Today’s must-read for serious political junkies is a post by pollster.com’s Mark Blumenthal about factors that might be temporarily inflating Republican advantages in polls. It’s a fine refresher course on the various objective and subjective factors used in screening for likelihood to vote, and suggests that “attentiveness” to an election campaign is one factor that tends to decline in importance just prior to Election Day.
If we could see everything about how pollsters model likely voters and all of their data, my educated guess is that the more these models rely on self-reported attention paid to the campaign, the more they tend to produce outsized Republican leads. My sense is that polls that have depended more on other means to model the likely electorate, including the internal campaign polls that also make use of previous vote history gleaned from voter files, have produced results that have been more consistent over time.
So maybe those Gallup and Pew generic ballot polls that have so badly frightened Democrats over the last few weeks are a bit exaggerated after all.