From an academic standpoint, this paper provides the first credible statistical evidence I have seen that something is amiss with electronic voting machines, specifically in Florida. But it is not a smoking gun. The relevant result of the paper to the electoral fraud discussion is the following, no more and no less: when controlling for most relevant demographic and political factors, the presence of electronic voting raised the Bush vote. In other words, Bush increased his 2000-2004 lead in counties with electronic voting more than he did in counties without electronic voting. These counties tend to be the most Democratic counties. These results are statistically significant at the p<0.01 level. In other words, we can be over 99 percent sure that the results are not due to chance.
Before both sides of the election fraud debate attack me for taking the other side, let me say that what this diary is intended to be is sort of a translation into layman's terms of the Hout, Mangels, Carlson, and Best paper "The Effect of Electronic Voting Machines on Change in Support for Bush in the 2004 Florida Elections" and a discussion of what the paper does and doesn't say. We've already got enough diaries up on whether or not there's election fraud. If you want to comment on that general topic, go to one of them.
This paper goes about as far as anyone can without either the exit poll data that we've all been hoping to get our hands on or without going out and doing some additional data collection toward making a case that the election results, at least in Florida, are fishy. The authors have accounted for most of the demographic variables that you can take into account. This is a solid piece of social scientific research, utilizing techniques such as ordinary least squares (or linear) regression, which can measure the effect of an independent variable, in this case the presence or absence of electronic voting, on a dependent variable, in this case the percentage change in vote for Bush from 2000-2004. Other independent variables, such as demographic factors, are thrown into the mix to try to discern the effect of electronic voting independent of them.
The big question that has to be asked by any serious social scientist who is looking at this paper is why? Well, one obvious answer is the mechanism that Hout, et. al, mention, which is "ghost voting," i.e., electoral fraud. It's not the only possible answer.
We also have to look at other factors that might covary with the presence of electronic voting. Did the GOP, for instance, simply do better GOTV and registration efforts in the largest counties in Florida (which happen to be the ones that adopted electronic voting machines)? Are there underlying demographic changes in the makeup of the electorates of these counties, as opposed to other counties in Florida, between 2000 and 2004 that would account for the difference in support for Bush?
Social science, unfortunately, is not conducive to the production of smoking guns. Journalists and lawyers do a much better job of this. Because of the limitations of statistical techniques, there's always a caveat. Things do, however, begin to look suspicious when study after study begins to build up. Each study has its own holes, but when you accumulate all of the studies, the chance that the caveats in all of them came to pass is so minuscule as to render the supposed outcome a virtual certainty.
I will take the typical social scientist's way out of this and suggest that more research is needed. However, in the light of this paper and the Freeman paper that was discussed in the blogosphere a couple of days ago, I will say that while I am not yet a convert, things out at the end of the T-distribution are beginning to emit a scent that is somewhat reminiscent of aquatic creatures.
For anyone who's curious, my background: I have passed Ph.D. qualifying exams in political science research methodology but have not completed my doctorate. In fact, I've put completion of the doctorate on hold to go to law school. I also have 12 years of experience in Washington as essentially a professional observer of the U.S. Government.
If you want to read the paper, it's here.