I've been seeing some complaints about the methodology of this poll, there is one good write-up presented here. Basically what it comes down to is that the New York Times not only cherry-picked from their own data, but they also polled heavily from Conservative voters rather than getting averages based on the demographics of each state.
It makes better headlines than the reality (both candidates are pretty evenly matched in each of the swing states with no clear winner), but this information is pretty much worthless other than hopefully getting more people to vote this year.