You have written a few articles saying that up to 1/3 of Obama or Clinton supporters will not support the other if they become the nominee. Yet every major poll - contrasting a head to head match up with Obama-McCain or Clinton-McCain, has both Obama and Clinton winning. Why don't you include such central data or (considerations) when making these assumptions. Am I the only American who notices this glaring contradiction? And please don't tell me you are only polling "the non divided Obama and Clinton voters."

The drum roll....

I've wondered about that, too. And not just in this election.

First, I think that the national polls typically sample a relatively small group of people. Some of the people who say they won't vote for Clinton or Obama are included, of course, but it would not be unusual for them to be either under- or over-represented in any given poll. The polls don't stratify their samples bassed on things like that.

Second, national polls are irrelevant in the general election. The polls showing that Obama supporter won't vote for Clinton (or vice-versa) are going to be mostly made up of ppl in states that won't matter in the general election. E.g., the voters in NY, TX, and CA have 120 electoral votes (out of the 270 needed to win). NY is almost certainly going to go to the Dems no matter how many Obama (or Clinton) supporters sit out the vote or vote for McCain. CA will probably go to the Dems no matter what the Obama or Clinton supporters do. Similarly, TX is probably going to to the the Repubs no matter what the Dem voters do. So just looking at these three states, 4/9 of the possible "defections" probably won't matter. And there are many other smaller states where this is true.

Third, I think the polls are getting at somewhat dfferent things. The polls about possible "defectors" say more about how many voters are likely to be available for pick-up by the Repubs. If the RNC can identify and target those Clinton or Obama supporters who say that they won't vote for the other candidate in the general election, especially in states where the vote is likely to be close, they'll have a better chance at picking up those states. They have a good chance of persuading those people to vote for McCain. Even if they don't get them to vote for McCain, they may well be able to turn them off to the whole presidential election. And the RNC is very good at targeting specific voters. That's why the 1/3 figure is important, even if both the Dem candidates are now beating McCain in national polls.