Those results were pretty close to the polls thus disproving your above hypothesis if you wish to rely on that.
do you mean, close to the polls right before election day, or close to the polls two months out?
Largely the case for NJ and VA, not so for MA-SEN. Still, the fact that come-from-behind victories
can happen does not make them
likely to happen. O'Donnell could win with a scandal on Coons' behalf or a much larger than expected GOP wave, but neither would "prove" that nominating O'Donnell was correct, because neither was supported by the evidence at the time of the primary.