Well Rasmussen’s own Party Affiliation survey for August showed a +4.3% R gap, he never used his own data in his own normalization of the numbers.
In 2006 the dems had a big edge in party affiliation they expanded that gap by ~1.5% in the subsequent election.
In the 2010 midterm the gap in party affiliation was 1.3% R, what has happened to move the needle to the dems by 4%+ ?
Of course he has been getting hammered by Axelrod.
First, polling has never been an exact science, far from it.
Modern polling methods started with Gallup in 1936 just in tine for the Roosevelt-Landon election. Gallup picked the winner and embarrassed the Literary Digest out of business.
Yet even back then, Gallup wasn’t accurate. From 1936-1952 they were mediocre to terrible(especially Dewey-Truman in ‘48). From 1956-1976 they were pretty accurate with very good results in 1960 and 1968. However, from 1980-2012 with the exception of 1984 and 2004, they have been simply crap.