The Elusive Mandate: Searching for Meaning in Presidential Elections

by Larry Bartels on January 30, 2013 · 5 comments

in Campaigns and elections,Political science

That’s the title of a lecture I delivered recently at Harvard’s Radcliffe Institute for Advanced Study. The talk—an opinionated overview of political science research on American presidential elections—is now posted on the Radcliffe Institute’s website. Here’s the summary:

Almost 50 years ago, eminent Harvard political scientist V. O. Key Jr. described in his book The Responsible Electorate an electorate “moved by concern about central and relevant questions of public policy, of governmental performance, and of executive personality.” Bartels assesses how well Key’s optimistic portrait of the American electorate holds up in light of the subsequent half-century of electoral research. He concludes that presidential election outcomes are mostly determined by factors unrelated to central and relevant questions of public policy and governmental performance.

There’s nothing about the 2012 election, except to note that it was quite ordinary.


Andrew Gelman January 30, 2013 at 9:22 am


Could you post the slides?

Larry Bartels January 30, 2013 at 4:05 pm

I know. You came to class and took good notes. You just want the slides to study for the final. Here you go: If you have any questions, feel free to come to my office hours.

Andrew Gelman January 30, 2013 at 5:26 pm

Thanks! The slides are so much clearer when I can sit and read them while drinking a beer.

Andrew Gelman January 30, 2013 at 6:45 pm


Regarding your point that incumbents are punished for bad weather: But Hurricane Sandy is a counterexample, no? (Not that this invalidates the theory, it’s just one data point on the other side.)

Larry Bartels January 30, 2013 at 7:11 pm

Perhaps–Obama gained a bit in NJ and NY relative to neighboring states that were less directly affected by Sandy. My evidence (from joint work with Chris Achen) is derived from sustained wet or dry spells, not storms, and covers the entire 20th century. In general, I think it’s likely that (1) there’s a distribution of electoral responses, (2) responses are more positive when politicians are _seen to be_ addressing the problem energetically, and (3) the mean of the distribution is negative, which does nothing to promote accountability unless you think incumbents can prevent wet or dry spells.

Comments on this entry are closed.

Previous post:

Next post: