Reality Check: NY-23 Poll May Seek to Alter, Not Reflect, Reality

There's a lot of buzz in the conservative blogosphere today about a new poll put out by the Club For Growth which shows Conservative Party candidate Doug Hoffman leading both Democrat Bill Owens and Republican Dede Scozzafava in the special Congressional election in New York's 23rd Congressional District. The poll shows Hoffman at 31 percent, Owens at 27 percent, Scozzafava at 20 percent, and 22 percent undecided.

It's a bit disturbing how credulously the conservative blogs, most of whom are rooting for Hoffman, are taking this poll. Here are few of the concerns that a more critical observer might ask about it:

-- The Club for Growth endorsed Hoffman and just last week threw $300,000 into the race on his behalf.
-- The sample size is tiny (300 people).
-- The pollster that Club for Growth is using, Basswood Research -- I'm sure does perfectly good work for its clients -- but is not that well known** and is therefore not taking too much reputational risk with this poll. (** I'm told that Basswood does in fact do a decent amount of work for Republican candidates like Tom Coburn and Mark Sanford.)
-- The number of undecideds in the poll -- 22 percent -- is unusual for a poll just eight days out from an election and is higher than what the public polls show.
-- The poll was conducted entirely over the weekend. Although Sunday is a fairly good polling day, Saturdays are not.
-- The narrative that Club for Growth constructs around the poll is that Hoffman is taking votes from Scozzafava, but the poll also shows the Democrat Owens polling quite a bit lower than he does in the public polls. It seems unlikely that Owens voters are defecting to Hoffman. Rather, if Scozzafava's support is indeed collapsing, I'd expect Owens to be picking up some of that support in addition to Hoffman.
-- The poll was released at a time when the NRCC, which has endorsed Scozzafava, is defending its position by citing the polling evidence, and so the incentive to put out some contrary evidence to alter the inflection of the media narrative is quite high.
-- The poll shows that 59 percent of so-called likely voters have no opinion (or haven't heard of) Owens, 48 percent have no opinion of Scozzafava, and 56 percent have no opinion of Hoffman.
-- Only 14 percent of the likely voters in this poll are age 40 or under, as compared with about 40 percent in the Research 2000 poll.
-- Previous polls put out by Basswood Research and the Club for Growth in this race featured highly leading question wording, although that does not appear to be the case here.

Are any of these red flags, unto themselves? No, just a lot of yellow. To be clear, this is very probably not a case, a la Strategic Vision, where the numbers were simply fabricated. But there's an awful lot that a pollster can do short of making up numbers -- asking leading questions, applying implausible likely voter models or demographic weightings, selecting an unorthodox sample frame, etc. -- to produce a result that fits its desired narrative.

It would also not be totally shocking if Hoffman won this race (although I think he'll need to get to at least 37-38 percent of the vote, which is about the minimum that I'd expect Owens to get). But this poll would not give me particularly more confidence in that outcome -- except to the extent that it alters rather than reflects reality. Likewise, those blogs that report on this poll without addressing some of its shortcomings are more interested in advocacy than reporting.

Note: I originally was not aware that the questionnaire were in fact available for this poll, which I've subsequently been alerted to; they answer some questions while raising some fresh ones.