Gallup is making news because their most recent poll has McCain cutting Obama’s lead to two. Good news. And I’m happy about it.
But in this recent election cycle, Gallup has disgraced itself because they have all but openly admitted that they have absolutely no idea what’s going on. Instead of just releasing one poll and basing their reputation on its accuracy, Gallup is issuing two polls with two completely different turnout models.
They’re issuing a traditional model and an “expanded” version (which is kind of like New Coke, I guess.) You see, Gallup is saying that they don’t know how many people are going to turn out in this election so in the traditional model they’re expecting a record turnout of about 60 percent of registered voters to vote in this election. That would be a record. The “expanded” model, however, is based on something like 2/3 of voters turning out. This is the ‘we love Obama’ model that says Obama voters are going to come out in not just record numbers but they’re going to shatter records.
Maybe the expanded version is right but isn’t it the job of a polling company to figure out who’s coming out to vote and who isn’t? I mean, that is their job.
But here’s the result of issuing two polls. Today, according to Gallup’s traditional model, McCain is down two. Good news, like I said. But according to the expanded model McCain’s down seven. And there’s a margin of error of three on each. So if McCain wins by one Gallup can say they were right on because it was within their margin of error. And if Obama wins by ten, that too will be in the margin of error.
If I want to know how the race is going within an 11 point estimation I can pretty much talk to anybody. It’s a little silly for Gallup to be no more precise than regular folks like me who are just guessing. Gallup should just rename themselves the CYA pollsters. No matter what happens they’re covered.