Tuesday, January 08, 2008

Addendum to the Lack of Wisdom of Crowds

This news item from the CNN site, gives the appearance that I was a ditz in what I wrote yesterday. The story says that about 20% of those voting in New Hampshire today made up their minds about whom to vote for day of (or they were seriously willing to reconsider their choice). I said in my previous post that you could tell what happened in the Iowa Caucus well in advance of that event.

I may very well be a ditz in a host of other matters but on this one, I don't think so. There is a problem in how these pollsters frame their questions and so they throw out a lot of information that would be useful in making predictions. Here's the scoop.

I pretty much know who I'm going to vote for on February 5th (Super Tuesday) but certainly leave open room to learn new things about the the candidates between now and then and depending on what I learn, I could change my mind. The likelihood of that happening is low, in my view, but it is possible. Say I'm 90% confident of choosing my currently preferred candidate, leaving a 10% chance that I might change my mind. Does that make me an undecided? If quite a bit of the population is like that and further if they have the same candidate who they a likely to vote for, then that candidate should be the favorite going in. But the way we measure things, we wouldn't know.

There is a further confirmation of this way of thinking about what is happening. Ask yourself the following, other things equal. Which person is more likely to vote, the one who if he did vote would likely choose one particular candidate or the other who if he were to vote is presently completely undecided - each candidate is equally likely to be selected? The news is talking about record turnouts (some of which, to be sure, is the weather but is probably not all of it). In my view that's because these voter know who they wanted to vote for and they've known that for a while.

No comments: