On Polls

A CNN/USA Today poll just published tells us weird and wonderful things about the group of Americans that pollsters call 'likely voters'. First, 51% would choose Bush as the president and 47% would choose Kerry if only these two men were offered for s/election. Second, if Nader is added to the menu, 49% would pick Bush, 45% Kerry and 4% Nader. Third, half of the respondents believed that the Iraq war is a valid part of the war on terrorism. Fourth, president Bush's job approval percentage rose after a week of negative media headlines about the way the Bush administration handled the terrorist threat before 9/11. Fifth, 80% of likely Kerry voters trust Clarke's testimony more than the administration's, while 81% of likely Bush voters trust the administration's testimony more than Clarke's.

Well, not all of this is so weird and wonderful. The first result, for example, has a margin of error of plus or minus four percent, so it's even possible that Bush and Kerry would get the same number of votes if the elections were today and if all voters resembled the likely voters in this poll. Unless Nader entered the race, of course. Though, surprisingly, Nader seems to take votes away from both of the main runners. The fifth result isn't that unexpected, either, as the decision to vote for a particular candidate is pretty highly correlated with the opinions one has of that candidate. Of course, we can't tell which of these is the egg and which the hen...

But the remaining two findings are indeed miraculous. Think about it: there is no evidence that Iraq had anything to do with the terrorist attacks against the United States. Yet half of the respondents know something the world's elite intelligence professionals don't: the Iraqis had their fingers in the pie! Why aren't these people questioned by the 9/11 commission? Or better still, why aren't they hired to do the intelligence work for their country?

Even more awesome is the likely voters' reaction to Clarke's testimony: they now like Bush even better than before! I could understand Bush's job ratings not declining among those who are going to vote for him come whatever may, but for these ratings to rise? Once again, these likely voters are blessed with some extremely valuable inside information, and I want to know what it is.

Polls are silly things, on the whole. Imagine trying to make people answer a poll like this one: how many refusals do you think the pollster gets for each willing respondent? The answer is a lot. Then imagine trying to squeeze definite statements from people who often have but the fuzziest idea of the issues they are being quizzed about. The only scary thing about all this silliness is that it's rather reflected in the elections themselves: most stay away and those who participate often don't know what they are choosing.