In the Right

In last week's survey, Corzine was holding 79 percent of Democrats, but just 30 percent of Independents, and just 7 percent of Republicans; Christie was holding 79 percent of Republicans, 8 percent of Democrats, and 45 percent of Independents; and Independent candidate Chris Daggett was drawing the support of 9 percent of Republicans, 10 percent of Democrats, and 20 percent of Independents.

In this week's survey, Corzine is holding the support of 77 percent of Democrats, 32 percent of Independents, and 10 percent of Republicans; Christie is holding 78 percent of Republicans, 6 percent of Democrats, and 47 percent of Independents; Daggett is drawing the support of 9 percent of Republicans, 12 percent of Democrats, and 17 percent of Independents.

So if Corzine and Christie are holding about the same percentage of their base vote as they were last week, and Christie still maintains a 15-point advantage among Independents, what explains the 7-point swing?

A look inside the poll reveals what's behind the top-line ballot movement -- the composition of the survey sample has shifted significantly between the two surveys.

Where last week's survey sample was 40 percent Democrat, 29 percent Independent, and just 25 percent Republican, this week's survey sample is just 34 percent Democrat, 30 percent Republican, and 29 percent Independent.

That's a 6-point drop for the Democrats, and a 5-point gain for the Republicans. The result is that the Democratic advantage in this survey drops from 15 points to just 4.

With a typical New Jersey gubernatorial turnout of roughly 2.3 million voters, one percent equals 23,000 voters. So Democratic turnout dropping from 40 percent of the electorate to 34 percent of the electorate would mean a loss of 138,000 Democratic voters at the polls -- and with Democrats breaking 77 percent for Corzine, 6 percent for Christie, and 12 percent for Daggett, that would mean 106,260 fewer Democratic votes for Corzine, 8,280 fewer Democratic votes for Christie, and 16,560 fewer Democratic votes for Christie -- in all, a loss of a 97,980 vote advantage for Corzine.

Meanwhile, the 5-point gain for the GOP translates to 115,000 more Republican voters at the polls than would have been the case if the turnout model looked like last week's. With Republicans breaking 77 percent for Christie to 10 percent for Corzine to 9 percent for Daggett, that would mean 88,550 more Republican votes for Christie, 11,500 more Republican votes for Corzine, and 10,350 more Republican votes for Daggett -- in all, a gain of a 77,050 advantage for Christie.

Add Corzine's loss to Christie's gain, and the difference in the survey samples between last week's survey and this week's survey adds up to a net swing of 175,030 votes from Corzine to Christie.

Now, does that mean I'm predicting a 175,000-vote victory for Christie tomorrow night? No, it doesn't.

I'm simply using the two survey samples from the same respected pollster, both from surveys conducted within one week of each other, to demonstrate a point we first made here quite some time ago -- a survey's predictive value is only as good as the construction of its sample.

If tomorrow's electorate has a relatively small, 4-point advantage for the Democrats, it looks like Christie will win. If, on the other hand, tomorrow's electorate looks more like last week's suggestion, and there's a 15-point Democrat advantage, it may be possible for Corzine to squeak out a victory.

But based on what I've seen on the ground over the last few days -- light turnouts for Obama in both Camden and Newark, a virtually non-existent ground game in Hudson, no excitement for Corzine anywhere -- I find it hard to believe this year's electorate would look like last year's.

The enthusiasm in the final 96 hours, and the polls, all seem to be moving Christie's way.

In this race where it really is "all about turnout," it's up to Christie to bring it home.

Follow me on Twitter!