Perhaps the only disturbing news for Obama in the survey is that most Clinton voters (56%) say they are not likely to vote for the Illinois Senator in the general election against John McCain. A month ago, 45% of Clinton voters said they were not likely to vote for Obama against McCain.

This sort of thing makes me wonder how much to trust these polling people. Surely we can assume that Obama’s increased support comes only from those who are likely to vote for him. A month ago, 40% would have voted for Clinton; now, 33% would. 45% of 40%, a month ago, is 18%. 56% of 33%, today, is 18.4%: practically the same. Perhaps the fact that it’s stationary is cause for some concern, but it’s hardly surprising.

If the people at Rasmussen don’t understand that, what else have they messed up?