I finally have an Arkansas data set ready for Dave's Redistricting App. To put it mildly, sorting out the 2008 election data for the state was a nightmare. In any event, I'll be forwarding it to Dave after I take a last glance over it tonight. I don't know when it'll actually get uploaded, but here are some relevant notes while it's fresh in my mind.
[Edit] The 2010 precinct data is an average of the votes for Lieutenant Governor, Secretary of State, and Commissioner of State Lands. These were the three closest statewide races and all three went narrowly Republican.[Edit]
A few relevant notes.
1) There is a scattering of voting districts where the 2008 figures add up to more than the 2010 voting age population. These are all in the Delta and appear attributable to two reasons. First, the Delta has been losing population, up to a quarter in some areas, with much of the loss toward the end of the last decade. Second, I had to do early vote distributions in a number of counties and it's clear that in several Delta counties these were heavily concentrated in urban areas. In any case, I only decided to 'tweak' the figures for two counties (noted below) where the deviation was systematic. In general, I avoid making arbitrary adjustments to my standard methodology. The number of votes affected are ultimately trivial in the broad scheme of things.
2) Pine Bluff: I used the list of precincts and polling locations to do the initial vote assignment. However, in many cases where neighboring precincts used the same poll location there would be one precinct with way too many votes and the other with way too few. I don't know if this is a reporting issue or a mapping issue (it seems to be a combination). I redistributed these precincts by VAP. In a few cases, this involved a heavily black and a heavily white precinct. I further adjusted these figures to match the partisan breakdowns of nearby precincts with the same racial composition. I did the exact same distributions for both 2008 and 2010.
Precinct 212 was double counted by the state (but not the county) in 2008. I excluded the duplicate votes. That's why my statewide total is different from the official total.
3) Russellville: In 2008 all of Russellville voted at Tucker Coliseum, so I just distributed those votes to the 32 relevant voting districts by VAP. In 2010 the precinct counts were reported separately but the precincts did not remotely match the census voting districts. I got a precinct map and distributed votes by VAP on the block level.
4) West Memphis & Helena: For the 2008 election, Crittenden County (W. Memphis) and Phillips County (Helena) did not assign early votes by precinct. When I did my standard distribution, the rural precincts ended up with far too many votes. In short, I reweighted the early vote distribution to put most of the early votes in the urban/suburban voting districts.
5) Hot Springs: The election precincts and census voting districts also don't match up very well at all in Garland County. I had to use a street file to find the sub-precinct boundaries and distribute the votes accordingly based on VAP.
And a general note: Arkansas has made an impressive move to standardize its election reporting, which is now quite orderly. Sadly this was not the case in 2008. Some counties distributed all votes by precinct, some did absentees but not early votes, some did early votes but not absentees, some did neither. In 2010, almost all counties did a full precinct distribution. If in doubt, go with 2010, which is probably the more useful data set in any event, since those were actually close elections and the precincts more consistently matched the census VTDs (meaning far fewer adjustments were required on my part).
Finally, I'll be forwarding minor fixes for New Hampshire (Hampton and Hampton Falls are currently flipped) and for Oklahoma (the boundaries for precincts 201 & 202 in Logan County are wrong on the census maps, so I needed to do a distribution).
I also did a data set for 2012 Puerto Rico governor. I don't know how it'll be labelled on DRA, but Padilla (Popular Democratic Party) will be blue and Fortuño (New Progressive Party) will be red, because Padilla is a registered Democrat and Fortuño is a registered Republican.