In the immortal metaphor supplied by Gary Cruse on this blog this summer, the blogging world today wears a potato down the back of its swim trunks, instead of the front.
Yesterday, we believed the exit poll data. Mind you, so did both parties, with gloom at the Republican headquarters and premature joy in Boston.
These polls showed Kerry beating Bush by two to three percentage points in the popular vote and they showed him taking Pennsylvania and Ohio. Two of three surveys in Florida showed Kerry winning there, too.
In the words of the Jim Rutenberg of the New York Times, the exit polls invited us to believe that “Kerry [was] on the verge of winning the three states most pundits believed could sway the election.”
Now the question is this: why was the data unreliable? Maybe it’s a methodological problem: the pollers put the wrong questions, to the wrong people, in the wrong way, or something. We will hear lots on this possibility in the next few days.
I want to raise the issue I cribbed from Diane Francis a couple of days ago: that, in the pre-election surveys, many voters who were claiming to be undecided were in fact concealing their intention to vote Republican. It looks as if the same thing may have been happening at the exit polls. People who had just voted Republican were claiming to have voted Democrat.
Diane Francis suggested that in the case of pre-election surveys, Republicans had been shamed or brow beaten into lying. This is possible, but not compelling.
But alternate suggestions are not forthcoming. Could this have been a difference between what the voter wanted to do and what the voter felt he or she had to do? “I wanted to vote Democrat but I felt obliged to vote Republican.” Did they claim to have voted Democrat because this was the better, the more winning, the more becoming choice? Did they claim a Democrat vote as a fashion accessory?
None of the alternatives to Francis’ suggestion is very plausible. But clearly something very odd happened here, and we may have to resort to a very odd explanation to make sense of it. Why did people lie?
References.
The earlier post, The Bush Victory: you read it here second, may be found here
Diane Francis’ post from the National Post may be found here
Zogby (link above), while conciliatory, explains that their guesses were within the margin of error.
I’m prone to statistics, and so am guessing that the issue lies more within sample size than something as implausible as the weight of personal “peccability”.
I am definitely not an expert, but I think you can get different results based on the time of day you do the exit polling. The people on line before the polling place opens have a different profile than the ones who stroll in at 11:30, and are different still from the people who rush in at 7:55 p.m. So an exit poll that is reported at 12:00 noon, can only have those people who voted by about 10:30 am or so.
Nigel, thanks for the link. Zogby seemed to be a little soggy, and what in heaven’s name was he doing making so much of something within the margin of error? Thanks, Grant
Joe, thanks and good point. I thought the conventional wisdom was that white collar voted on the way to work and blue collar on the way home. This should have made the exit polls capture more Republicans. And perhaps they did, making the “deception” effect all the more remarkable. Thanks, Grant
I had thought there was some discussion of assymmetry in the people polled. It didn’t quite make sense perfectly for me, but the early polls were claimed to be 60/40 women…and that was said to have maid a difference. In any case sampling error was not unusual. Also, I read that hearing that exit polls were pro-Kerry in the morning didn’t particularly phase the Bush folks who were in-the-know. Why that is (4-years ago systematic error?) I’m not sure.
BTW, love the blog.
Another explanation:
http://www.tompaine.com/articles/kerry_won_.php
http://randomwalks.com/archive/2004/11/will_no_one_speak_of_this_part_2.php
Is it true? I have no idea. But I wont dismiss it out of hand, given different things I have heard about previous elections.
The key graph is this one, here:
http://img103.exs.cx/img103/4526/exit_poll.gif
Unfortunately, I don’t know how to assess how reasonable the argument is. But hopefully we will find out …
It wasn’t the exit polls that were wrong. It was that a percentage of the votes cast for Kerry were electronically transferred to Bush, intentionally. All I hear are comments that the exit polling, which by the way, has been a very accurate means of determining who the winner would be in the past, was somehow not reliable, all of a sudden, and that the problem lies in the exit polling itself. Bunk. Diebold provided and programmed the electronic voting machines, donated heavily to the Bush campaign, and the owner promised Bush to deliver the election to him. Wake up people. Democracy is gasping it’s last breath, and the media is just allowing it to happen.