What poli-sci can't quantify

When Eric Fehrnstrom came out with his indelible Etch-A-Sketch metaphor for the Romney campaign's intention to wipe the slate clean for the general election, I took issue with political scientist Brendan Nyhan's assertion that all such gaffes have "little electoral significance":

There are gaffes and gaffes, however.  The evidence that they don't matter is often gathered from polls taken shortly before and after the incident in question, showing little difference -- e.g., in this John Sides post cited by Nyhan.  Some campaign blowups sink deep, however, and some are gifts that keep giving for the opposition.  When making phone calls for Obama in the fall campaign in '08, I spoke to several people whose opinion of Obama had seemingly been shaped by the Jeremiah Wright affair or by his "cling to guns and religion" riff.  Perhaps their fears about him -- in some cases racist ones - -simply seized on those handy objects. But who's to say whether some such explosive objects-to-hand may not pack more charge than others? That anxieties about Obama's "black agenda," as one person characterized it to me, would not have been less intense if that particular fodder had not been furnished?  And when a potent negative perception works its way over time into people's overall perception of the candidate, is it detectable in polling?

It seems to me that Fehrnstrom has put a weapon with staying power in the hands of Romney's opponents, chiefly Obama.  Any time an antagonist wants to call attention either to a new tack-to-the-center policy shift or an old one, he or she can figuratively shake an Etch-A-Sketch

Well, the Obama campaign at least takes this view -- at the highest levels.  Obama's May 10 interview with Robin Roberts (famous for other reasons) included this exchange: