Abstract
The belief that a military veteran candidate receives an electoral benefit at the polls based on a history of military service remains a widely held assumption in American politics. However, this assumption of a veteran electoral bonus has rarely been studied by scholars and the limited literature displays mixed results. This article presents the findings of a new study that addresses the mixed results in the literature and presents evidence that demonstrates that certain types of military veteran candidates do gain a veteran bonus in congressional elections. This advantage over nonveterans is conditioned by party, the type of race, and the nature of military service. By analyzing general election races for the United States Senate over 34 years (1982–2016), the study uncovers support for Democratic candidates with military service receiving an electoral bonus at the polls. This electoral bonus is most widely enjoyed by Democratic veterans in open Senate races and with experience in deployed warzones. The key findings suggest that previous conclusions in the literature with respect to establishing a veteran bonus in congressional elections should be reexamined to expand the time period of analysis, restructure the characterization of military experience beyond a binary variable, and include both House and Senate elections.
“In one of the sharpest exchanges of the campaign, Mr. Webb and Mr. Allen squared off on the war in Iraq on ‘Meet the Press’ on NBC on Sunday, with Mr. Allen defending the Bush administration’s policy and denouncing the ‘second-guessing and Monday-morning quarterbacking’ of the critics. ‘We’re going to need to do what it takes to succeed,’ Mr. Allen said, when asked if he would support additional troops in Iraq, ‘because it’s essential to the security of the United States of America.’ Mr. Webb responded: ‘I know what it’s like to be on the ground. I know what it’s like to fight a war like this, and—there are limits to what the military can do.’… Mr. Webb also took several digs at what he called theorists in the administration and among its allies who know combat only in the abstract. Mr. Allen, like the majority of the current Congress, did not serve in the military.”
1
Introduction
In the weeks leading up to the November 2006 midterm elections, Tim Russert hosted debates for several key United States Senate races on the popular Sunday morning news program “Meet the Press.” The two general election candidates for the U.S. Senate in Virginia, incumbent George Allen and challenger James Webb, debated for an hour on national television on September 17, 2006. This one debate, while just a singular moment in a unique campaign for one Senate seat over a decade removed, laid bare the stark difference in the biographies of two candidates seeking election to the “world’s greatest deliberative body.” Each candidate represented the polar opposite of the other with respect to military service and political experience in elected office. The New York Times characterized the 2006 Virginia Senate race as “a campaign of contrasts…combat boots versus cowboy boots.” 2
In November 2006, after a campaign that garnered national media attention largely due to a major gaffe by Allen in the months leading up to the election, the political world was shocked by Webb’s narrow victory. 3 At his victory rally, Senator-elect Webb removed the combat boots he was wearing and emphatically held them in the air as the crowd applauded. The boots belonged to his son, who had followed in his footsteps as a Marine and was deployed to Iraq throughout his father’s first run for office. Webb had worn them throughout the previous months, a visible symbol that his military experience and national security expertise was the stated foundation of his long shot and ultimately successful campaign. 4
The 2006 Virginia Senate race is just one example that speaks to the long-held notion in American politics that a candidate biography that includes military service helps gain votes. The Allen–Webb race was decided by the interaction of numerous factors over many months. But the basic fact of the outcome remained that an established, experienced candidate with an overwhelming fundraising advantage was beaten by a decorated military veteran with no prior experience in elected office. Was Webb’s victory an example of an overarching “veteran bonus” in elections, or the idea that a candidate’s military service provides a boost with voters? Or was it simply an outlier data point representing one successful race in an election year known in the press in part for being the year of the “fightin’ Dems”? 5
The continuous reinforcement of the military biography of veteran candidates in campaign strategy over many decades supports the widespread belief in the existence of a veteran bonus in elections. Surprisingly, this belief that military veterans possess an advantage at the polls has rarely been tested by scholars. The literature that does examine this question focuses largely on presidential contests (Somit, 1948; Teigen, 2007, 2018), a limited number of elections for the House of Representatives (Somit & Tanenhaus, 1957; Teigen, 2008, 2013, 2018), and hypothetical Senate races (McDermott & Panagopoulos, 2015).
