Conservative Arithmetic

It appears that in the conservasphere Rasmussen polling is called “Rasmussen the most accurate pollster.”  In fact Rasmussen had the best performance in 2008.  This was widely noted.

However Rasmussen performed terribly in 2010. 

If one focused solely on the final poll issued by Rasmussen Reports or Pulse Opinion Research in each state — rather than including all polls within the three-week interval — it would not have made much difference. Their average error would be 5.7 points rather than 5.8, and their average bias 3.8 points rather than 3.9.

So I googled [ rasmussen “most accurate pollster” -2008 ] for the claim without the qualification which makes it accurate.  The results were impressive with a huge number of sites confidently claiming that Rasmussen is (not was is) the most accurate pollster.  To be honest, I haven’t clicked the links (I notice that the text around one refers to the President as “Obumbler”).  To me this is a glance into conservatives’ separate reality.

Somehow the new data from 2010 is irrelevant.  Once a claim has been accepted as fact, it becomes immortal and invulnerable to new data.

There is a simple explanation for why Rasmussen used to be accurate and later had a huge Republican bias. They don’t call cell phones (they are a robopoller and not allowed to call cell phones).  Back in the good old days of 2008 when almost everyone had a land line, this wasn’t a huge problem.  Now it is.  Clearly in 2010 they didn’t remove the bias from not polling cell phone only households.  As far as I understand it, they weight using a 3 month average of Rasmussen polls.  This removes noise but does nothing to bias.  In any case the weird kids these days who don’t have normal phones may be different from well me not just because they tend to be young and poor but also because they … well I just don’t understand them and I think Scott Rasmussen doesn’t have a clue either.

So there is a plausible story for why Rasmussen might have been accurate in the past then became inaccurate.  This hypothesis is overwhelmingly supported by data in the public domain.  And it hasn’t penetrated the conservabubble at all.