Thursday, October 22, 2009

A Short Comment About Polls

I will post more later, But there is something that has been bothering me for a while about the way polls are presented.

Numbers don't lie, but numbers can mislead. A non-political example here:

Auto insurance commercials claiming that the 'average person who switched saved' some hundreds of dollars. It's a non-statement. Of course those who switched saved money. Those who called for a quote and found they wouldn't save money didn't switch. It's akin to saying "90 percent of my friends think I'm a decent guy."

This pertains to polls because, again, numbers can be misleading. When you see job approval ratings, they might be close to accurate. When I see a news broadcast covering healthcare, and they ask, "Do you approve of Obama's plan for healthcare," or "Do you agree with Obama's handling of healthcare reform," or any version thereof, it's too ambiguous to mean anything.

There are false negatives, but no false positives to balance the poll. If you think the public option or the broader term 'reform' is a bad idea, you say no to the question. If you think it's a good idea, but still think he's not doing enough to reach the goal, you say no. There are people who want a public option, or want Obama's goal to be reached, that will say no because he's not doing enough. There is no scenario that would make a person opposed to reform or a public option answer 'yes' to that poll question.

Because of those false negatives, with no balancing false positives, I'm skeptical about the common wisdom and the interpretation of these polls.

No comments: