As the current (as I write) narrative moving from "Happy Gingrichmas" to "Newt is Over (If You Want It)", pundits and other horse-race handicappers are scrambling for any correlation to assume is causal. And so now we're going from the idea that debates, not TV ads, are where it's at (see here) to once again assume that negative ads work. Maybe they do, maybe they don't.
What do we know, empirically? Gingirch saw the latest surge of anybody-but-Mitt (ABM) sentiment, after the Bachmann-Perry-Cain Overdrive; Paul's numbers are up; Gingrich put on a happy face (it is rumored that he peeled it off of a small child himself) for a nice-guy ad; everyone else has been putting up TV attack ads against Gingrich. These may not be in strict chronological order.
It could be that some of these have a causal relationship. It could be that the attack ads are being seen by the same people polled about potential voting behavior. It could be that those polled people (the PP) only saw debates and on that basis moved away from Gingrich (actually, this is a canard; what are the odds, really, of the same people being hit by Quinnipiac or the WSJ or Pew, etc.?). Could be that the people polled who picked Gingrich we picking based on what they thought others were choosing (this is a real theory in Poli Sci), or that these people are just now learning who Newt is, or blahblahhypothesiscakes. Aliens. Or we could blame Majestic 12. They're always up to something.
I suppose my real question is: does there need to be a connection between all these facts? Are we so trained by reading fiction from the Holmes stories to CSI that everything shown on-screen must be taken as a piece of evidence, as something that ties into a critical narrative? Sure, the question of "do political ads on TV work?" is an interesting one, and there's an awful lot of airtime earned by ads, and there's a lot of correlation -- but has anyone done a control?