ precincts.

But I'll figure out a way.

New Mexico politics from the local perspective.


MI-6
Barack Obama won Michigan's 6th District (Southwest Corner, Kalamzoo) by a margin of 54.8 to 43.5. The raw vote total is 160,393 to 127,432.  This is without counting the votes from Allegan county, which McCain won by only 5,000 and is split between the 2nd District.  Adding Allegan County is unlikely to have an significant impact. George Bush beat John Kerry by a margin of 53-46 in 2004.  This is a swing of nearly 9 points for the Democrat.  The PVI for this election would be Approx. D+2.2.  I don't know the exact 2004 PVI, but it's approximately R+2, so the new averaged PVI would be D+0 to D+1.

16, Male, MI-01


Copyright 2003-2010 Swing State Project LLC

Primary Sponsor

You're not running for second place. Is your website? See why Campaign Engine is ranked #1 in software and support among Progressive-only Internet firms. http://www.mediamezcla.com/

Menu

Make a New Account

Username:

Password:



Forget your username or password?


About the Site

SSP Resources

Blogroll

Powered by: SoapBlox