Yesterday,
we talked about all of the noise in the recent state-by-state results.
With perhaps a couple of dozen polls coming out each day, from firms
with radically different views about how to model turnout, and in
regions of the country that are each reacting slightly differently to
the post-convention environment, there are bound to be results that
cause some cognitive dissonance.
Underlying all of this,
however, is a high degree of near-term stability in the race. For each of the past
six days, our daily point estimate -- formed from making inferences
based on all the national and state polls that are released on a given
day -- have shown Barack Obama with a national lead of between 2 and 3
points. This is what is represented by the big cluster of dots you see at the right endpoint of our supertracker chart:
The big takeaway, naturally, is not to get too hyped up about any one individual result. Rasmussen has Barack Obama 2 points ahead in North Carolina? Good for him. Gallup had John McCain draw back into a tie in their daily tracker? Good for him.
The polls can actually be more difficult to read when a race is
(relatively) stable than when it's shifting toward one or another
candidate. But you throw everything into a blender, and the race is
pretty much at a point of equilibrium -- which, of course, will
probably be ruined tomorrow if and when we have the opening debate in
Mississippi.
State polls, which I'm just going to touch upon briefly today:
The
best results for John McCain and Barack Obama are the polls in Maine
and North Carolina, respectively. The North Carolina number really
shouldn't be all that shocking. The state had been stuck on about a
3-point McCain lead for months, but with the Obama campaign having
worked the state harder than McCain, it's not surprising that we
finally had a poll that showed him with a lead there. Still, the model
is hedging on considering the state a true toss-up because of that
McCain +20 poll from SurveyUSA, and the McCain +17 from Research 2000,
each of which came out a couple of weeks ago. If SurveyUSA and R2K
resurvey North Carolina and give Obama a better number, he will gain
ground fairly quickly. Until then, caution is warranted. One other
thing to keep in mind -- North Carolina is a big retail banking center, so the candidates' positions to the bailout may get extra scrutiny here.
SurveyUSA
gives Barack Obama a relatively narrow 5-point lead in Maine, making
this the second consecutive poll -- along with Rasmussen -- to show the
state in the mid-single digits. Usually this is the point of the
campaign where the candidates start to consolidate their number of
states, but there's an argument that McCain ought to send Sarah Palin
out there, particularly to ME-2, where her backwoodsy charm could play
well (Maine awards one electoral vote to the winner of each district.
With that said, SurveyUSA revealed nearly identical results between the
first and second districts, and they have usually voted very similarly
in the past. Like many campaign decisions, this one will probably be
dictated by McCain's internal polling.
Finally, Pennsylvania and
Michigan appear to be drifting past one another like slowly-moving
tectonic plates. While Pennsylvania had appeared to be a stronger state
for Obama for most of the cycle, we now have it projecting as an Obama
+4.1, to Michigan's Obama +5.3. If an election were held today, Obama
would almost certainly win both -- but if things tighten again,
Pennsylvania rather than Michigan may be the focal point.
--Nate Silver