ACORN: Election Fraud? A Review of Election Fraud in the USA

With the election coming up soon in the US, there has been some claims about election fraud. The latest came with Republicans attacking ACORN for fraud, an accusation that FactCheck.org finds correct, although not to the degree that is stated. However, the main question is whether these claims will surface following the election in November.

The scenario is easy to imagine. Obama takes the election day polls, McCain wins and what then?

This already happened once, however. In 2004, John Kerry won in nine of eleven battleground states according to the original exit polls. With the final results in, Kerry only won five of those states. In other words, the polls determined the wrong winner in 33% of the battleground states.

As it turned out, in ten of the eleven states, Bush also did remarkedly better in the actual results than in the polls. In Ohio, for example, Kerry won according to the exit polls by 4,2%. The end count showed Bush 2,5% ahead. In New Hampshire, the exit polls showed Kerry ahead by almost 11%. Kerry kept the state but won by only 1,3%. These examples only mirrors the remaining 8 states.

While the media ascribed what happened to errors with the polls, I still find it hard to believe that in 11 of the 12 states, Bush fared better than the exit polls predicted. I find it hard to believe that in 33% of the polls, the state was given to the wrong winner. And I find it hard to believe that all those four went the same way.

One thing is certain about polls, they tend to have errors. Error should be random, however, and random errors should go both ways. It should not randomly favor one of the candidates. The alternative explanation is that there are structural errors. Something goes wrong in the estimation, like a Bradley-effect. It was speculated that Bush voters were more reluctant to specify their opinion, but it seems like an inadequate explanation. Yes, Bush was not popular, but was it really that bad to vote for Bush? And was it the same in all these states? It makes no sense to me.

Both the average people and the media found such ways to explain these disreptancies. The polls were wrong, the data had not been completely worked through and polls are generally weak. In other words, people don't trust polls anymore.

There are obvious flaws with these arguments. Why would trained professionals publish polls that had not been weighted properly? Pollsters wish to get the results right and they will do everything to get them as right as possible. Eventually these pollsters came to "correct" their numbers, but if so was needed, why didn't they do it prior to releasing them? Could it be that they actually thought they were right first? Zogby called the race to Kerry, but would he do it if he wasn't sure? Why would they not do what every pollster does to get the results as accurate as possible?

And why has exit polls previously been very successful at determining the results in US elections? Why is it so that in Germany, pollsters can get the results down to at most 1,5% off the actual results (and most often far below that), while in the US, results can be almost 10% off? The one plausible explanation I can get to match up is that pollsters in the US are dumber than those in Germany. I do, however, have a feeling that most Americans would reject this explanation, and they'd be right to do so. Frankly, there is no reason to think that those in Germany are much better.

Looking at the nomination race in the US now, we might have expected pollsters to deliver as bad results as in the election in 2004. However, my experience with those exit polls was that they were extremely accurate. Mostly, they tended to get the results down to 1% or 2% accuracy. I personally called races based on CNN exit polls earlier than the channel did, simply because those polls are trustworthy. The reason they are trustworthy is that they are fairly easy to get right.

In the nomination process, even polls prior to the election tended to be right. While some polls might have been slightly off, those did not interview people who had actually voted, but rather people who were registered or were deemed likely voters by the pollsters (or they were simply registered voters). Thus, it should be expected that such polls are worse indications than the actual exit polls. The results were, however, predictable. The poll of polls at Real Clear Politics was right in every race, except the one in New Hampshire. Occasionally the results were slightly off, but this is expected due to some groups of voters failing to show up.

It is extremely difficult to draw any conclusion based on the exit poll experience last time. Based on a study of exit poll data, Professor Steven F. Freeman concludes that election fraud can not be rejected as an explanation. While I agree with this conclusion, I believe that the real test is whether exit polls are equally weak at predicting outcomes this time. If they are, election fraud is a very plausible explanation. After all, similar finds in Georgia and Ukraine has made people conclude it was election fraud there, and it shouldn't be rejected simply because this is the USA. Wasn't it so that all Americans are Georgians?