For the first time since shortly after clinching the Democratic
nomination, we now have Barack Obama as less than a 60 percent favorite
to win the election. Our simulations presently project Obama to win the
election 58.4 percent of the time, with McCain winning the remaining
41.6 percent.
The main culprit for the decline are the new numbers out of Ohio, where Rasmussen
shows John McCain jumping into a 10-point lead. We have already
discussed this particular poll at length. Are the changes caused by
differences in measuring party identification? No, not really.
Rasmussen assumes a slightly redder electorate than other pollsters,
but Obama's numbers had declined among Democrats, Republicans, and
independents alike. Could there be problems related to the sampling of
young voters in this survey, who went surprisingly strongly for McCain?
That is a more viable explanation. But it still would not account for the entirety of the decline.
There is also new polling out from American Research Group,
which has Florida and New Hampshire moving in John McCain's direction.
In Florida, Obama now trails by 2 after having led by 5 points, and in
New Hampshire, he leads by 2 after having led by 12.
The sky is not falling for Obama in Colorado,
where Rasmussen has him retaining a 3-point lead (the lead is 7 points
before leaners are factored in). This is consistent with the polling in
Colorado throughout the election; Obama's leads have generally been in
the small single digits, but he has almost always held one.
When
we throw everything into the 538 blender, what we find is Ohio rating
strictly as a toss-up. The fact that Ohio appears to be polling a point
or two behind the national numbers for Obama rather than a point or two
ahead has significant implications across the map. Viable 'Plan B'
states like Colorado, Virginia, Iowa, New Mexico, Nevada and perhaps
Montana become even more important, as they, rather than Ohio, may now
represent the path of least resistance toward an Obama electoral
victory.
--Nate Silver