What do veteran endorsements of presidential candidates accomplish?

A new report that I have co-authored with James Golby and Kyle Dropp was released yesterday by the Center for a New American Security. The study, titled Military Campaigns: Veterans’ Endorsements and Presidential Elections, can be found here. The New York Times also reports on the study.

We tested whether telling voters that "most members of the military and veterans" support one candidate or the other had an impact on voter’s preferences for President Obama or Governor Romney. We found that in the aggregate, such endorsements did not seem to move vote choice by a statistically significant amount, but that they did have a statistically significant effect on voters who claimed to be independents and especially on independents who claimed not to follow foreign policy very closely.

President Obama received a statistically significant bump in support from those voters who were told of an endorsement, whereas Governor Romney did not. We believe that is because a military/veteran endorsement of Obama would be surprising, given public perceptions of the military as a conservative organization and the historical advantage Republicans have had on national security.

We go on to argue, however, that such endorsements are not good for civil-military relations because they involve the military in partisan politics. Of course, retired military may exercise their first amendment rights just as any other citizen could. But we argue:

"Retired senior officers may think they are drawing fine distinctions between the formal institution of active-duty military and their own views as retired citizens. But the truth is that no one, especially not the campaign team, is very interested in their views as private citizens. Rather, it is their symbolic role — their role as spokespeople for the military — that gives their endorsements significance."

We worry that the cycle of high-profile endorsements could in the long run help erode public trust in the military as a non-partisan institution and we find some suggestive evidence that such worries are reasonable.

The reactions to our report have been interesting. One person wrote complaining that our descriptive background section focused on the high-profile endorsements, such as Admiral John Nathman’s cameo at the 2012 Democratic National Convention or General Tommy Franks’ cameo at the 2004 Republican National Convention, but then our survey prompt was more general: "most members of the military and veterans."

We followed this research design for several reasons. First, as we explain in our study, campaigns prize high-profile endorsements because they symbolize something larger than simply a single citizen. That is why they tend to cluster them and announce them as a group (think Admiral Crowe standing in front of all those retired generals and admirals endorsing Clinton in 1992). The individual endorsements are meant to symbolize the endorsement of the larger institution. Our survey prompt captures that idea.

Second, as a practical matter, we had to design and conduct the survey during a campaign season but in advance of when the campaigns had publicized the lists of their endorsements. It would have contaminated the study to create false names or false endorsements.

One person also wondered why we didn’t ask the question directly: "If a candidate is endorsed by a retired general, would you be more or less likely to vote for that candidate solely because that?" However, the whole point of survey experiments is to capture latent effects whether or not the respondent is aware of them. Social scientists have developed these tools because asking direct questions distorts or obscures the underlying phenomena they are seeking to study.

Another person wrote to complain that we had misidentified Jason Dempsey, author of Our Army, as an "Obama supporter." It was a minor point, but I am inclined to think the critic was right. Jason wrote a piece for Huffington Post analyzing a different poll and concluded that younger military personnel were not as Republican leaning as older cohorts. Several Obama supporters pointed us to that article (and others like it) as we were doing our research and argued to us that Obama had a decisive advantage among younger military and veterans, but we did not find similar results in our own survey, which is why we wrote what we did. However, since Dempsey has gone to some pains not to be identified with one candidate or another – and, indeed, has written persuasively of the danger of the military developing a partisan identity — we should not have referenced him as an Obama supporter even though Obama supporters relied on his analysis to make their case to us.

We are grateful so many people are taking the study seriously and look forward to a lively debate. And, as is always the case in academic research of this sort, we end with a call for more funding for more research to follow all of the interesting lines of inquiry that commenters raise!