In Search of a Possible Effect of Romney’s 47% Comments

by John Sides on September 26, 2012 · 4 comments

in Campaigns and elections

{ 4 comments }

Martin Vandernort September 26, 2012 at 9:50 pm

The conventions give rise to “bounces” not trends. The recent slide in Romney’s numbers are unambiguously 47% related. I don’t see why this is surprising. What he said on the tape was … grotesque. Anybody can see that. Everybody was talking about it. Everybody. They’re still talking about it. It was a gaffe that pervaded the entire national dialogue for a week. And rightfully so.

John Sides September 26, 2012 at 10:33 pm

Martin: The effects of conventions are bumps, not bounces. Read Erikson and Wlezien, The Timeline of Presidential Campaigns, or see my post:

http://fivethirtyeight.blogs.nytimes.com/2012/08/26/how-will-party-conventions-affect-the-presidential-race/

Statements like “unambiguously” reflect your opinions of Romney’s statement more than what the data say. Most of Obama’s recent gains in the polls occurred after the DNC but *before* Romney’s comments. On average, the polls have moved 1% since Romney’s comments.

Martin Vandernort September 27, 2012 at 6:41 pm

So a convention (i.e. informercial) can cause an increase in the polls but a horrendous gaffe that monopolizes media air time for a week cannot.

But in point of fact, conventions “bumps” or “bounces” cause a large spike in their immediate aftermath followed by a decline. There may be some residual gain (bump).

But look at the charts. Romney got a “bounce” which was then counteracted by an Obama “bounce” and then after the 47%, Romney numbers dropped.

A horrendous gaffe matters. The congressman who liked to dress up as a Nazi — bad gaffe. And it mattered. Sarah Palin spouting incoherent nonsense in a succession of interviews. Gaffes, and they mattered.

Yes, candidates behavior matters in elections. It’s not all “structural”.

Josh R. September 27, 2012 at 11:56 am

Are aggregate shifts (e.g. X% support prior to comments to X+/-Y% support afterwards) the only/best way of looking for (or conceiving of) causal effects in this case?

Potential alternatives/additions:
1. Voting Intentions (comments might not have changed number saying they’d vote for Romney, but might change but still led to a decreased probability that they will actually turn out – my background here is the argument by Yanna Krupnikov that the effects of negative ads depend on their timing – difference between ads affecting decision whom to vote for and ads affecting whether to vote).
1a. (Related): Intensity of support/opposition which could turn up in…
2. Non-electoral behavior (donations, volunteering, etc.)
3. Media coverage (affirmation of Romney out of touch story line, which keeps it alive in press accounts, with potential downstream effects not observed in immediate aftermath of debacle)
4. Elite responses (e.g. elite Republicans jumping ship/preparing the ground for eventual defeat, etc)
5. Campaign behavior (Romney shifting grounds to try and look more empathetic; might lead to changes the issues campaigns attempt to prime in the last few weeks of election and/or reception of those priming efforts)

Comments on this entry are closed.

Previous post:

Next post: