It is difficult to gauge the immediate reaction to yesterday's
financial meltdown in the polling, particularly as several other
aspects of the campaign, such as last Friday's debate, are also still
echoing onto the numbers. Barack Obama lost 2 points in the Gallup Tracker, bringing his lead back to a +6, while gaining ground in the other national trackers.
Either
way, Barack Obama remains in a commanding position nationally -- but it
is less clear how much of his advantage is translating into particular
swing states.
Firstly, we have a trio polls from ARG,
each of which show Obama a couple of points behind in some critical
swing states. Oddly, ARG's national poll has Obama doing just fine,
moving into a 4-point advantage nationally. The thing about ARG polls
is not necessarily that they're "bad", but that their behavior tends to
be erratic, jutting around in ways that aren't particularly correlated
with one another or with other polling firms. So, while I'm happy to
include ARG's numbers in our model, I'm not going to invest a
particular amount of time in trying to decode them.
On the other hand, the SurveyUSA
polls in Indiana and Ohio may well mean something, as they show Obama
gaining ground, but remaining slightly behind in those states. The
reason the model still projects Obama with a slight advantage in Ohio,
when the two most recent polls (SurveyUSA and Rasmussen) gave McCain a
1-point lead, is because those polls (especially Rasmussen; less so
SurveyUSA) have tended to be toward the lower end of Obama's projected
performance in Ohio. And those polls have
shown gains for him -- just not enough to give him a lead. The model
assumes that when pollsters like Quinnipiac or CNN or National Journal,
which had been friendlier to Obama before (shown him tied or ahead in
Ohio), refresh their data, they too will gains for Obama, and will
probably show him several points ahead.
If the model's guess is
wrong, it is wrong, and it will quickly correct itself. But a lot of
times, people think they are seeing "movement" in the polls when all
that they're really seeing is the mix of pollsters in a particular
state change. Obama has been gaining ground everywhere lately; there is
no particular reason to conclude that he isn't
gaining ground in Ohio, when his numbers have been improving in
neighboring states like Michigan, Pennsylvania and even Kentucky. The
model does not assume, by the
way, that the movement across different states is uniform; instead, it
searches for demographic patterns that might explain it. For the time
being, none of those demographic clues would exclude Ohio from the fun.
With all that said, Florida, where PPP today shows Obama ahead by 3 points, may well be on the verge of surpassing Ohio as a potential tipping point state.
--Nate Silver