To be clear at the outset, political scientists have concluded to date that there is not a blanket veteran bonus for all candidates when considering most types of federal elections. After all, if such a widespread veteran bonus existed over time in American elections, “there would be 535 veterans on Capitol Hill” (Teigen, 2008, p. 123). Beyond ruling out a systematic electoral bonus for all veteran candidates, the literature to date presents limited understanding of the conditions that may lead to an electoral veteran bonus. In the examination of congressional contests, prior studies exhibit four key limitations in seeking to establish the existence and nature of the veteran bonus: (1) the analysis of short time periods in American electoral history; (2) the examination of elections exclusive to the House of Representatives; (3) the lack of candidate quality data beyond incumbency status; and (4) the homogenizing of military experience in the binary coding of veteran and nonveteran candidates.
This article addresses these four limitations in the current literature and provides evidence of the situational nature of the veteran bonus in congressional elections. First and foremost, the results of the study strongly suggest that the veteran bonus in U.S. Senate elections over the last three and a half decades is specific to Democratic party candidates. Second, the evidence suggests that the veteran bonus increases in cases where the Democratic candidate is running in an open race. Finally, the veteran bonus is also conditioned by the type of military service, with Democratic candidates with experience deployed to a warzone gaining the largest veteran bonus overall. These findings suggest that prior studies should be reexamined, and the study provides new insights into military veterans running for election to the United States Congress.
Military Veteran Candidates and Elections
The first major study of military experience and elections was put forward by political scientist Albert Somit (1948). His “Somit thesis” simply stated that for presidential elections, “a party nominating a military hero rather than a civilian would be enhancing its chances of winning the election” (Somit, 1948, p. 200). Somit’s methodology and data have since been challenged (Karsten, 2012), and a recent comprehensive study of military veterans seeking the presidency found no veteran bonus when modeling presidential election data from 1948–2012 (Teigen, 2018). Obviously, the biggest challenge in evaluating the Somit thesis with presidential contests is the fact that the number of cases in any analysis is so small in number. For this reason alone, studies of congressional elections provide better data. As noted many decades ago, “by patient analysis of congressional voting, recurring patterns can be identified, patterns that disentangle to some extent the components of nationalism and parochialism in a national election” (Key, 1964, p. 546).
Turning to congressional elections as a better testing ground with more numerous cases and twice the number of election cycles, additional research to test the Somit thesis is focused almost exclusively on the House. The first of these studies by Somit and Tanenhaus (1957) examined a one-third sample of House races from 1950 to 1954 by region of the country. Somit and Tanenhaus concluded, with a limited sampling of data, that the electorate overall shows no preference for veteran candidates over candidates without military experience. In a more recent study of House races from 2000 to 2006, Teigen (2008) found that Democratic veterans in the 2002 election and Republican veterans in the 2006 election did gain a small veteran bonus at the polls (between 1 and 2 percentage point advantage in vote share). However, this effect in the Teigen (2008) study only appeared in two of eight election cycle cases (i.e., the study involved four election cycles from 2000 to 2006, testing for significance of both Democratic and Republican candidates as veterans). Teigen (2018) updated this House data and analysis to encompass House races from 2000 to 2014, and again found no systematic impact of military biography on election outcomes.
The only known study regarding military service and Senate elections was completed by McDermott and Panagopoulos (2015). This experiment within the 2008 Cooperative Congressional Election Study (CCES) found that in a hypothetical Senate race, a military biography only positively impacts a Democratic candidate with Republican voters and those with more interventionist views. McDermott and Panagopolous’ experimental research built upon Teigen’s (2013) experiment with potential voters, which concluded a military biography helps a candidate in voter perception of their ability to handle defense issues but not in other areas. While both McDermott and Panagopoulos (2015) and Teigen’s (2013) recent experimental work was a huge step forward in the understanding of military veteran candidates and the impact on voters, especially with their treatment of party impacts, each of these studies included important limitations. Specifically, Teigen’s (2013) experiment dealt only with voter perceptions of a candidate and did not measure the impact of military service on actual vote choice. More importantly, both experiments did not include any distinction of the types of candidate military experience beyond the binary measure of veteran versus nonveteran.
