Tuesday, October 26, 2004

Myths

The next seven days are Halloween, Thanksgiving and Christmas tied up in one for bloggers. And yet, since I expect to be very busy, my blogging may be light. Oh well, I never wanted to unduly influence this election.

Every election, certain analytical myths get trotted out. Occasionally, a new one arises. The latest, for instance, is that since a growing number of people use only cell phones, and pollsters call land lines, that polls are biased against Democrats. Or something like that. Now it's true pollsters will have to eventually adapt, but this argument is, at best, overstated.

My favorite myth, and it's been around at least a few election cycles, deals with "weekend polling." For some reason, many believe that polls done on the weekend are considerably different from those on the weekdays, and less reliable to boot. As far as I have been able to determine, there is simply nothing to this.

If anyone has any evidence I'm wrong, please send it in. By evidence, I don't mean "everyone I know believes it" or "here's the reasons weekends are different." The former is an argument from (questionable) authority and the latter is giving a reason for a non-existent phenomenon. What I would regard as proof is statistical research, with a large sampling, that shows weekends and weekdays provide predictably and significantly different results

web page hit counter