The New Independents Explain Polling Paradox

Rush Limbaugh has mentioned the paradox of the overall statistical tie (in most polls) for the Presidential race  even though Mitt Romney is leading among independents.   Limbaugh suggests that this is evidence that most pollsters are oversampling Democrats relative to Republicans.  His reasoning is that if Republicans and Democrats existed in about equal numbers, and independents lean Romney, then Romney ought to be winning.   Only two polls show this result.  Most show a dead heat.

Well, I have an alternative explanation that leads to a disturbing conclusion for Romney supporters.    I don't think the independents now are all the same group that were independents in 2008.   I think a lot of people who considered themselves "Republican" in 2008 now call themselves "Independent."    They are independents now not because the GOP is too conservative, but because it is too much like the big-government Democrats.    They will hold their nose and support Romney, but they don't imagine he will implement the changes they want.   I am doing a lot of door to door campaigning, and this is what I am finding.

Why does this bode ill for Romney?  Well, it would mean the polls showing him ahead by seven are wrong and the ones showing that this race could go either way are correct.   Imagine the electorate in three roughly even sized groups- Democrats, Republicans and Independents.    If that represents the sampling they use in the polls, then Romney winning the R's and the I's and Obama winning among the D's would mean that Romney would have the lead (if they were winning their base by the same amount).  

But now imagine that one of every five Republicans was so disgusted by the corporate-financed tilt away from limited government in the Republican party that they started IDing as independents.   In that case, instead of being 33% of the total pool, Republicans would be only 27%.  Independents would be under-represented, 33% when they were actually closer to 39%.   That pool would be shown as moving right, but what really happened was a bunch of existing right-leaning votes simply changed designations.   That 33% did not get more conservative, rather 6% of the right-leaning voters changed their label.

If the pollsters still think the Republicans are 33% of the vote when they are only 27%, then instead of the Democrats being over-sampled, it would be the Republicans who were over sampled.     If this is correct then the polls that say the election is very close would be correct.   They may be correct for the wrong reasons (they may over-sample Democrat turnout rather than catch that more former Republicans are calling themselves independents) but the bottom line is the same- a very close election.