In the light of Monday's Rasmussen poll, which showed Perry with a ten-point lead—one point less than he had in November—one has to wonder: What was the Hutchison campaign up to when they announced on January 12 that their internal polling showed KBH with a two-point lead? (Hutchison 34.9%, Perry 32.9%, Medina 2.7%) I chose not write about this poll for the following reasons: (1) It reeked of a certain commodity found in cow pastures. It was too much of an outlier from all other polls in this race. (2) The KBH campaign did not release the kind of supporting information that might have buttressed the credibility of the poll. If supporting documentation existed, the news that she had taken the lead would have been such a huge boost to her campaign that she would have released everything the campaign could muster to make the poll credible. The campaign's failure to do so strongly suggested that the poll was suspect. (3) As Harvey Kronberg and others have pointed out, the proportion of undecided voters—26%—was highly unusual for a race featuring two well known candidates who between them have been polling around 80% of the vote (81% in the November Rasmussen poll). (4) The Perry campaign had previously released a Mike Baselice poll in December showing Perry ahead by 13 points, 49% to 36%. While this poll likewise did not provide a lot of documentation, it was in line with previous polls.
So, what really happened here? Did the campaign pressure its pollster to provide a favorable poll? Didn't they know that Rasmussen has been polling the race every two months and that their poll would be exposed within days?
There are many ways in which a poll could be tilted to produce a desired result. Kronberg mentioned several in the link I provided above. (He did not suggest that any of these techniques had actually been used):
First, the sampling could have been of one time Republican primary voters. The GOP primary doubled in size last year, so that could be a rich environment. One long standing thesis has been that the larger the turnout, the better Hutchison will do.
Second, the pollster could have simply tested self-identified Republicans, whether or not they ever voted in a primary. This group would likely have more undecideds than traditional primary voters.
Third, [the polling firm] could have used an “informed ballot” which is different from a push poll although the two are often confused. An informed ballot typically tests the candidates’ head to head numbers on the front end, but then asks a series of questions that challenge the respondents perceptions about the candidate. The informed ballot is legitimate in that it tests questions that can move voters.
However the poll arrived at its result, it seems to me that only two possibilities exist for why it produced the result it did. Either it was very bad work, or it was very good work, if you get what I mean.
- 1 week