Poll Suggests Clinton Backers Still Wary of Obama; I Suggest Poll is Stupid

A new AP poll out yesterday suggests that Obama is still having a lot of trouble convincing former Clinton backers to come around and support him. Specifically, the poll found that only 58% of Clinton supporters have made the switch to Obama. Boy, that seems like real bad news for Obama, right?

I don’t think so. For two reasons. First, this AP poll is almost certainly understating the percentage. This poll uses a somewhat strange methodology. Instead of taking a new sample every time, they took one sample of about 2,000 people back in November of 2007 and then they go back to the same sample every few months to gauge how their opinions have shifted. That is a serious mistake, in my opinion. The reason that pollsters take new samples each time is because the make-up of the electorate, both demographically and in terms of opinions, is not static. Indeed, it is extraordinarily unlikely that the electorate looks the same today as it did 10 months ago. Think about all that’s happened since November 2007. For one thing, there has been a huge addition to the pool of possible voters, in the form of new registrations. A sample collected 10 months ago doesn’t, by definition, include any of those new registered voters (including those people who registered so they could vote for Clinton in the primaries). If a pollster came to us and said, “I took this sample in 1998, it should still be good as a representative sample today,” we’d laugh them right out of the room. Of course, there is a difference between 10 years and 10 months, but the sample is stale none-the-less.

Furthermore, the people in this stale sample have, of course, been asked these same questions multiple times by the same survey organization (that’s the whole idea with this ‘methodology’). It seems to me that the strong psychological imperative to appear “consistent” might be at play here much more than out in the regular world.

So, there are strong reasons to discount this poll out of hand. But let’s assume, for the moment, that the poll is generally accurate. Does it mean that Obama is doomed? Actually, I’d argue the opposite is true. According to this poll, Obama is getting 60% of former Clinton supporters, while John McCain is getting about 25%. That means that, assuming Clinton supporters make up about half of all Democrats, Obama is currently ahead in the national polls even though he only has the support of 80% of Democrats. John McCain is, apparently, winning about 12.5% of Democrats, so that leaves 7.5% of Democrats who are still undecided. If Democrats make up about 35% of the electorate, then that would mean that among the approximately 5% of the whole electorate that currently tell pollsters they are undecided, according to this poll, about half are actually Clinton Democrats. Essentially, this poll, if correct, means that most of the undecideds out there lean heavily towards the Democrats, and are likely to either end up voting for Obama or staying home. Sure there’s some resentment there, but if they haven’t moved to McCain yet, they probably never will, and most of them will vote their party when they get to the ballot box.

So that poll would be great news for Obama, if it weren’t terribly flawed.