try this to begin with
A nonpartisan statistical approach to Rasmussen data
My statement was to relate the perception that people within the political consultant industry have about Rasmussen. Nobody - on any side of the political wars - NOBODY can say if a poll taken far out from an election date is accurate or not accurate. That is an impossibility. So you are asking for evidence which is impossible for anyone to get regardless of what side of the issue they are on. It is impossible to gather any verifiable data to show that a poll issued two months or two weeks before an election is wrong. As you well know, the test of a polls accuracy is how it does compared to the actual election results. However, when somebody like Rasmussen is noticeably different by the vast majority of polls and that difference is almost always in the favor of the conservative or Republican candidate or position, it makes a statement. And within the world of political consultants, Rasmussen is known as
"in the bag" for the GOP and an organization which will carefully mold its methodologies and skew its results accordingly.
this should help on the perception issue that I talked about
The Rasmussen Problem | The New Republic
this article has an excellent chart which clearly shows the outlier aspect of Rasmussen
Political Polls, Data Analysis and Outliers: Discuss | The Big Picture
Of course, some will tell us that Rasmussen could have been correct and every other one of those on that chart could have been wrong. Which reminds me of the proud mother at the parade who noticed that her son in the marching band was the only one not in synch marching with the others as he went by as a few giggled noticing his error. Not to be discouraged, the proud mother puffed out her ample bosom and loudly pontificated to the crowd
"everybody is out of step but my Johnnie!"