Here we seem to have the answer to the question I posed earlier about 'likely' and 'registered' voters.

According to CNN's story on the new CNN/USAToday Gallup poll ...

The poll, released Monday, found that among likely voters, Kerry was the choice of 52 percent and Bush 44 percent in a two-way matchup, with a margin of error of plus or minus 4 percentage points ...

Among registered voters, Kerry's lead over Bush narrowed from 8 percentage points to 5 points in a two-way race and from 6 points to 2 points in a three-way race.

That is because Democratic voters are indicating they are more likely to vote than the overall electorate -- something that has rarely happened in past elections and may be fueled by the interest in the recent Democratic primaries.

So there you have it. <$Ad$>When the pollsters restrict their count to 'likely' voters Kerry does better.

I'm not enough of a poll maven to give a precise enumeration or history, but that's really uncommon. Restricting down to 'likely' or even 'very likely' voters almost always gets you a more Republican-friendly number.

Notwithstanding the CNN story's reference to interest generated by the primaries, it seems to me that the most straightforward explanation of this is that Democrats and Democratic-leaning independents are more energized, more committed to getting out and voting this year.

Of course, that energy could be a product of the primaries. But I suspect that gets the order of events, at least in part, wrong.

The fact that primaries turned out so well for the Dems -- and I mean this only in the limited sense of not producing a lot of intra-party dissension -- is a product of the energy and unanimity among Democrats over the central and overriding importance of ousting the president.