Yes, I'm still at it! On Sunday I sent off a data set for 2008 New Jersey presidential figures to replace the one currently on Dave's Redistricting App. I also sent off a file with presidential results for Rhode Island. So, while the usual caveat applies that they aren't yet uploaded to DRA and won't be until Dave has the time to do it, I want to go ahead and post a few remarks while it's all still fresh in my mind.
New Jersey: My long New Jersey ordeal is over! New Jersey has an unusual hybrid election system where mapping and reporting duties are shared by counties and municipalities. The degree to which counties or municipalities determine precinct boundaries also varies between the counties, as do the guidelines that are used, and the reliability of their VTD reporting to the Census Bureau. There are also differences in which county or municipal departments do what (County Clerk v. Board of Elections; Municipal Registrar v. Planning Dept., etc).
Might I add, I'm astounded at the degree of pointless bureaucratic redundancy throughout New Jersey, which I can tell extends to virtually every level of government beyond just elections and doesn't seem to have any correlation with whether an area leans Republican or Democratic. I just thought I'd throw that out there! The short of it is that, for my purposes, New Jersey was a nightmare to properly map for DRA.
So, on that note, here are the fixes to expect in the new data set. I don't know to what degree this actually affects the 2008 presidential results on the congressional district level. I expect that it will have a significant effect on a number of legislative districts. Also, as is usually the case, I didn't bother with the 'average' election data.
1) The current data set doesn't include the "Other" votes for minor candidates. I added them.
2) The current data set includes absentee votes for some counties and not for others. I've included all the votes except for a handful of countywide federal ballots in Morris and Sussex counties (428 and 68 total votes respectively).
3) I've corrected minor reporting errors in three counties (Atlantic, Essex, Monmouth) so there will be slight differences in my countywide total from those on the NJ SOS website.
4) The current data set has zero votes for "East" city wards. This is especially noticeable in Newark. I think this was a GeoID translation error having to do with the letter "E" being in the GeoID. In any case, I've obviously added the vote figures for those precincts.
5) There are 64 municipalities in the current dataset, including several large cities, where the total Obama/McCain votes for the town or ward are just distributed to VTDs by Voting Age Population. This usually involved towns or wards with precinct splits or mergers not reflected in the VTD map. Whatever the case, I have allocated the votes to their proper VTD. I only do distributions between neighboring precincts where necessary to account for shapefile errors, as detailed below.
6) I didn't count them up, but there are a number of municipalities scattered across the state with almost no votes assigned to them in the current data set. I'm not sure why that happened. For instance, the current data gives Lakewood Township in Ocean County just 72 Obama votes and 725 McCain votes. The correct figure is 8242 Obama votes, 19173 McCain votes, and 144 Other votes. I thought it might be one precinct getting distributed to the entire township but the figures don't match for that (e.g., there is no individual Lakewood precinct with 72 Obama & 725 McCain votes). In any event, this is obviously fixed in the replacement data set.
7) There are many townships where the number of precincts stayed the same between 2000 and 2008 but where precincts were renumbered (e.g., Precinct 12 from 2000 had become Precinct 9 by 2008, etc.). For a number of townships/wards the census VTD numbering was not updated. In other words, the VTDs are numbered as the precincts were in the 1990s, not in 2008. The data set currently on DRA assigns precinct votes in these cases based on the old incorrect numbering, especially in a number of city wards. My replacement data set assigns precinct votes to the correct geographic census VTD.
8) Finally, there are shapefile errors all over the state. One type was easy to identify: where the VTDs are the 2000 precincts, not the redrawn precincts used in 2008. I flagged these early on and got 2008 maps to make the necessary adjustments. The other type jumped out when I thought I was done in January. I did my usual final check of turnout by VAP and flagged several dozen more shapefile errors all over the state. I had to go through another round of collecting maps to fix those, which took another couple months (NJ clerks range from very friendly and helpful to downright hateful).
Some of these errors are very subtle, involving a boundary street that votes entirely in one district but that is split between census VTDs (putting one side of the street in the wrong VTD). This can be trivial or it can be a major discrepancy, in some cases shifting over 1000 voting age residents to the wrong VTD. The short of it is that, while I surely missed some of these where the effect on the vote counts was trivial, I've fixed those I was able to identify, and certainly all the ones that make a meaningful difference at the legislative or congressional district level.
So, there you have it. I presume that Dave will indicate on the DRA homepage when the new data is uploaded. If nothing else, NJ should move from the list of states done by others to the list of states done by me.
See after the jump for an update on where things stand on the remaining states.
Rhode Island: As noted, I sent a file of 2008 presidential data for Rhode Island. I got 2008 precinct maps for every township in Rhode Island and distributed the precinct votes by VAP on the census block level. Very tedious!
Montana: 18 Montana counties did not participate in the census VTD program, so I had to get precinct maps from the counties. I've collected all of them and started to gradually do a precinct-to-census conversion of the 2008 presidential results, using the same method as with Rhode Island. Again, this is a very tedious and time-consuming process. For now, I'm too busy to really work on it much. I don't expect to really get back to it until May.
As an idle aside, someone in DC was coincidentally collecting Montana precinct maps a couple weeks before I did. From what I gather, he or she failed to get maps for several counties. If that person is reading this, here's a tip: When all else fails, get the clerk to take a photo of the map with a cell phone and send it to you as a text message attachment.
Alaska: Alaska participated in the VTD program and I'm hoping Dave might yet get a VTD map up on DRA. I want to do 2008 president and senate and I can do it quite swiftly if I have a VTD map. If not, then it will be a pain in the ass like the other block group states, and again, I don't expect to do it before May. On the bright side, if I have to do the conversion I can superimpose VTD and block group shapefiles on a GIS viewer, which will make it go a lot quicker than RI and MT counties without VTD shapefiles.
Oregon: Ah, Oregon... Such a lovely state with such unhelpful election procedures so far as my purposes.. Only Multnomah County participated in the census VTD program so I need precinct maps for all the other counties. Moreover, the state doesn't collect results by precinct, just total certified figures, so I need the precinct breakdown from every county. I set aside my precinct data/map collection effort when I realized that I wasn't done with New Jersey in January, but I plan to get back to it soon. These are the counties that I still need from Oregon.
2008 Results by Precinct: Baker, Malheur, Wallowa, Wasco
2008 Precinct Maps: Baker, Gilliam, Grant, Klamath, Lake, Lincoln, Malheur, Morrow, Union, Wallowa, Wheeler, Yamhill
If anyone reading this happens to have any of that, then please message me! Meanwhile, I'll start contacting these counties again now that I finally have New Jersey out of the way. If I had to guess, I won't have the time to do Oregon until June in any event, which should make it the last state to get election data on DRA. Yay!
Miscellanea: I do still plan to look into getting "Other" votes for Florida and a more complete data set for California. Someday..