Today's "Good News about the Presidential Polling" post

[womzilla]

Over the last few days, several news venues have announced that Bush had made some small but measurable gains in nationwide opinion polling. An article from Thursday on The Emerging Democratic Majority Weblog gives some reason to believe that this improvement is relatively unimportant. In the "purple states"--the states which, realistically, could go either for Bush or for Kerry, Bush's position has either not improved or has actually worsened since the end of March, and Kerry's position has either stayed stable or improved.

And now check out these just-released findings from the same ABC News poll that contributed to Democrats' anguish about Bush being ahead. According to data in The Hotline... Kerry is ahead of Bush by 4 points in the battleground states (50-46). He's even ahead of Bush by 2 points in these states with Nader thrown into the mix and drawing a ridiculous 7 percent.

Note also that Bush's approval rating in the battleground states is 49 percent, 2 points under his national rating and that his approval rating on the economy in these states is just 41 percent, 3 points under his national rating.

Interestingly, if you look closely at recent Gallup poll results, there are also signs of poor recent Bush performance in battleground states (or, as they call them "purple states"). Their latest poll had Bush ahead overall among likely voters by 5 points. But he is only tied with Kerry in the purple states. Moreover, that represents a 6 point decline for Bush in the purple states compared to Gallup's March 26-28 survey.

One must be cautious about these data, of course, because of sample size and other problems (though note that the ABC News battleground states sample is probably 300 or so, which is a pretty decent size). But they do lead me to a hypothesis about Bush's recent improved performance in trial heat questions. Instead of getting more votes where he needs them--in the battleground states--his posturing is mostly driving up his support in the hardcore red states, where he doesn't need them. If that's true, Democrats should definitely not be intimidated by recent poll results. Bush is preaching to the converted--which can make him look better in a national poll--but he's not winning many new converts where it counts.

Bush is pumping a lot of money into advertising already--he's spent an estimated $50 million in the six weeks since Kerry effectively clinched the nomination. But it's not doing a lot of good in the battleground states.

It would be nice if the wave of terrible news over the last few weeks would make Bush's ardent supporters less supporting. But that's not going to happen. About 40% of the American electorate is going to vote for Bush under almost any likely scenario, just as about 40% is going to vote for Kerry. It's the remaining 20% who will decide the election, and the evidence is that they're leaning Kerryward in the places where it matters.

It's still six and a half months.

Comments

[dougo.livejournal.com]

About 40% of the American electorate is going to vote for Bush under almost any likely scenario, just as about 40% is going to vote for Kerry.

Aren't these numbers more like 20%? Does "electorate" mean "people eligible to vote" or "people who usually vote"? Maybe that's being pedantic, but I wonder if Kerry should concentrate on the 50% who don't usually vote rather than the 10% who do but are undecided. But I'm guessing that's a lot harder (and Kerry's probably not the guy to do it). (Howard Stern is doing his part, (http://www.salon.com/news/feature/2004/03/12/stern/) but I doubt much of his audience is in the purple states...)

[womzilla.livejournal.com]

I was using "electorate" to mean "likely voters", which was indeed a bit sloppy of me. The actual number of eligible voters who vote is 55-60%, by the way, and has been consistent at that level for several elections. Most people who are eligible to vote who don't vote aren't slavering for a chance to vote, either; they feel alienated from the process, for a variety of reasons including active suppression (in the case of many minorities), the pervasive sense that "the system" is broken, or because they've been convinced there's no difference between Republicans and Democrats so there's no point in voting because nothing makes a difference.

[lavendertook.livejournal.com]

OK--here's some thoughts on polls. It just doesn't seem right to me that Bush's percent should be up after all the news that has been coming out lately--perhaps steady, but it makes no sense that he's gained. And going with the idea that these polls are legit *grumbles*, what makes a poll legit is its being representative of a wide cross-section of the population, which would include people who do not normally vote and probably won't turn out for the Nov election. So that really skews the polls from reflecting who will really be voting is that in order to have legitimacy, the polls have to try to be more representative than the voter turn-out will be.

And perhaps what's skewing them toward Bush is the percentage of those nonvoters who are apathetic or do not like to make political waves--if cornered with a poll they're going to affirm the status quo. However, they won't be at the polling boths in Nov. Thus representative polls will always favor the status quo--and in effect helps buoy it. Hence people who are likely to vote right now may really be favoring Kerry even if the people polled don't reflect this, but the poll itself favoring Bush serves to hurt Kerry's chances by making it look like he's not in the lead. Hence, my conclusion that polls suck.

But this would be assuming that the kind of nonvoter I described is more numerous than the nonvoter who is too stressed trying to survive to vote and the one who has no faith in the possibility that they can make a difference or would be counted--and I don't know if that is a safe assumption.

What say you?

[womzilla.livejournal.com]

Polls often distinguish between "likely voters" and "potential voters", for just the reasons you outline. "Potential voter" polls are cheaper to conduct, because you don't have to call as many people; any citizen over the age of 18 can count as "potential", while a "likely voter" is someone who self-reports as a likely voter, or who has voted in the last few elections, or fits some other criteria. The broader polls--the nationwide polls--tend to be "potential voter" precisely because they're less targeted.

And I agree with you that there's a strong status quo tendency, especially on "national security" issues. I'm appalled to learn that even a large percentage of New Yorkers think that Bush is better than Kerry on "national security", which amazes me considering how much Bush has done to actively harm the security of the United States over the last four years, including insulting foreign leaders while campaigning in 2000.