May 22, 2005

MEG KREIKEMEIER WRITES IN THE CHICAGO TRIBUNE on the uses and abuses of polls:

In early October 2004, Newsweek released a poll immediately after the first presidential debate, which showed a dramatic shift in public opinion in favor of John Kerry.

Did Kerry narrow the gap with his debate performance? Was he really the closer that many in the media had suggested he was? Newsweek was basing its headline "The Race is On" and accompanying story on a comparison between its two most recent polls. The problem, though, was that the polling data was inconsistent.

The October Newsweek poll sampled more Democrats than it did Republicans.

And the first poll, conducted in September 2004, sampled more Republicans than Democrats, not at all reflective of the historical composition of registered voters. . . .

Given the swing in demographics between the two Newsweek polls, of course Kerry saw improvement in his results. In fact, if he hadn't he would have been in deep trouble.

And while President Bush's support among Republicans eroded a bit between the polls, his support among Democrats actually increased. Kerry's support among Republicans went up slightly, and his support among Democrats remained flat.

So why the breathless headlines?

Why did the news media report the data without first thoroughly reviewing it?

Why did the change in poll results pique the curiosity of a stay-at-home mom like me but not the much-ballyhooed investigative instincts of the reporters covering the election?

Read the whole thing. And congratulations to the Tribune for addressing the issue.