How Hurricane Sandy Could Matter on Election Day

I’ll speak to three possibilities, just for starters:

1) Sandy hurts Obama.  As Larry Bartels wrote about a few months ago, incumbent presidents can be punished for natural disasters, including floods.  Of course, a lot hinges on whether serious damage hits the key battleground states in Sandy’s path: Virginia and Pennsylvania, plus probably New Hampshire and maybe parts of Ohio.  Given Obama’s apparent lead in Pennsylvania, it’s less likely that Sandy’s damage could swing that state against him.  So the focus should be on the other states, I think.

2) Sandy helps Obama.  Some research by John Gasper and Andrew Reeves finds that while both presidents and governors are punished for natural disasters, they are rewarded for a proactive response to those disasters.  Presidents can overcome the effect of even the most severe weather damage in a particular location by declaring that location a disaster area (and therefore eligible for federal disaster funds).  Here’s a graph that illustrates this, with the key part highlighted:

The open question, however, is whether any disaster declarations could happen quickly enough and visibly enough to matter on Election Day, only about a week after Sandy makes landfall.

3) Sandy limits early voting, possibly hurting Obama’s chances.  David Axelrod is already suggesting this could happen. Alec MacGillis wonders if this could make a split between the popular vote and the Electoral College more likely, by potentially depressing turnout in blue states.   Perhaps some early voters who are stymied by Sandy would then fail to vote later on, including on Election Day.

I don’t think this is likely.  What the literature on “convenience voting” suggests is that measures like vote-by-mail and early voting tend to make it easier for habitual voters to vote, rather than stimulating turnout from marginal or infrequent voters.  See this piece by Adam Berinsky.  I’ll quote from the summary on his webpage:

Reforms designed to make voting “easier” exacerbate the existing biases in the composition of the electorate by ensuring that those citizens who are most engaged with the political world continue to participate.  That is, voting reforms encourages the retention of likely voters from election to election rather than encouraging new voters to enter the electorate.

So, for example, those Marylanders inconvenienced by the cancellation of early voting on Monday will likely show up at the polls anyway, especially since this is a presidential election.

I welcome other thoughts and especially citations to relevant research in comments.

7 Responses to How Hurricane Sandy Could Matter on Election Day

  1. Jonathan Baron October 28, 2012 at 3:48 pm #

    Is it possible that the hurricane could help Obama by causing massive power outages and road closures? These might make voting more difficult in suburban and rural areas, but they are less likely to affect cities. I can’t find (in Google Scholar) any studies of the effect of outages on voting, or preponderance in cities vs. suburbs. Apparently the Department of Energy keeps records of these things, but I can’t find them. (I live in Center City Philadelphia and support Obama, so I may be engaging in wishful thinking.)

  2. Neil Malhotra October 28, 2012 at 7:35 pm #

    Hi John:

    Andy Healy and I have an Annual Review of Political Science piece coming out at the end of the year covering a lot of this research agenda, hopefully in time for the next natural disaster that strikes the country!

    Some relevant research that we cite:

    1. Jowei Chen has a terrific paper in the AJPS showing that the Administration can boost turnout among co-partisans with disaster relief efforts, and lower turnout among opposite partisans (by demobilizing them from supporting the challenger):

    2. I think the more relevant citation than Achen and Bartels here is Gomez, Hansford, and Krause who in the JOP did a copious analysis showing that aberrant weather is more likely to depress turnout among Democrats. Achen and Bartels looks at year-long aberrant weather while GHK look at weather shocks around/on Election Day (although the Berinsky point about convenience voting is quite well taken):

    3. Lastly, Andy and I have done some work on disasters, but we interpret the data a bit differently. Like Gasper and Reeves, in our APSR paper we find that relief efforts can earn the incumbent votes, but preparedness spending does nothing, which creates some perverse incentives:

    Finally, in our QJPS paper we find that tornado damage (but not deaths) lower incumbent vote share, more consistent with economic voting than emotional voting. Tornadoes are a bit more empirically tractable than hurricanes since they are usually spatially contained in a county (although Chen does some nice geospatial modelling to study hurricanes):

    • John Sides October 28, 2012 at 7:50 pm #

      Many thanks, Neil. It was an oversight on my part not to cite this work — some of which we’ve even mentioned on this blog before. Chalk it up to writing quickly in response to events.

  3. anne October 29, 2012 at 9:09 am #

    Gods will be done….this does seem providential or is it presidential.

  4. Eitan October 29, 2012 at 2:18 pm #

    Hi John –

    In a 2010 QJPS piece, Bernard Fraga and I find that rain and snow only affect voters in uncompetitive jurisdictions (see here:
    We re-analyze the terrific dataset compiled by Gomez, Hansford, and Krause (cited by Neil above) and find a really heterogeneous effect with respect to weather. In states that are close, there is no measurable negative effect of weather on turnout.

    We think that the campaign activity and voter interest in competitive places compensate for any nuisance brought on by bad weather. If our results are generalizable here, we might expect to see very different effects in PA than NJ, for instance.


    • John Sides October 29, 2012 at 4:05 pm #

      Thank, Eitan! Just tweeted that.

  5. Eli Rabett October 29, 2012 at 7:24 pm #

    Wrong. Sandy is hitting the democratic (eastern) end of PA.