Close Elections

As campaigners become more expert in their use of statistical analysis applied to polling, they have the power to become more efficient in their campaigning. Start with an easy example: if Candidate Smith is leading by a 20% margin in Tarball County over his nearest competitor, it's obvious that he won't campaign there anymore. He's won Tarball County.

But even if he's leading by a tiny margin, statistics can still tell with measurable confidence whether he will still win. For example, a lead of just a 1.5% in Tumbleweed County, population Y, based on random sampling of Z registered voters, might translate into a 90% chance of winning that region. No, it's not a 100% sure thing, but getting that extra 10% would be very expensive, particularly in terms of opportunity cost in other regions where there's only a 55% chance of winning. He'll probably skip further campaigning in Tumbleweed County.

If a candidate has confidence in his pollsters, he can balance factors like certainty of winning, dollars/time spent campaigning, and margin of victory. I am sure that candidates do this, and that their tools are more accurate than ever. Dean was the leader entering into the Iowa race; in the absence of unexpected, sudden events, he's likely to be able to maintain that lead. But he may have decided that there's no need to crank up the decisiveness factor, and instead conserve money and time. I think that's what we're seeing: an efficient campaign drawing in part on proper use of statistics.