Tuesday, September 07, 2004

DEAD-CAT BOUNCE? I know, I know. If I obsess over the daily polls, I'm going to go nuts, and you're going to stop reading. Still, some results from Gallup bear scrutiny.

Yesterday's USA Today reported that George W. Bush had "widened his lead" over John Kerry to seven points. Susan Page wrote:

As the campaign enters its last eight weeks, a USA TODAY/CNN/Gallup Poll taken Friday through Sunday shows Bush at 52%, Kerry at 45% and independent candidate Ralph Nader at 1% among likely voters. Before the convention, Bush led Kerry by 2 percentage points.

As Page noted, that's among "likely voters." But in her next paragraph, she reported that among "registered voters," Bush led Kerry by just two points, 48 percent to 46 percent, with Nader at four percent. Hmmm. Who are these creatures that USA Today considers likely unregistered voters? Don't you have to be registered to vote? Yes, there are some unregistered voters who will register before November 2. But obviously these are the least motivated of potential voters, and neither candidate should count on many of them stumbling to the polls on Election Day.

For Kerry, it gets better. Late yesterday afternoon Bloomberg moved a story reporting that Gallup now shows the gap between Bush and Kerry among registered voters to be just one point - 49 percent to 48 percent. After five weeks of Kerry-bashing, culminating with Bush's widely praised convention week, the race is still essentially where it's been since March: in a dead heat. At least among those who care enough to be already registered to vote.

Thanks to Bloomberg Radio's Michael Goldman for passing this along.

2 comments:

Anonymous said...

I think you may have got hold of the wrong end of the stick here. As you point out, it's extremely unlikely that an unregistered voter will cast a ballot for either Kerry or Bush in November (although there's still time for people to register). But that implies that these aren't "likely voters."

It's much, much more likely that the "likely voters" in the poll are a subset of registered voters who the pollsters think will actually cast ballots in November. (The question then becomes what additional criteria the pollsters applied to make that determination of likelihood.) The intimation here is that a significant proportion of registered voters won't vote. Unfortunately this has been true in past elections and probably will be true this year. So increasing voter turnout would be good for Kerry - my impression is that it's usually better for the Democratic candidate - but in the absence of weekend elections, a strong civic sense in the population, or mandatory voting, it's proven tough to do.

Anonymous said...

Keep in mind that there are a lof of garbage registrations nowadays with the motor voter laws.

Check this for more on the poll discrepances over the last few days:
http://rasmussenreports.com/Poll%20Differences%20Sept%206.htm