I had a conversation with Nate Silver in Pittsburgh last year about Rasmussen Polling. It wasn’t an easy conversation because it took place during the Netroots Nation afterparty in a noisy bar. But I basically implored him to look into their polling practices because it seemed obvious to me that they were creating dishonest product and lots of it, and that it was skewing polling averages on issues and approval ratings that were hurting Democrats in virtually every news cycle.
Not long after our conversation, Silver did expose a Republican leaning pollster, but it turned out to be Strategic Vision instead of Rasmussen. I has happy to see the statistical master take them down, but I was frustrated that he didn’t seem able to nail Rasmussen, who I felt and still feel are far more pernicious and influential in distorting the debate in this country.
So, I am extremely happy to see Silver has started asking tough questions of Rasmussen. In essence, he can’t say for certain that Rasmussen is wrong, but he can expose their house effect. Here’s the thing to remember. In order for pollsters to have credibility and attract clients and have influence, they absolutely must do a good job of predicting the outcomes of elections. But their performance is judged by their last poll in the field. They can skew results in a Republican direction in the weeks and months before an election to help candidates look viable and raise money (and do the opposite to the Democrat) without paying any price for it. They can make the Democrats look more unpopular than they really are and make proposed legislation look like it is loathed by the public. They can do this because there is never an accountability moment on those types of questions. And that is what I firmly believe that Rasmussen is doing. It’s not a bug, but a feature of their polling that it is about 6% more favorable to Republicans than the average of the rest of the pollsters. And I expect that house effect to slowly dwindle down to almost nothing by election day because they want to be as accurate as possible in predicting who will win. After all, on election day, we have winners and losers, and we can judge the pollsters strictly by whether they made the right predictions. We can’t do that on the president’s or health care’s popularity.
