Since the election, Democrats have offered various theories, all of them ludicrous, to attack President Bush’s victory. One of the more recent is a “study” done by some people at Berkeley which found that in certain counties in Florida, votes for President Bush increased faster than they would have predicted, based on 1996 and 2000 results. In those counties, electronic voting had replaced punchcards. The conclusion is obvious: Bush stole Florida! Again! A remarkable claim, given that the President carried Florida by what, five points?
Dafydd ab Hugh, who actually knows something about statistics, dissected the Dems’ latest theory in an email to California Insider:
I have just read through the UC Berkeley paper you linked about e-voting. There is a pretty significant pair of errors in it.
The assumption is made (but never disclosed) that in an honest election, e-voting patterns would mirror those of paper voting. But in fact, there is no way to know this. For example, marginal Democratic voters — and since the error they purport to find is about 2.5%, we are talking about the margins — in the Democratic counties could well be older retirees from New York and more frightened by electronic voting, while Republican voters in those Democratic counties could be younger and actually attracted by e-voting; even younger Democratic voters could have been more moved by 9/11 than older Democrats, and might also be less afraid of e-voting.
The second assumption is that if there is an actual discrepency between the results as obtained by the e-voting=1 model (use of electronic voting machines) and the e-vote=0 model (no e-voting), that the correct way to interpret this is that there were between 130,000 and 260,000 “excess votes” for Bush. But mathematically, it’s equally valid to suppose that there was a Republican suppression factor in the 2000 and 1996 elections — that is, that the Democrats cheated in counting punchcards in heavily Democratic districts in past elections — which they were unable to do in 2004 with the electronic voting machines.
In fact, it is easier to cheat with punch cards: for one easy example, if you take a stack of ballots and push a long stylus through the “Gore” hole, this will have the effect of turning Bush votes into uncountable Bush and Gore double votes, while leaving Gore votes undisturbed. This requires no computer sophistication at all (or even a high-school diploma) and can be done by the lowest level of poll worker, long before the cards even leave the individual precinct.
All that the study actually found was that Bush’s support in Democratic counties of Florida increased more than the level one would have expected from projecting Republican support in 1996 and 2000, and that this increase seems correlated to electronic voting. It does not and cannot tell us whether those “excess” votes are in fact legitimate or a result of fraud… for all that the authors pretend that’s what they have found.
Without having analyzed the study, my guess is that it doesn’t prove anything at all. But Dafydd’s observation that if the study really is meaningful, it probably shows that the Democrats were cheating in 1996 and 2000, is interesting.
The Democrats have been stealing elections for generations. One happy consequence of the Dems’ current tinfoil-hat obsession with (nonexistent) Republican voter fraud could be that it may be hard for the Dems to resist efforts at real reform. If the Republicans push hard for reform at the national level, it might be possible to make fraud harder between now and 2008.
UPDATE: Patrick Ruffini has taken a stick to the Berkeley authors as well. It is amazing how you can get people to pay attention to the most transparently stupid hypothesis, as long as your point is to bash President Bush and the Republicans.