Bush variation from exit polling
Interesting figures. Let's take a group of states:
Arizona
Louisiana
Michigan
Iowa
New Mexico
Maine
Nevada
Arkansas
Missouri
Illinois
If we take the average for these states, Bush was leading by 2% at the
2pm exit poll (50.5 to 48.5)
At the 4pm exit poll it was down to 1.5% (50.2 to 48.7)
In the official results, his lead was 2.3% (50.7 to 48.4). That is
well within the margin of error. In essence, Bush did better by 0.8%
compared to the exit polls in these states. His biggest gain
compared to the exit polls in this group of states was in Maine, a
gain of 3 percentage points. Got that? Ok. Now let's look at
another set of states.
Wisconsin
Pennsylvania
Ohio
Florida
Minnesota
New Hampshire
North Carolina
Colorado
In the 2pm exit poll, Kerry had an average lead of 7.5% in these
states, ranging from 20 points in Pennsylvania to -3 in Colorado. In
the 4pm exit poll, Kerry's average had dropped, but he was still
leading by 3.88 percentage points. At that time, he was only
trailing in North Carolina and Colorado, and had leads in the other
states ranging from 17 in New Hampshire to 1 in Florida (the 2pm lead
in Florida for Kerry was 3).
So, in these states, we find
2pm Kerry 53.38 Bush 45.88
4pm Kerry 51.50 Bush 47.63
But then in the official result, we get this:
Kerry 48.38 Bush 50.75
Remember, in the previous set of states, Bush did better than the exit
polls to the tune of 0.8% on average.
But now, in this set of states, he does better than the exit polls to
the tune of 6.25% on average. That's nearly eight times better than
his gain relative to the exit polls in the first set. And well
beyond the margin of error.
What could account for this difference. Well, paperless voting was
far more common in the second set than in the first.
Ironically, in October, the New York Times wrote about the importance
of exit polling to establishing the integrity of elections...
Exit Polls to Protect the Vote
By MARTIN PLISSNER
Published: October 17, 2004
WASHINGTON -- Since the 1960's, the exit poll, that staple of election-night television, has been used along with other tools to declare winners when the polls close in each state, and its accuracy is noted later when the actual vote count proves it right. A landmark exception, of course, came in 2000, when the networks initially gave the decisive Florida vote to Al Gore.
But now exit polls are being used in some places to monitor the official vote count itself, either to validate the outcome or to mount a challenge to it.
That has happened in several countries in the last year, and in the United States one organization plans to use exit polls in five closely contested states in November to measure whether there have been impediments to voting.
Last fall, an American firm, whose polling clients have included Al Gore and John Edwards, was hired by some international foundations to conduct an exit poll in the former Soviet republic of Georgia during a parliamentary election. On Election Day, the firm, Global Strategy Group, projected a victory for the main opposition party. When the sitting government counted the votes, however, it announced that its own slate of candidates had won. Supporters of the opposition stormed the Parliament, and the president, Eduard A. Shevardnadze, later resigned under pressure from the United States and Russia.
Full article here:
nytimes.com |
