By John Aloysius Farrell, Thomas Jefferson Street blog
The liberal and conservative bloggers here at Thomas Jefferson St. throw a lot of poll numbers at you in the course of the year. Predictably, we cherry pick the numbers we like, and flaunt them. I don't really know why--I guess to rally the troops with the thought that victory is at hand. Polls measure nothing but momentary popularity, not wisdom or courage or brilliance or innovation. They certainly won't answer the pressing questions of the day, like whether Sarah Palin has gotten breast implants. You go girls, says I.
(Since we are talking about polling, you can weigh the visual evidence and cast your vote--Did she? Or didn't she?--at Huffington Post.)
Anyhow, over at the FiveThirtyEight website, ace political analyst Nate Silver has published an analytical survey of the better-known polling firms in America, and ranked them on accuracy in the final weeks of an election. Here are some results that jump out. You may want to remember them the next time one of us acts like a know-it-all, using a poll as evidence.
The big names deserve their fame. Field, Mason-Dixon, Gallup, Pew, Marist, and Rasmussen all did very well. So too, did most of the bigger media polls, especially the Associated Press, the ABC-Washington Post poll, and the NBC-Wall St. Journal poll. And one impressive result, to me, was the high marks that SurveyUSA got. They conduct hundreds of polls for media clients, apparently quite accurately.
The suspects in the media category: Fox (Shocked!), the Los Angeles Times, the Hotline and Newsweek.
But even the sloppiest media polls shine in comparison to partisan polling firms.
I hereby vow never to cite Public Opinion Strategies (GOP) or Democracy Corps (Dems) in a blog again. And if I do, don't believe me. I don't know if the Tarrance Group is a statistical outlier, or just generally more accurate, or are bungling their assignment as Republican shills, but they are the exception among partisan firms.
And, finally, don't bother with results from "interactive" polls. I never cite them, and Silver's analysis shows why. They're not ready for prime-time yet, especially the Zogby Interactive survey, which, inexplicably, gets a fair amount of press coverage in each campaign season.