Exit Poll Data Does Not Match Pre-Election Polls

It is now well known that the final exit polls from the November 2 Presidential contest between George W. Bush and John Kerry were adjusted (weighted) once actual vote tallies arrived in order to match the reported vote. This paper will first demonstrate that the weighting process was invalid. In other words, it is mathematically impossible to match the exit polls to the reported results of the popular vote. Then this paper will compare exit poll results to pre-election poll trends over the last month leading up to the election and then provide some other clues regarding which states may have fraudulent results.

Weighting of the Exit Poll Is Impossible

The final fixed exit poll shows how the electorate of 2004 broke down compared to the voting in the 2000 election. And what it reveals is that in order for Bush to win, a virtually impossible thing happened: every single Bush voter from 2000 also went out and again voted in 2004. That is, no Bush voter passed away from 2000 to 2004 or for whatever reason, could not vote in 2004. It is perhaps the greatest electoral miracle that Karl Rove has ever performed!

A very large number of people voted – 122.6 million. The 'fixed' exit poll says that of these, 43 percent voted for Bush in 2000, 37 percent voted for Gore in 2000, 3 percent voted for Nader/Other in 2000, and 17 percent did not vote in 2000.

Translating this into numbers means that of the 122.6 million voters in 2004, 52.6 million voted for Bush in 2000, 45.4 million voted for Gore in 2000, 3.7 million voted for Nader/Other in 2000, and 20.8 million did not vote in 2000. Really?

In 2000, Bush received only 50,456,169 votes. So 104 percent of Bush's 2000 base returned to polls, compared to 89 percent of Kerry's base. This is impossible! And this is important, because the exit polls show that Kerry won new voters, Kerry won voters who did not vote in 2000 (54 to 45), and Kerry overwhelming won voters who voted for Nader or someone else in 2000 (71 to 21). Also, the exit poll shows that Bush and Kerry swapped about an equal number of voters in 2004 -- ten percent of Gore voters went for Bush in 2004 while nine percent of Bush (2000) voters went for Kerry in 2004.

So the only way that Bush won the election in 2004, was by having a better turnout of his base. His turnout was so good, that it was mathematically impossible! First, obviously some Bush voters passed away from 2000 to 2004. Let's be conservative and say that only 2 percent of Bush's 2000 voters died between 2000 and 2004 - that is, just over 1,000,000. That leaves us with at most 49,450,000 potential Bush-2000 voters. This means that even if every single Bush voter from 2000 returned to the polls in 2004, it could only be 40.3 percent of the electorate, not 43%. And even that assumption is highly unlikely.

The fixed exit polls are trying to convince us of a Bush win based on a mathematical impossibility.

Equally important, once you change the weighting of the poll, the whole thing, all the questions, need to be re-weighted. This can not be explained by the margin of error. Once you do the re-weighting, the reported results will be outside the margin of error of the exit poll. In other words, the national popular vote total is impossible. It is far outside the possible margin of error of the national exit poll survey.

Finding the Clues – Where Did Fraud Occur?

The exit polls were manipulated to produce at least three results. One is to get the exit polls to match the "actual" Bush margin of victory in key battleground states, the second is to match results in non-key states where the reported vote did not match the initial exit poll in order to boost Bush’s popular vote “mandate” and the third was to get the exit polls to show that there was not a major swing toward Kerry during the last 24 hours. The latter was necessary because if Kerry really had been winning 60 or 65 percent of the undecideds on election day (the people who made up their mind on the day of election), then it would be really hard to explain how Bush wound up winning both the popular vote and the electoral college. Instead, the exit polls were fixed to state that nationally Kerry won the voters who decided on the day of election by a scant margin of 52 to 45. The exit pollsters had a major challenge though – if they fixed the exit polls in the same way for all the states, it would be too obvious that something was amiss.

The first question is how legitimate are the results in the key battleground states that Bush ended up winning. According to the exit polls, the reason that Kerry lost Iowa, New Mexico, Ohio, Nevada, and Florida is not that he didn't close well but rather that he was coming from so far behind. For instance, among voters who made up their minds in the last week, we see that Kerry was winning the undecided vote quite well – anywhere from a low of a 56 to 43 margin in Iowa to a high of a 59 to 38 margin in New Mexico.

Clearly the problem was not the lack of a strong finish on Kerry’s part. Rather, according to the fixed exit polls, the problem was that Kerry was too far behind and had too much ground to make up. Of voters who made up their minds more than one month before the election, Kerry lost big: He was so far behind by the time of the first debate on September 30, he could not make up the difference. According to the exit poll data, his deficit one month out ranged from seven points down in New Mexico to 12 points down in Florida.

But was Kerry really that far behind with a month to go? In order to analyze this question, pre-election polling data was examined. Some caution is in order because the pre-election surveys are worded to tease out responses that may not be very solid. But at least a similar bias should be shown in all states. The polling data from the state-wide polls of likely voters carried out between Sept. 15 and Oct. 2 (one month to 7 weeks before the election) was examined and compared to the data above from the exit poll. The state-wide pre-election polls showed a virtual tie in New Mexico and a four point lead for Bush in Iowa and a five point lead for Bush in Ohio, Nevada, and Florida. More importantly, the November 2 exit poll data was underestimating Kerry’s percentage one month out from the election by 4 to 7 points, with the highest totals being New Mexico and Florida.

The state-wide pre-election polls conducted five to seven weeks before the election would suggest that the exit poll data is off – that it consistently was over-estimating the margin between Bush and Kerry among voters who decided more than one month before the election who they were going to vote for. In the three states with the most polls done – Iowa, Ohio, and Florida – the exit poll over-estimated the Bush lead by 4 to 7 percentage points. Most significantly, in Florida the exit poll gave Bush a margin of 56-44 among voters who decided at least 30 days out. Yet the pre-election polls conducted between Sept. 15 and Oct. 2 suggest that Bush had only a 4 or 5 point lead then.

One can argue that Bush voters who responded to the exit poll made up their minds earlier than Kerry voters, so that there is a built in bias in these polls. This argument would say that the difference in pre-election poll data and exit poll data simply reflects that a number of Kerry voters told pre-election pollsters in September that they were going to vote for Kerry but then (collectively) told exit pollsters on November 2 that they actually made up their minds sometime during the last month before the election. This is possible.

One way to test out this argument is to look at the five closest battleground states where Kerry won. How much bias was there in the pre-election polls in those states? The same methodology was used – gathering the data from all the state-wide polls of likely voters and finding the average for Bush and Kerry. On average there was no bias in the polls. Only one state, Michigan, had a bias similar to the five battleground states won by Bush, although the bias was smaller than in any of the five “red” states. On the other hand, Wisconsin showed the opposite bias. There, either the pre-election polls were over-estimating the Bush lead 30 days out or the exit poll was over-estimating Kerry’s support one month before the election.

This is worth repeating – in the five battleground states won by Kerry the November 2 exit poll data matched the pre-election polls regarding voters’ inclination more than one month before the election!