A theory on Colorado and wider...

The poll we're releasing tomorrow shows Barack Obama's approval rating in Arkansas ever so slightly higher than what he received at the ballot box there last fall. At the same time our polling in Colorado last week found him lagging his November vote share by five points. What explains that dichotomy?

Like many things involving Obama, I think it comes down to race.

Our national poll last week found that 88% of whites who voted for Obama approve of the job he's doing- pretty impressive. But 99% of the African Americans who voted for him like how he's doing- in terms of raw data it was 92 approving of him and 1 unsure.

It stands to reason then that regardless of how well Obama did in a state last fall, where his approval rating now falls in relation to his vote share in that state has a lot to do with how large its black population is.

There are two states where we've polled a standard electorate (we're looking at a 2009 electorate in Virginia and New Jersey) and found Obama's approval running four points or more behind his vote share. Both of them- Colorado and West Virginia- have populations that are only 4% African American. That leads me to believe that if we polled a New Hampshire or Iowa- other states favorable to Obama but low on black voters- he might have weaker approval numbers in those places too.

On the other hand we've found Obama with approval ratings significantly better than the percentage he got in November in a couple of deeply red but also heavily black states- Louisiana and Alabama.

I could be wrong, but this is the best data driven theory I can think of for why his numbers are surprisingly poor in Colorado.