I just took a look at some of the vote tallies in New Mexico, another hotly contested state. Guess what I found? Dixiecrats! Or maybe you'd call them New Mexicrats out there.
There is a pro-Republican vote swing in small counties in New Mexico that's very similar to those so-called Dixiecrat small counties in Florida. According to voter registrations these counties are heavily Democratic, but in 2004 and 2000 they voted overwhelmingly for Bush. Even so, their margin for Bush increased dramatically this year.
Here are some examples that compare the proportion of voters registered Democrat vs. Republican to the percentages won by Kerry and Bush this year. (For simplicity, I eliminated "No Party" and Green Party registrations when calculating relative percentages for DEM and REP registered voters)
|County||DEM||REP||Kerry (DEM)||Bush (REP)||vs. Expected||Tech|
|De Baca||987||72%||381||28%||280||28%||705||71%||-44%||Optical Scan|
De Baca county, for example, shows a complete reversal. 72% of major party registrants there are Democrats, but only 28% of the voters in De Baca went for Kerry. Kerry's performance in DeBaca is 44% less than expected. Of course De Baca county went for Bush in 2000 as well, but only by 62% to 36%.
Why are there so many Democrats voting Republican in these counties? The southern Dixiecrat explanation doesn't work in the wide open Southwest.
Just like in those small Florida counties, all the New Mexico counties above use ES&S Optical Scan voting machines.
Just like in Florida, Kerry did not exceed the expected percentage in any county in New Mexico. Not one.
I also compared all Optical Scan counties against the E-Voting counties. The Optical Scan counties only account for about 10% of voters, so I'm not sure how valid these trends are. But here goes:
- In Op Scan counties, Kerry performed 21% worse than expectations on average. In E-Voting counties he only did 10% worse.
- Turnout in Op Scan counties was 8% higher than the rest of the state, on average. (71% vs 63%) This could be another indicator of vote inflation in the small counties.
There are some other anomalies with the New Mexico counts too. Many of these might be explained by the fact that vote counting in New Mexico is still underway, and new data is posted to the website each day.
- The sum of the vote totals in each county don't add up to the total number of ballots counted statewide. Perhaps the absentee ballots are not yet allocated to the counties or something?
- None of the county counts sum up to the "Number of Voters" posted for the county. I assume this generally means that provisional or absentee ballots haven't been fully counted yet. However in a couple cases the counts are higher than the "Number of Voters". In Valencia County, for example, the ballots total 25,795 but the Number of Voters posted is 25,435.
- Turnout was strangely low (56% or less) in three counties. The statewide turnout was 64%. All three counties with strangely low turnout were heavily Democratic.
County Turnout % DEM vs REP McKinley 49% 79% Cibola 55% 76% San Miguel 56% 84%
I've posted a county-by-county summary of these spreadsheet results here.
I think we're less likely to see pro-Bush fraud in New Mexico because the Governaor and the Secretary of State there are both Democrats. But still, anomalies like these cry out for good explanations. Anybody want to propose some?
I've gotten some emails asking why we should bother analyzing the data like this when Kerry has already conceded. As Kevin Drum has pointed out, we need to keep the pressure on so those pollsters or government officials who have the power to explain results like these will take some time to do it. Also, we need to gather data points, discover trends, and help folks like Ralph Nader, David Cobb, Bev Harris figure out which battles to fight.