Lies, Damn Lies & Demographics

Not long ago, I mentioned Bush's polling troubles. It seems that when I bring it up in the cafeteria, the most common response is, "polls don't mean anything." Hrm. This is obviously untrue. What is an election but a poll? And the election results you see on TV aren't the votes being counted, but the results of exit polls. But people dislike the idea of polling, because they dislike the idea of demographics. A David Horsey cartoon from the Seattle Post-Intelligencer:



Because, if people can be neatly racked into a cartoonist's caricatures, it sort of challenges their individuality. I used to do telephone interviews for Dan Jones & Associates, a Utah polling firm, and when you call someone up, they become uncomfortable at answering questions. This is for the same reason people dislike the whole science of demographics.

No matter that the hundred million or so Americans who vote this November will have something on the order of at least two hundred millions reasons for voting the way they do. We will either vote for George Bush or John Kerry and that is all that matters. We will sort ourselves into, if I may borrow Horsey's alliteration, Bush Bumblers or Kerry Kiakamzies.

So it's easy to understand why people dislike demographics (if you can get past my horrible alliteration). But, to contradict everything I just said, the real value of statistical analysis doesn't appear until you understand it just a little better. Ryan Lizza at TNR blogs:

We can already assess the effect of the two big strategic moves of the pre-convention period. The Bush campaign's decision was to spend some $60 million in an attempt to discredit Kerry as a viable alternative to the president before the race really started. The Kerry campaign's decision was to concentrate on fundraising and allow events in Iraq and 527 spending to parry the Bush assault. Conventional wisdom among nervous Democrats outside the Kerry campaign, as well as much of the press, was that Kerry was making a Titanic mistake and Bush was making a bold and brilliant move similar to Clinton in 1996.

But the results are in. Kerry leads Bush in almost every national poll. His fundraising is astronomical, and he is pumping up his ad campaign just as Bush is ratcheting his down. The two main assumptions of the Bush campaign--that Kerry would be seriously under-funded and that he could be crippled by advertising--have proven to be wrong.

This is a good example of how political analysis works. You look at 1)what the candidates are saying, 2)where they are spending their money and 3)what the results are. But without demographics, the first two are pretty meaningless.

Sure, you can just see who wins an election and draw conclusions from that. We run an election cycle that lasts about 18 months in America, and a presidential election has a budget about twice that of making the Lord of the Rings trilogy. (Yeah, that's right: all three.) Campaign managers want to be able to make course corrections. Wouldn't you?

Anyway, this post seems to have wandered a bit more that my usual writing, with no particular underlying point. Sometimes the bear gets you.

Post a Comment