Even though the available evidence from scholars does not offer any strong support for Somit’s thesis in either presidential or congressional elections, the assumption that military experience is helpful to candidates endures. One only needs to look at current and recent campaigns to observe this fact. Campaign messaging consistently highlights military service if a candidate is a veteran, and this often manifests itself as the leading biographical characteristic put forward above all other experience. This is frequently the case even when military experience is limited in duration as compared to other personal and professional background characteristics of the candidate. Military experience in some instances is the sole message or primary focus of campaign advertisements.
Veteran Bonus Theory
Regardless of whether the literature supports the idea of a blanket veteran bonus, the fact that campaigns continue to stress military experience over the last few decades suggests that some veterans receive an electoral advantage under certain conditions. The foundation of a “veteran bonus theory” is the association of candidates with the highly respected institution of the military. The military remains at the very top of the most-trusted institutions in American society since Gallup began polling citizen institutional confidence in the early 1970s, 6 and this leads to the logical assumption of a positive impact with voters. If over 70% of Americans view the military in a favorable light, but <10% have confidence in Congress, then a Senate candidate such as Rep. Martha McSally’s (R-AZ) stressing her military experience and not her service in the House during a campaign passes a common-sense test at first glance. It also suggests that military service may constitute a separate and distinct measure of candidate quality presented to voters, beyond the traditional measure of achieving prior elected office.
It is known that voters deal in low information environments and rely on heuristics, or mental short cuts, to evaluate candidates (Popkin, 1994; Lupia, 1994; Kahn & Kenney, 1999). If the military service of veteran candidates is continually stressed in campaign advertisements and media coverage, then it follows that this military cue may have an impact on electoral outcomes in the form of a veteran bonus. The limited existing literature suggests that any veteran bonus might be conditioned on the timing of the election cycle (Teigen, 2008), as well as the party identification of the veteran candidate (McDermott & Panagopoulos, 2015).
As far as the political party of veteran candidates, the literature suggests that a veteran bonus may impact both Democrats and Republicans. The first reason for this is the fact that candidates do better with voters when campaign messages are based on personal background and experience (Sellers, 1998). Therefore, a candidate from either party that served in the military could reap a bonus based on campaign messaging that highlights this personal experience. Additionally, political parties have developed issue ownership over time (Petrocik, 1996; Petrocik et al., 2003; Ansolabehere & Iyengar, 1994) that leads to the potential for mixed results with respect to veteran candidates. The Republican Party has typically “owned” the issue of national defense since the 1980s (Damore, 2004; Sides, 2006; Holian, 2004). A Republican candidate with military experience is in position to reinforce personal expertise in an area that is already favorable to his or her party. In times of high salience for defense issues, voters could be drawn to a Republican military veteran as the most experienced candidate in the political party with the best reputation for handling national security matters. In essence, a Republican veteran candidate could receive an enhanced veteran bonus from both personal experience and party reputation.
On the other hand, there is research supporting the ability of candidates to successfully “trespass” on issues owned by the other party (Damore, 2004; Holian, 2004; Hayes, 2005). If Republicans are viewed by the electorate as owning national defense issues, then a Democrat with military experience could be viewed as the ideal trespass candidate (Sellers, 1998; Sides, 2007). Recent history in congressional races shows that party strategists are drawn to this line of thinking. Despite mixed electoral outcomes, the Democratic Party continues to highlight its recruitment of military veterans in the last decade (Scher, 2017). With no existing research to resolve these competing approaches, this study tests both hypotheses:
Finally, returning to Somit’s (1948) thesis that a war hero candidate receives an electoral bonus, this original hypothesis as defined has never been tested in congressional elections. Current studies of military veterans in congressional races code military service as a binary variable, veterans and nonveterans, without further defining the type of service of veterans into multiple categories or utilizing the war hero label. This follows the use of a simple binary veteran variable, veteran or nonveteran, in other studies of legislative behavior (Lupton, 2017; Bianco, 2005). Some scholars do speculate that this dichotomous coding of military service may not suffice, as it “may overly homogenize the concept of military service” and create a situation where “a former supply clerk at Fort Dix and a Congressional Medal of Honor winner…appear identical…but could yield very different campaign benefits” (Teigen, 2008, p. 123). In the same way a variable for candidate quality differentiates between levels of elected office, a categorical variable for military service can be classified beyond a binary measure.
