I'm not so sure it's a positive, but the pattern is there. From TPM Reader SS ...

H- makes an interesting point regarding the lack of variance in the polls--namely that each poll with approximately 43% of the voters who have decided, and "undecideds" break overwhelmingly for Clinton. The same was clearly true in Ohio and Pennsylvania, and I believe in a few other states as well.

Do you attribute this "late break undecided" votes to something other than the voters had not made up their minds? Specifically, given the states (strong blue-collar white vote), do you think that at the end of the day that there is either (a) a Wilder effect at play here, or (b) that the Wright controversy makes Obama a little too "risky" a vote for a generally fairly conservative voting base?

If so, this could be viewed as an overall positive for Obama in the General Election, if one presumes that these "undecideds" are more concerned about policy than they are about race or the "riskiness" of voting for Obama. Clinton is, in some ways, a relatively safe alternative for Democrats (her policies are virtually identical to Obama), making it easy to justify not voting for the "other" Democrat.