Comments on: Right Back Where We Started!

By: Matthew Dickinson

Matthew Dickinson — Thu, 11 Oct 2012 15:53:11 +0000

Hi Tom,

Thanks for the link – I saw the headline but didn’t read it. Interestingly, several pundits were tweeting yesterday about how the debate results show once again that you can’t trust political science!

By: Tom Campanella

Tom Campanella — Thu, 11 Oct 2012 15:48:02 +0000

I am sure that you saw this article yesterday… nice shout out to the political scientists (though also raises the debate issue). Anyways, thought I’d share, given its relevancy to this blog…

http://www.washingtonpost.com/blogs/ezra-klein/wp/2012/10/10/in-excitable-pundits-vs-political-scientists-ill-take-political-scientists-every-time/

By: Matthew Dickinson

Matthew Dickinson — Thu, 11 Oct 2012 15:38:39 +0000

I don’t think that the political scientists believe “debates don’t matter.” Instead, we tend to believe that they are far less influential than pundits believe, and that they rarely persuade very many people to change their vote. In terms of polling, history suggests a “big” debate effect may be in the order of 3.5% points difference between the pre- and post-debates polls (cumulative impact of all the debates), and even here it is not clear how long that “bump” is sustained. I think it safe to say that, by historical standards, Wednesday’s debate had a bigger impact than most political scientists anticipated, but note that its full effects have yet to be measured – how long will the Romney bump last, and will it be eroded by subsequent debates?

By: El Criador de Gorilas

El Criador de Gorilas — Thu, 11 Oct 2012 12:58:12 +0000

The race is where the models predicted, but would it be right to say that the mechanism for such outcome has not been one that the literature anticipated? (“debates don’t matter”).

By: Matthew Dickinson

Matthew Dickinson — Thu, 11 Oct 2012 12:43:10 +0000

George,

Your point is well taken. But,fairly or not, that’s what the forecast models “assume” – that most people give credit, or blame, for the fundamentals (state of the economy, progress in war on terror, etc.) to the President. In other words, presidential elections are, in large part, a referendum on how things went during the president’s time in office. Now, some of the models do adjust for how long one party has controlled the White House. In this respect, since he’s only been in one term, the model apportions less “blame” on Obama than if the Democrats have controlled the presidency for two or more terms.