Earlier this week, I asked whether the drought in the Midwest could end up costing President Obama votes this fall. At first glance, it seems like a preposterous question—why would voters punish a politician for a natural disaster?

And yet, there's some evidence that voters often do punish politicians for freak weather events. Here's a 2004 paper (pdf) by Larry Bartels and Christopher Achen: "We find that voters regularly punish governments for acts of God, including droughts, floods, and shark attacks." It may be irrational, but the effect seems to be more than a mere coincidence. Bartels and Achen even suggest that severe droughts and floods may have cost Al Gore up to 2.8 million votes in 2000, enough to sway the presidential election.

Other political scientists, however, have detected some important nuances at work here. In a 2010 paper for the Quarterly Journal of Political Science, Neil Malholtra and Andrew Healy found that voters in regions hit by tornadoes do tend to vote against incumbent parties, yet for more rational reasons than Bartels and Achen suggested. Voters seem to care more about the government response than the event itself:

First, voters do not punish the incumbent party for tornado-caused deaths, which governments likely do not have the power to address with effective policy. Second, the incumbent party only appears to lose votes when no disaster declaration takes place in response to the tornado. Thus, voters appear to be rewarding and punishing government with respect to its performance in handling the disaster, as opposed to blaming the government for these natural events.

This suggests that governments can win over voters by responding competently to natural disasters. Here's another paper by Andrew Reeves and John Gasper noting a similar effect: "We find that electorates punish presidents and governors for severe weather damage. However, we find that these effects are dwarfed by the response of attentive electorates to the actions of their officials.” (Oddly enough, subsequent research (pdf) has found that voters reward politicians for disaster response, but not for disaster preparedness.)

By the way, for those wondering how the government is responding to the current drought, Bloomberg has a relevant report: "This year’s once-in-a-generation drought may leave many crop farmers largely unscathed as they are protected by taxpayer-subsidized insurance, a program Congress is moving to make more generous." Of course, that's assuming Congress can actually pass a farm bill this year—which is still far from assured.

So when it comes to natural disasters, at least, subsequent research has found that voters aren't quite as irrational as Bartels and Achen' paper suggested. Sporting events, however, are a different story. Here's the gist of another surprising paper (pdf) by Healy and Malhotra:

[W]e explore the electoral impact of local college football games just before an election, irrelevant events that government has nothing to do with and for which no government response would be expected.We find that a win in the 10 days before Election Day causes the incumbent to receive an additional 1.61 percentage points of the vote in Senate, gubernatorial, and presidential elections, with the effect being larger for teams with stronger fan support.

So, to sum up the relevant advice for any incumbent politician trying to win reelection: Natural disasters can cost votes, but a competent disaster-relief effort will usually help with voters. There's no electoral upside to making preparations beforehand for a disaster. And better hope the local football team does well in late October.