The bias might narrow slightly as more votes are counted; late-counted votes tend to be Democratic in most states. - Nate Silver, 538, "Senate Polls Had A Significant Democratic Skew"
Does the data bear that assertion? (this is coming from Nate Silver. Of course it does :)
If so, what are the proposed plausible explanations? I'd prefer explanations that are backed up with data, or at least are testable/falsifyable.