At TNR, Nate Cohen issues an important quadrennial warning about early polls that are said to tell us something significant about the 2014 elections: polls without likely voter screens are going to include a lot more Democratic-leaning voters than the actual midterm electorate.
In both 2010 and 2006, polls of registered voters significantly overestimated the Democrats. There’s a good reason: non-white and young voters, who disproportionately support Democratic candidates, are also disproportionately likely to stay home during low turnout, midterm elections.
In 2010, the application of likely voter screens turned a tight race into a Republican rout. Pew Research’s 1 point Democratic edge among registered voters, for instance, became a 6 point Republican lead among likely voters. Perhaps turnout will be a little better for Democrats in 2014, but it seems unlikely that Democrats will completely defy history and retain the entirety of their advantage among likely voters.
Yeah, we get it. But that’s why the early evidence of a pro-Democratic trend among older voters is so interesting: these are voters who are more likely to survive a likely voter screen and actually turn out in 2014. The disparity in propensity to vote among different elements of the electorate is a relatively stable phenomenon. Their attachment to one political party can change, though, and if that happens with old white folks the GOP’s assumed big advantage in midterms could fade significantly, and with it the partisan gap between RV and LV polls.
