Sunday, December 17, 2006

Inappropriate Use of Polls

"Do you think California will be struck by a major earthquake in the next ten years?" Why is that a poll question? You could make some argument that asking geologists who have studied this area in specific what they think the likelihood is of an earthquake. That might have have some value for what resources we devote to preparation for that. But even if 100% of the geologists said it would, or 100% said it wouldn't, that won't affect the earthquake.

"Will the earth be struck by a major asteroid in 2036?" Not really a poll question. There is a fringey sort of answer that would say if enough people think the asteroid that is coming is going to hit us, we will take steps to make sure that it doesn't. In that sense, the poll indirectly leads us to some useful information.

Is OJ guilty? Will this be a snowy winter? Is Chicago west of Detroit? Some things are just not poll questions.

I bring up these obvious errors because most of the poll questions in the major media are not much better, but we don't tend to notice that. "Do children study more or less than they did thirty years ago?" "Is there more crime than there was ten years ago, or less?"

Pick a country and a disease at random. I pick Turkey, and diabetes. What if next week Newsweek starts in with polls "Are we sending too much aid to Turkey, not enough, or the right amount?" None of us knows how much aid we send to Turkey. But who is going to say "send more?" Unless there is something about Turkey in the news that suggests we should be sending them more, such as a natural disaster, everyone is going to say either "about right" or "send less." On the other hand, who is going to say that we should spend less on diabetes research? Not that we know how much is spent now or how much is needed, but the fact that the question is even asked starts a whispering campaign for the answer that someone wants. Well, let's keep that up for a few months. Tim Russert asks Dick Cheney "Are we sending too much to Turkey?" "Are we spending enough on diabetes research?" Other magazines and newspapers start commisioning polls about it. You can eventually start to get the answer you want.

Oh yes you can. That's why disease research even has PR campaigns, isn't it? That's one of the resons that Turkey even has an embassy here, isn't it? To move opinion. The objection that Newsweek or USA Today or Tim Russert wouldn't engage in that sort of random manipulation isn't the point. Random manipulation isn't the point. The point is, do whispering campaigns work?

"Are we winning the war in Iraq?" That is not a poll question. It is in fact a question that cannot help us win, but might help us lose the war in Iraq. If enough people say "no," then politicians make noises about having to do something different. Plus, our allies wonder if they can trust us. Further, our enemies wonder if they can win by just hanging on a little longer.

Most people in America haven't the faintest idea whether we are losing or winning, but most of them have an opinion anyway. At the moment, most people think it's a stalemate, with smaller percentages saying "we're winning," or "we're losing." To try and improve the quality of the polls, sometimes they will ask people who look like they might know more than the rest of us. History professors. Retired State Dept. officials. Think about it. They don't know more than we do. And we don't know anything. Experts in related fields by necessity have to look at these questions from data which they know which is only part of the story. "Well, the Sunni tribes came to Anbar province later than the Shia..." So what? Even if it's true and the historian is right on the money, it's 2% of the picture.

I have folks who I correspond with who would at this point claim I was advocating suppression of speech critical of the Administration. Really. In fact if you scurry around the comments sections of the blog sites, and the postings of the big leftie blogs like Kos and HuffPo, you can find a lot of people who say that. I work every day with people who believe that. They don't say it every day, but they have said exactly that in the past year. Some people, by which they think they mean someone other than me, want to cut off debate about the war altogether. They don't want to allow anyone to criticise the president. It's scary.

No, it's scary that people believe that.

1 comment:

Woody said...

Well, the could do even worse, like the BCS (college football championship) and combine polls along with computer rankings made by techno-nerds who program data that gives results contrary to all common sense. I can see CNN saying that according to its latest poll and its computer analysis that space aliens will invade us next year. Sadly, many people would believe it.