And that “Bradley effect”? Didn’t happen.
That’s the word Wednesday from two elections experts, Yale political scientists Donald Green and Alan Gerber.
Analysts — or, if you prefer, hopeless political junkies and wonks obsessed with this year’s presidential election — were closely watching the polls not just to see whether Barack Obama or John McCain would win, but whether we can still trust the accuracy of traditional polls.
While the final margin is still being calculated, Obama appears to have defeated McCain by about 6 percentage points among the national electorate, squarely within the margin of error of the final tracking polls.
“Pollsters actually came out looking OK in this election,” said Green (pictured). “They didn’t have egg on their face the way they did in 1996” when they overstated Bill Clinton’s reelection margin.
Copious megabytes and bottles of ink were devoted during the election to the question of whether the polls were ignoring a “Bradley effect” — secret anti-black sentiment that voters weren’t revealing to pollsters.
“There was originally a hint of that in Virginia, a state that was projected to be for Obama. When more and more results came in, he had a narrow lead … and he won a narrow victory. He was trailing a bit in Missouri [in the polls], and it ended up a dead heat. The Florida results, the Ohio results, the Pennsylvania results, all of these were in line based on what you expected,” within the major polls’ margin of error.
Cell Phones & Robopollsters
Pollsters were under scrutiny as well because of changes in technology. It was suggested, for instance, that the polls may be under-reporting Obama’s support because pollsters survey voters over landlines. They don’t use cellphone numbers. And many people, especially younger people inclined to vote for Obama, have only cellphones now.
At least that was the theory.
In practice, Green said based on the data available so far, the polls appeared to have captured an accurate reflection on how young people voted. It may have turned out that views of young people with only cellphones tended to conform to the views of young people with land lines.
A second trend being watched: the increasing use of automated polling. Outfits like Rasmussen Reports and SurveyUSA used computer generated phone calls asking people to punch in answers, rather than human employees calling voters and typing in their answers. Rasumussen predicted a 52 – 46 percent vote, which is what may well turn out to be the final result.
“It appeared to work well” with results as accurate as those of live polling, Gerber (pictured) observed of the new method. “It’s extremely inexpensive compared to live answers. That will make it a pretty compelling story.”
Both Gerber and Green cautioned that it’s premature to draw definitive conclusions. Votes remain to be reported on the West Coast, and much detailed analysis of voting groups remains to be done.
Donald Green and Alan Gerber will offer a more comprehensive and updated analysis of the 2008 elections at a brunch at Congregation Beth El Keser Israel, 85 Harrison, Sunday. The brunch begins at 9:45; the program starts at 10:15 a.m.