We should always recognize the error inherent in polling, whether exit polling or any other kind! New Jersey Network has called the governor's race for Christie. He is up about 100,000 votes as I write this, about 5.7 points with 74% of the vote reported.
Clearly either the turnout estimate in the exit polling is wrong -- the most likely reason -- or the reported breakdown of the vote by partisans is wrong.
I'm guessing there are more Republicans in the mix than the exit poll showed. Either they didn't sample some areas that had (relatively) massive Republican turnout or Republicans were less likely to talk to exit pollsters. Or something else... :)
For example, if Republicans made up 33% of all voters (compared to the exit poll report of 31%) and Democrats were 40% (instead of the 43% in the exit poll), Christie gets a 3 point win in the calculations.
It simply looks like Republicans turned out - especially in Ocean and Monmouth Counties, and Democrats just did not keep up the pace in comparison.
So again, polling gives us insights in how something happened, but in the end we have to count the votes!