This is a guest post by Aaron Strauss, Ph.D. He is a freelance data specialist who consults for Democratic and progressive organizations. He has worked on several presidential campaigns since 2000 and was the DCCC’s Targeting Director in 2012. He tweets at @aaron_strauss.
In the immediate aftermath of the 2012 presidential election, some academics were skeptical that the Obama campaign deserved so much post-election media hype. This skepticism was certainly reasonable as nearly all journalistic “inside the campaign” accounts did not justify their conclusions through campaign outcomes (other than “Obama won”) to justify their conclusions. Since then, political scientists have estimated campaign effects on Obama and Romney’s vote share in counties and media markets. Now I have some new analysis showing that the Obama campaign had a more effective GOTV program than did the Romney campaign.
To further this discussion, I examine how individually targeted get-out-the-vote (GOTV) efforts affected turnout at the individual level, not at the county or media-market level. Thus, I am ignoring tactics such as persuasion (e.g., TV ads), voter registration drives, and mass GOTV efforts (like Sarah Silverman videos). Persuasion is an especially important campaign strategy, and may help get out the vote too, but the secret ballot makes measuring persuasion more difficult.
My data comes from state voter files matched with political variables on Catalist’s database. The gold standard for quantifying campaign effects is a randomized experiment. Unfortunately, it’s not possible to randomly assign specific voters to either the entire Obama or Romney GOTV campaign strategies. Instead, I compare the turnout of Democrats, Republicans, and independents in both battleground and non-battleground states. We’d expect higher turnout in battleground states, where there were actual GOTV campaigns, than in non-battleground states. Using 2008 as a benchmark, turnout decreased in 2012, but it decreased less in these battleground states than in other states.
Then the crucial question becomes: whose turnout was most affected in these battleground states, Democrats or Republicans? The answer provides suggestive evidence of the strength of Obama’s and Romney’s GOTV efforts.
Here’s what I found: The Obama campaign was indeed more effective than the Romney campaign at turning out their supporters. This effect is clearest among those who were not highly likely voters to begin with — who are exactly the voters that a GOTV campaign would target.
Just as the Obama and Romney campaigns did, Catalist modeled the likelihood of turnout using information in the voter file. Among those whose turnout probability was below the median likelihood, independents (whom neither campaign targeted) were actually slightly less likely to turn out in battleground states than non-battleground states. By contrast, Obama campaign’s mobilization tactics increased turnout among targeted Democrats by about 1.7 percentage points, relative to independents. The Romney campaign was also effective, but less so: their supporters were about 0.6 percentage points more likely to turn out than non-targeted independents.
Thus, the evidence shows that the Obama campaign had a larger GOTV effect than the Romney campaign—about 1 percentage point per voter in the battleground states. What does this mean for vote margins? To take the closest battleground state in 2012—Florida—I estimate that targeted GOTV efforts by Obama netted an additional 42,000 votes, while Romney’s efforts turned out an additional 7,000 votes, for a difference of 35,000 Obama votes. Those extra votes account for about half of Obama’s winning margin in Florida.
My findings parallel what John Sides and Lynn Vavreck’s The Gamble shows at the county level: the presence of a GOTV campaign, as measured by the establishment of field offices, was associated with additional votes for both Obama and Romney—but more for Obama. My findings are also consistent with those of Ryan Enos and Anthony Fowler, though I find smaller effects for the straightforward reason that Enos and Fowler capture cumulative effect over time of living in a battleground state, rather than isolating the impact of the 2012 campaign.
These new data help quantify what practitioners already know: GOTV is a crucial part of winning close campaigns, and some campaigns do it better than others.
(Methodological notes are available here.)