Last night, I noted how NBC Newswas swimming against historical trends in pushing a poll with a deeply partisan skew in its sample. The poll, produced in conjunction with The Wall Street Journal, had a D+11 sample, i.e. 46% of the sample identified as Democrat, 35% identified as Republican. In 2008, a very big year for Democrats, the electorate was D+7, 39% Democrat, 32% Republican. In fact, you have to go back nearly thirty years to find an election with such a partisan skew.
This morning, on MSNBC's Morning Joe, NBC political director Chuck Todd admitted their poll had a skewed sample (around 3:50 mark) and stated that, if this poll were weighted similarly to their last poll, the race would have been unchanged, instead of Obama extending his lead over Romney to 6 points. (Of course, this invites the question as to why they didn't use the same weighting in this sample.)
Credit to Mr. Todd, then. Of course, when he previewed the polls results on a local NBC affiliate yesterday, he didn't mention anything about the skewed sample. And, of course, his acknowledgement that the poll oversampled Democrats by a wide margin, hasn't stopped him from opining about lots of other information "found" in the poll.
Note to Mr. Todd: A highly partisan skew doesn't just affect the head-to-head match up. It affects pretty much every other finding in the poll. Obama and Romney's likability? Who is better to handle particular issues? Every single "result" will be skewed due to the flawed sample used in the poll. There is simply almost no real information to be accurately "learned" from this poll.