I’m still reasonably certain that the selection of Alvin Greene over an almost equally unknown former state legislator in the South Carolina Senate Democratic primary can be explained by a combination of factors and not a conspiracy. However, I’m becoming less and less certain about that when I read more about the raw totals in the race. Tom Schaller at 538 has been all over this angle, both here and here. The irregularities are numerous:

• Unusual turnout rates. The Vic Rawl campaign asserts that turnout ended up much higher than their expectations.

• Non-white registrant share. There seems to be no correlation between Alvin Greene’s share of the vote in white and non-white counties (although, if the two were unknown, that may stand to reason). However, at the precinct level, there may be a small correlation.

• Absentee versus Election Day ballots. This is a bigger concern. Politico reported that Vic Rawl did much better in the absentee vote, whereas Greene crushed him on Election Day. In one case, in Lancaster County, Rawl won the absentee vote 84-16, but lost the Election Day vote 58-41. Similarly massive spreads occurred all over the state. Overall, Rawl did 11 points better among absentee voters than among Election Day voters, a spread unlike any other race in the primary election.

• More ballots than voters. In lots of precincts, the total number of ballots for Greene and Rawl exceed the number of voters in that precinct. The same thing appears to be true in a few precincts on the Republican side. Sometimes bad data like this comes out on the first count, only to be smoothed over in a later canvas. But this data looks real bad.

Benford’s Law. The Rawl campaign has released the findings of a forensic analysis on the election data, and that’s where this gets wild:

Dr. Mebane performed second-digit Benford’s law tests on the precinct returns from the Senate race. The test compares the second digit of actual precinct vote totals to a known numeric distribution of data that results from election returns collected under normal conditions. If votes are added or subtracted from a candidate’s total, possibly due to error or fraud, Mebane’s test will detect a deviation from this distribution.

Results from Mebane’s test showed that Rawl’s Election Day vote totals depart from the expected distribution at 90% confidence. In other words, the observed vote pattern for Rawl could be expected to occur only about 10% of the time by chance. “The results may reflect corrupted vote counts, but they may also reflect the way turnout in the election covaried with the geographic distribution of the candidates’ support,” Mebane said.

This at least suggests the possibility of precinct-level tampering with the vote.

Why, you may ask, would such tampering exist, when Jim DeMint is a virtual lock for re-election? Actually, Vic Rawl was polling at least within striking distance of DeMint previously, whereas Greene has pretty much no chance. And maybe a candidate as embarrassing as Greene (that pornography charge was discovered very quickly after the election, you know) would set Democrats back throughout the ticket. I actually think DeMint would have been fine against Rawl, as well, but when the numbers look this odd, you grab for an explanation.

Perhaps the easiest one involves a pure malfunction. Election blogger Brad Friedman does indicate that South Carolina uses ES&S touch-screen DRE voting machines with no paper trail. Even if the above evidence showed clear signs of fraud, I’m not sure that a recount would show anything but a mirror of the initial Election Day vote.

I think the lesson here is that, even after Bush v. Gore, after the Help America Vote Act, we still don’t have a very secure or stable voting system in many parts of this country. And you run the potential for not only fraud, but from broken machines, to change the course of an election.