My friend A.I. bad mouthed the Rasmussen poll in a recent comment, something I noticed because my friends at Badlands Blue and Madville Times recently did the same. Here is A.I.’s comment:
Does this mean you will stop quoting Rasmussen, create-a-positive-Republican-narrative, robo-call polls as though they are worth more than a pitcher of warm spit?
Here’s Travis at BB:
Rasmussen’s not credible for a couple of reasons. Most notably, he worked for George W. Bush and the Republican National Committee. He’ll also be joining Karl Rove and other conservative big shots as a featured speaker for a National Review post-election cruise in November. So, it’s not surprising that after a thorough analysis of a number of Rasmussen’s polls, Nate Silver at fivethirtyeight.com finds that Rasmussen’s “polls have tended to show substantially more favorable results for Republican candidates” than other surveys.
Here’s how it works…In order to publish as many “polls” as possible that show Republicans in the lead, and therefore convince the electorate that the Democrat doesn’t stand a chance (credit to the Daily Kos for that analysis), Rasmussen does them on the cheap.
That’s an ad hominem, followed by an unsupported accusation involving a contradiction. Lots of professional and respected pollsters have party affiliations, and anyway you can’t discredit a poll by telling us the pollster is biased. You have to show the poll was biased. And surely the Daily KOS is just as biased as Scott Rasmussen, so how can that site confirm Travis’s accusation?
To be sure robopolls are controversial, but it isn’t at all clear why they are unreliable. Are people less likely to answer recorded questions? Are they more or less likely to fudge when talking to a real human being?
Let’s do what Travis did not, and take a look at recent elections and Rasmussen’s record.
Final Rasmussen Poll
|New Jersey Gov
In the Massachusetts special election, Rasmussen’s last poll still showed a Coakley lead. However, the difference between that one and previous R polls clearly showed a strong surge for Brown. R underestimated the surge by bit with a week to go. In the case of the New Jersey Governor’s race Rasmussen gets it about right (a three point win for Christie) while overestimating, perhaps, the third party candidate. Rasmussen’s last poll in Virginia is a bit low on McDonnell’s final tally, but right on Deeds’ number.
In all three cases, Rasmussen’s polls were about as accurate as good polls tend to be. Only Survey USA, I think, did consistently better. In none of these polls did Rasmussen exaggerate Republican support or the gap between the winning Republican and the losing Democrat. The same is true of the 2008 President election, which Rasmussen nailed.
In short, Rasmussen looks to be a very reliable poll in recent elections, and there is no evidence at all in these cases of a Republican bias. But I can see why A.I., Cory Heidelberger, and Travis Dahle all want to believe that the R polls are unbelievable. Rasmussen has Daugaard leading Heidepriem 52% to 36%. We won’t know how accurate this polling is until we can compare the last R poll with November’s result, but gentlemen: isn’t this what South Dakota gubernatorial elections look like?
R’s 52/36 fits comfortably into that series. If anything, Daugaard’s advantage looks a bit low in comparison. South Dakota Democrats haven’t won the state house since 1974. If I were a betting man, I wouldn’t bet on it happening this year.
Finally, there is Rasmussen’s post-primary polling showing Kristi Noem leading Rep. Stephanie Herseth-Sandlin 53% to 41%. That twelve point lead looks surprising to me, even as a post victory bounce. It is hardly outside the range of probability. We live in a largely Republican state. All the polls, even the “legitimate ones,” Travis, are showing a strong advantage in enthusiasm among Republicans. I can’t help but point out that, not too long ago, my friend Mr. Heidelberger was rooting for a Democratic challenge to H-S.
Moreover, other polls show a dramatic shift in support to the GOP, especially among independents. Nate Silver at FiveThirtyEight, whom Travis cites approvingly, backs up the NPR poll.
A new poll from Public Opinion Strategies and GQR for NPR, which shows Democrats performing badly in vulnerable House seats, is making a lot of waves this morning. It certainly contains bad news for Democrats — however — it is the same bad news that was already implied by generic ballot polling.
So far as I can tell, Herseth-Sandlin’s seat has been considered vulnerable by most analysts all year. And then there is the recent Gallup poll showing independent voters favoring Republican candidates by a good ten percent. Considered together with the enthusiasm gap and the Republican advantage in South Dakota, Rasmussen’s last House Race poll was hardly unbelievable.
I am not making any predictions here. My only point is that Scott Rasmussen has a good record and his current polling in South Dakota is supported by a lot of additional evidence. I can understand why my friends on the other side of the aisle want to believe otherwise.