In line with Somit’s original war hero distinction, the most obvious next step is the bifurcation of the veteran variable between those who served in a military conflict and those who did not. This distinction is also readily apparent in campaign messaging and strategy, as war veterans seeking elected office almost unanimously adopt language to highlight this fact. In recent campaigns, numerous congressional candidates that served in Iraq or Afghanistan make this specific part of their biography very visible to voters in the first few lines of websites and campaign advertisements. Media coverage follows suit in stressing the service of candidates in a theater of war. Martial achievement through service in war is expected to be viewed more favorably by voters, as it represents a higher level of service and sacrifice within the positively rated institution of the military.
7
The final hypothesis for the nature of the veteran bonus is therefore:
From a theoretical perspective, testing these three hypotheses is best achieved with Senate elections. First and foremost, Senate elections “occupy a niche somewhere between presidential and House elections” (Fiorina, 1989, p. 116). Political scientists tend to study the House of Representatives, but it is Senate elections that are often more competitive and provide more variance in outcomes. Senate incumbents win re-election less consistently than House incumbents, and Senate elections typically involve more campaign resources, media attention, and quality opponents facing off in hard-fought contests (Jacobson & Carson, 2016; Kahn & Kenney, 1999). In summary, “U.S. Senate elections are an ideal laboratory for examining the dynamics of campaigns” because they provide a “variance (that) cannot be found in other federal elections” (Kahn & Kenney, 1999, pp. 30–31).
Data: Senate Elections (1982–2016)
This study tests the three veteran bonus hypotheses with Senate election data from 1982 to 2016, which represents a significantly longer period than any prior study. Of note, this period coincides with the decline of the veterans serving in both the Senate and in Congress overall. Senate election data is utilized for four primary reasons that relate to key differences between Senate and House elections (Jacobson & Carson, 2016; Abramowitz & Segal, 1993; Kahn & Kenney, 1999): (1) Senate elections are more likely to have high-quality challengers and be more competitive; (2) Senate elections rarely have uncontested races; (3) Senate races are often more visible in the media to voters; and (4) the geography of Senate races is stable over time and not impacted by redistricting/gerrymandering. Additionally, the practical reasons for studying Senate elections in a study that desires to examine longer periods of electoral history cannot be ignored. Collecting biographical information, especially on challengers, for Senate candidates in 33–34 races every two years with higher visibility and media coverage is simply a more achievable task than the same effort in more numerous and less visible House races.
Summary of Dataset
The unique Senate general election dataset developed for this study contains 18 election cycles over 34 years (1982–2016). Each Senate class (I, II, III) is represented 6 times. The dataset overall contains 591 contested Senate races and 1182 general election candidates. Each of the 1182 general election Senate candidates and 591 contested races from 1982 to 2016 were coded for the following characteristics: election year; state; candidate quality (0–5, dummy variable ranking from no previous elected office held to incumbent senator); gender; party ID; basic military service biography (binary variable, veteran, or nonveteran); detailed military service biography (nonveteran, common veteran, and war veteran); general election Democratic vote share; the Democratic candidate’s previous margin of victory/defeat in the most recent Senate election for each seat (i.e., 6 years prior); the Democratic presidential candidate’s margin of victory/defeat in each state in the most recent presidential election; and the percent of spending in the race by the Democratic candidate. 8 Details of the coding of each variable in the data is available in the Supplementary Appendix.
Dependent Variable
The dependent variable in the study is the Democratic vote share as the measured outcome of each Senate election from 1982 to 2016. The use of Democratic vote share (D votes/R+D votes) is arbitrarily chosen and follows the methodology of two previous studies of the impact of veteran status on House races (Teigen, 2008, 2018). A positive regression coefficient for Democratic vote share represents a positive vote share increase for the Democratic candidate. A negative regression coefficient represents a vote share decrease for the Democratic candidate. If Republican vote share (R votes/R+D votes) is utilized in the same models, the results are the same with opposite signs on the coefficients.
