Cox Guarding the Henhouse
For Whom the Polls Toll
As it turned out, better media coverage probably wouldn't have made a difference anyway -- looking at Bush's 38% margin of victory, it appears that he would have had to kill babies to make this election go the other way -- but Mauro had a good point, and one that applied even better to the other races on the ballot.
The news media would do its readers a service by not obsessing over the polls so much, and spending more time tackling election issues. An election should not be viewed as a horse race, where we analyze who is in "the lead" and who is "behind." It is a decision. And it is a decision that is made only once -- on election day. While the polls may give us a hint of what is to come, ultimately, they matter not. Only the last poll, at the election booth, does.
And what do we really learn from these polls? Not as much as those who produce and report them would like us to think.
The poll that kept getting mentioned in seemingly every election article in the Austin American-Statesman over the two months leading up to the decision was a Texas Poll released on Labor Day. Conducted from Aug. 12-27, the Office of Survey Research at the University of Texas asked 718 Texans who identified themselves as likely voters what their choices would be.
The resulting article's headline trumpeted, "GOP leads all but 1 major race -- but barely." On the surface, that might in retrospect appear to prove the Texas Poll's accuracy -- indeed, the GOP swept the statewide offices, mostly with close margins of victory.
But examine those poll numbers a little more closely. Toss out the gubernatorial and attorney general races, and the average percentage of those polled who said they were "undecided" in the other five races is 30%! With that many undecideds, can we really say that we know anything about the future of these races?
Let's look at those closer races: In the lieutenant governor's race, Rick Perry supposedly had a six-point lead over John Sharp -- a gap which closed to two in the actual election. Similarly, GOP comptroller candidate Carol Rylander's five-point lead over Paul Hobby closed to a mere one point. But the huge undecided bloc -- a quarter to a third of the total in those two races -- could have fluctuated in any number of directions. You can't just assume that they're going to split 50-50, and the AG and gubernatorial races are the strongest proof. If we assume that the Labor Day poll's numbers were valid, then all 10% of the undecideds swung to Mauro. And of course, the AG poll was completely blown -- Mattox not only lost the entire 18% of undecideds to Cornyn, but also several who had already made up their minds to vote for him!
So the truth is, nobody really had a "lead," despite what the media reported. Nobody has a lead until the actual ballots are counted. While the pre-election polls are somewhat interesting, they do not merit the importance which reporters place on them.