Robert Waldmann

The Rasmussen polling agency has two reputations. They have a good record predicting the vote in elections (which means especially primary elections give schduling and attention and stuff). On general public opinion, they ask questions which sure sound like Republican push polls. The theories are that they want to become Republican’s favorite pollster and/or that Scott Rasmussen allows his Republican preferences to guide his polling decisions.

I have the impression that Rasmussen generic house ballot polling should be classified with general opinion polling and not other forecasts of voting.

Look at this table from pollster.com

All but four of 52 polls by pollsters other than Rasmussen show the Democrats ahead — more people declare their intention to vote for a Democrat for the House of Representatives in 2010 than declare for the Republicans. All but one show the Democrats ahead and one of the three is a partisan onmessage/RNC poll which shows a 36% to 36% tie.

In contrast, the last 16 Rasmussen polls in a row show the Republicans ahead of the Democrats. Overall other than Rasmussen has 48 D, 2 R 2 tied and Rasmussen has 21 D 20 R and 4 tied. This difference is uhm significant.

Partly Rasmussen polls likely voters. 4 of 6 non Rasmussen likely voter polls are Democratic partisan polls (all show gthe Democrats ahead) the remaining two show ties.

In the 11 non Rasmussen polls which include data from September and/or October the Democrats lead by an average of 7.09 % in the 6 Rasmussen polls the republicas lead by an average of 3.33 %

The perception that the gap is narrow is entirely due to Rasmussen.

Notably, the average congressional voting intention poll is hard to reconcile with exceedingly low approval ratings for Republicans and Republicans in Congress. It has already been widely noted that the average of generic house ballot polls (which is about half Rasmussen) is anomalous given answers on other questions.

I note that pollsters are evaluated using the last few polls before an election. The hypothesis that Rasmussen plays it straight, when they know they will face a moment of truth, and slants Republican, when they know they won’t, fits not just issues polling but also the generic house ballot more than a year before an election polling.

update: Grammer cleaned up a little.