One of the favorite past-times of political junkies is following polls. I, for one, believe that we have reached a polling saturation point. This year one could almost garner a poll to say anything about the election. They moved up and down akin to a vibrant stock market. But, again, I think polling is over-rated as a predictive device. The technicians have some technical explanations of the decreasing utility of polling. For example, a good many in the generation under 30 have cell phones as their primary phones - they are left off the polling rolls. A second concern is that some polls over-rate one group or another in their samples. A third criticism is that polls seem to purport more certainty than they actually have - the wild swings in results before an election probably do not happen but are more accurate representations of the sentiments of the polled groups which are afterall a sample of the electorate.
Jon Cohen of the Washington Post did an excellent article this morning on the foibles of the polls. But there are two things I used in this cycle which I think, for me, made a bit more sense of their seeming gyrations. The first is one of the systems which aggregated polling results. I like Real Clear Politics, whose editorial content is decidedly conservative, but whose polling averages were just about right in most of the important races. In some cases, for example the North Dakota senate race, where the GOP failed to recruit a strong candidate, there was no polling, but the RCP averages helped to smooth the data. Second, the futures markets, originally developed by an Iowa scholar, which allow political junkies to buy and sell futures were pretty efficient predictors of what would happen in individual elections and in the general trends. I think they were much more accurate than the "generic poll" which seems to be governed by some many variations in interpretation as to be almost useless.
Cohen concludes his article with the old southern saying about trying to put lipstick on a pig. (In my region of the country the saying dealt with teaching a pig to sing - but you get the idea.) Polling is something, like any other market, that will continue to proliferate with all of the consequences of that kind of competition. But those two smoothing devices will help to make sense out of the Babel of pollsters.
Tuesday, November 14, 2006
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment