Scott Rasmussen’s automated polls have come under increasing criticism from Democrats (see a recent defense in the Wall Street Journal). His Colorado polls published since September of last year have shown Republicans in the lead for governor and senate.
His latest Colorado Senate poll showed Republican frontrunner Jane Norton up 12 points over Democrat Michael Bennet. Norton leads Andrew Romanoff by the same amount.
Bennet rapidly countered by releasing his pollster’s latest data only showing him three points behind Norton. Also, a recent Daily Kos poll (liberal political website) shows Bennet ahead by one point.
The usual explanations for different polling results involve a long list of possible factors, including simple statistical variation. Most discussion at this point in campaigns focuses on sample and weighting techniques. Rasmussen’s trademark approach, preferred by conservative blogs and talk show hosts, uses sampling and weighting techniques producing a likely voter model designed for an off-year election, like 2010. The approach is especially restrictive, and includes more Republicans than more broadly drawn samples of registered voters. Arguably, the approach may offer a better preview of November 2010 than a less restrictive approach. However, it also is not a description of current voters’ views, but a subset of voters who more dependably turn out.
It may be more accurate to describe the current senate race as closely balanced between the Republican and Democratic frontrunners.