Do Veterans Decrease Polarization in Congress?

If the timing of this post doesn’t make it obvious, the use of “veteran” refers to lawmakers with prior military experience, not the length of one’s tenure in Congress.  Speaking of which: Happy Veterans Day!
Chris Day—a colleague of mine at the College of Charleston—contributed to this post.  Chris was listening to NPR’s “Morning Edition” when Cokie Roberts offered an intriguing explanation of polarization in Congress.  Chris writes:

On today’s Morning Edition, Cokie Roberts observed that only about 20% of today’s Congress are military veterans when compared to the nearly three quarters of the post-Vietnam era.  She suggested that that this decrease may account for the increase in polarization currently facing Congress.  In other words, with the exodus of those with such shared experiences, both on and off the battlefield, so departs the patter, the mutual respect, and the “getting things done together” solidarity associated with veterans.

Chris, a comparative politics scholar who writes on conflict in Africa, notes the theoretical appeal of Roberts’ hypothesis:

We like to think that on Veteran’s Day, Americans of all political varieties can come together to confer well-deserved respect upon America’s armed forces…We particularly admire the fellowship of veterans.  Indeed, recent books such as Sebastian Junger’s “War” remind us of the shared sacrifice and camaraderie so central to surviving extreme experiences in violent environments – they get through things because they get through them together. So let us not forget political polarization.  Even our reverence for military camaraderie raises interesting questions.  In making the transition from soldier to civilian, just how durable is this fellowship?  For those that continue a career in public service after leaving the military, does having such a background matter?

At the blog we’ve detailed the reasons for polarization in Congress in various posts (see here, here, and here).  But while Roberts’ hypothesis isn’t one of the “usual suspects” cited by congressional scholars, it’s certainly testable.

I’ll match the theoretical appeal noted by Chris with the empirical side of things.  According to data from the Congressional Research Service, in the 96^th Congress (1979-1981) and 91^st Congress (1969-1971), 55% and 75% of representatives (respectively) had past military service.  In the current Congress, by comparison, only 19.9% of representatives previously served in the military according to the data collected for this post.  Now on its face, this supports Roberts’ claim, as military service has declined more or less linearly along with polarization.  However, we’re risking an ecological fallacy here, so we better check the individual-level data.

In the forthcoming analysis, “Polarized”—the dependent variable—is measured as the absolute value of a lawmaker’s DW-Nominate score.  The data is for representatives serving in the 112th Congress.  “D_Polar” represents the polarization of a lawmaker’s district using the absolute value of Obama’s two-party vote share in 2012.  The variable “Party” is coded 1 for Republicans and 0 for Democrats.  And finally, “Military” is coded 1 for representatives with prior military experience and 0 otherwise.

A multivariate regression model is reported below.  Overall, the model performs well, explaining 47% of the variation in polarization in the 112th House of Representatives.  We find (not surprisingly) that Republicans and representatives from partisan-leaning districts are more ideological than Democrats and representatives from moderate districts.  However, and most importantly, military service has no statistically significant effect on polarization in Congress.  The coefficient on military service is negative—consistent with Roberts’ suggestion—but insignificant (p=.27).  Thus, contrary to Cokie Roberts’ hypothesis, veteran lawmakers are no more or less partisan than their non-veteran brethren.  While her hypothesis has some intuitive appeal, and it’s supported by aggregate trends, the effect isn’t borne out when we look at the roll-call record.