Thursday, October 14, 2004

The Polls Mislead

I have become a dedicated "opinion poll watcher". I check the web regularly for the latest poll results. Each day brings a new reason to see what the polls are reflecting.

It strikes me that the variation from poll to poll is, at best, confusing and, at worst, misleading. Is Kerry ahead by one point or two? Is Bush ahead by one point or two? On any given day both results are reflected. Yet, the polls consistently report that there is a margin of error, at 95% certainty, of 3 to 4 percentage points. Basically, with less than a four point spread, the polls are meaningless exercises in trying to make something out of nothing. Why don't the polls report that the results are dead-even when there is so little difference, insufficient difference to make a difference?

While on the subject of polls, I inherently distrust polls that do not expose their questionnaires, timing, weighting, and samples. I'm sorry. There is just too much opportunity for fudging and incompetence. So I guess we have to view the polls as a form of entertainment, with no more significance than a performance of "Last Comic Standing." I just hope no one is putting any stock in them.

By the way, what are "tracking polls" anyway? Do I understand that they are polls taken over several days with the accumulated results presented on the day after the most recent survey? How accurate is that any way, with important events interceding? I recall looking at one poll's results where there was almost a 180% swing in results from one day to the next. Yet the results were presented as an average of the two days. The mind boggles.

We yearn for November 4th, when polls will be returned to opinions about who the sexiest male star is. That is a level of importance that is appropriate.