So it's a single digit race between the two likely nominees... don't screw up Hillary.
High single-digit difference. For the Presidential election, anything over 4% is a comfortable lead, at least for the time. 4% is the margin of error in most polls.
Most likely there is some statistical correlation in binary races between the margin of popular vote and the electoral vote. Of course one would expect a similarity between this election
Roosevelt 53.4 (432 EV) - Dewey 45.9 (99 EV)
and this one:
Obama 52.9 (365 EV) - McCain 45.6 (173 EV)
The first is seen as a big landslide and the second isn't. OK, FDR had the South locked up and the Republicans had the South mostly locked up.
Another good analogue for 2008 might be 1908:
Taft 51.6 (321) - Bryan 43.0 (162)
Obama and Taft both did well outside the South, but very badly in the South. Taft generally won close elections Up North, except in Nebraska (Bryan's home state) and some western states in which mining still figured an important part of the economy.
(That would be good for an overlay. You guessed it -- I am doing that one next!)