Thursday, January 31, 2013

The Ground Game in 2012

Location of all Obama (blue) and Romney (red) campaign field offices in 2012.
Data from John Sides and Lynn Vavreck. Map created using Google Fusion.
Last fall, I wrote about Barack Obama's field office advantage in the swing states. The final tally shows that, across all the states, Obama had 786 field offices; Romney had 284. With the election behind us, we can now assess whether that made any difference in the election's outcome.

In my analysis of the 2008 election, I found that Obama's field offices conveyed an advantage for him, generating roughly six-tenths of an additional percentage point of the two-party vote in counties where he had a field office. I ran the same analysis on the county-level vote in 2012 for all fifty states. The field office locations were provided by John Sides and Lynn Vavreck, who are using the information for their new book. I obtained the county-level election results from David Leip. I controlled for a variety of demographic characteristics of the counties.

What the new results show is that Obama's field offices did help him, but that effect was smaller than it was in 2008. While the effect was about six-tenths of a percentage point in 2008, it was about three-tenths of a percentage point in 2012. This finding is consistent with what Dan Hopkins found about advertising effects -- the effect was about three times as large in 2008 as it was in 2012 or in 2004. It's also consistent with what John Sides and Alan Abramowitz found in their preliminary studies of the 2012 campaign, which basically had a difficult time identifying any campaign effects at all. (I found no statistically significant effect for the Romney offices, just as there was no effect for the McCain offices four years earlier.)

There are several possible reasons why the 2012 Obama campaign (which, by most accounts, was even more technologically sophisticated than the data-driven 2008 version was) might have been less effective than its 2008 predecessor:

  • As Hopkins suggests, elections in which an incumbent is running (e.g.: 2004, 2012) may just experience less effective campaigns. Campaigns are likely to have more influence when voters know less about the candidates and need the ads and voter contacts to fill in the gaps. Voters knew plenty about Obama by November of 2012 and could render a verdict on his first term without help from the campaigns. 
  • It's also possible that, even if Romney's offices weren't able to move voters in his direction, they were more effective at checking Obama's efforts than McCain was. Notably, the correlation between Romney's and Obama's field office locations was .53; it was only .45 between Obama's and McCain's. Not a huge difference, but suggestive that the Republican approach was somewhat more effective this time around.
  • Finally, it may just be that Obama's innovations just weren't so... innovative this time around. His campaign's use of sophisticated voter targeting and social media were truly novel in 2008. But novelty wears off quickly, and voters become accustomed to being contacted that way. The advances in 2012 over 2008 were notable, but not enormous.

At any rate, the effect of these offices in 2012, while small, was not negligible. My analysis suggests that it was actually enough to change the outcome in Florida, which was the state settled by the narrowest margin last year. Had those voters who were motivated to vote for Obama by his field offices instead voted for Romney or not voted at all, the state would have voted, very narrowly, for Romney.

I'll have more to say on all this at a symposium next Monday at the Harkin Institute at Iowa State. If you're anywhere near Ames, Iowa, next Monday, be sure to stop by!


  1. The new tactics Obama used in 2012 were certainly innovative but I thought the coverage of them has been a bit overblown. Yes there were new ways to go after voters with social media and online data but most of the tactics that an Obama field office did were very similar to what they did in 2008 in my experience. That is you had a walk packet or call sheets for a given area that came out of the Voter Activation Network (or VAN which is a database of all voters in a given state that your field office would pull from for it's given area) database and would do direct voter contact based on that. You could canvass with your iPhone in some places but that was just a way to enter data about how John Doe at 101 Maple Street was going to vote that didn't involve writing it down on paper and entering it into a computer back at the office. An improvement, but not exactly revolutionary when it comes to field work. For my money the biggest overall change was the introduction VAN database, which depending on the state happened in 2005-2006, meaning Obama's 2008 campaign was the first presidential campaign to use that. Before that, like in 2004 or 2000, databases for Democrats were very campaign specific and could vary a great deal from area to area. In addition, I'd highlight Obama's tenancy to go after voters everywhere in a state with a lot of offices compared to Kerry or Gore where there was a tendency to target just a few geographic areas. In 2004 in Minnesota the Kerry/State Party outfit ("Kerry-Edwards Victory" was their cool sounding name) just did field work in the cities of Minneapolis, St. Paul and Duluth. In 2008 Obama and the state party had offices all over the state. So I'd say Obama's 2008 campaign was much more innovative overall compared to it's predecessors than his 2012 was.

    1. Really good points there. It may well be that the real innovation was VAN, which had nothing to do with the Obama campaign (although that campaign exploited it very effectively).

    2. I agree - I wrote a book that spends a chapter detailing the creation of the VAN (and all the work Howard Dean did as chair around data). It was an immensely important piece of infrastructure that is at the heart of Obama's two runs (as well as races across the country):