Independent Variable
The independent variable of interest is the veteran status of each of the major party candidates in Senate races from 1982 to 2016. In the dataset, all major party candidates in the 591 Senate races from 1982 to 2016 were first coded with a binary veteran status (veteran or nonveteran) based on any type of military service. The major party candidates were then coded with a more detailed ordinal variable to better delineate the nature of military service beyond a binary variable (0 = nonveteran, 1 = common veteran, 2 = war veteran). A war veteran coding represents military service that included deployment to an area of conflict (Vietnam, Middle East, Iraq, Afghanistan, etc.), regardless of whether or not the candidate experienced actual combat action. A common veteran coding represents any military service that did not include a deployment to an area of conflict. A detailed description of the definition of the coding process for the data with respect to war veteran and common veteran is available in the Supplementary Appendix.
Control Variables
The control variables include commonly used factors that affect election outcomes, including candidate quality measured by level of previous elected office held, gender of the candidate, previous vote share in the last Senate election, the latest presidential vote share as a measure of overall state partisanship, and campaign spending. Of note, these control variables are similar to Teigen (2008, 2018) with the exception of candidate quality. Whereas previous studies simply use a binary measure of incumbency or prior elected office as a measure of quality, the dataset in this study controls for a more detailed dummy variable for candidate quality (0 = no experience in elected office, 1 = other minor elected office, 2 = statewide elected office, 3 = former U.S. House member or big city mayor, 4 = former Governor, and 5 = incumbent Senator). Any “quality” measure of a candidate categorized in this way is arguably a crude measure, but this represents a step beyond a simple measure of incumbency. Of note, no experience in elected office is used as the reference category for this study. Given that Senate races are statewide elections, it is important for example that any model account for the variation in candidate experience between a small-town mayor (candidate quality = 1), a state Attorney General (candidate quality = 2), and a former Governor (candidate quality = 4). Incumbency is still accounted for in this control variable as the highest level of quality (candidate quality = 5). The extended timeframe of the data and the inclusion of specific control variables follows similar approaches by both electoral scholars and election forecast models that focus on the Senate (Sides, 2014).
Pooling of Data
After completion of the dataset, the 591 races were pooled in two groups for analysis. The first grouping contained the aggregate data of all 591 races from 1982 to 2016. The second grouping contained the open races only from 1982 to 2016, which resulted in a subset of data of 140 races. Due to the well-known advantage of incumbency in American elections, it is important to examine the open races separately for effects of veteran status. The detailed summary of the data is outlined in the Supplementary Appendix. Figures 1 and 2 also illustrate key descriptive characteristics of the dataset. Veteran candidates in senate election cycles (1982–2016). Veteran candidates in senate election cycles, type of veteran breakdown (1982–2016).
Of note, the number of veteran candidates in Senate elections has declined in the last 30 years, which is consistent with the decline in overall veteran representation in Congress. Figure 1 displays the fact that Republican veterans outnumber Democrats in small numbers in a few election cycles since 1982. Figure 2 highlights the fact that common veterans outnumber war veterans in almost every election cycle in the dataset. A notable exception to this is the 2016 election cycle, which may be an indication that the lengthy wars since 9/11 are now beginning to result in more war veteran candidates in Senate elections.
Results and Discussion
The analysis of the results begins with Teigen’s (2008) method of initially grouping the data by the average Democratic vote share received in the four types of races. The four types of races when categorized by military service are: (1) Democratic veteran candidate versus Republican veteran candidate; (2) Democratic veteran candidate versus Republican nonveteran candidate; (3) Democratic nonveteran candidate versus a Republican veteran candidate; and (4) Democratic nonveteran candidate versus a Republican nonveteran candidate.
Mean Two-Party Democratic Vote Share by Candidates’ Military Service, U.S. Senate Elections (1982–2016).
Number of cases in parentheses.
Using the nonveteran versus nonveteran races as the baseline, Figure 3 displays the overall average veteran advantage in vote share in the aggregate data from 1982 to 2016. Of note, the baseline for each type of race is different in the data based on the mean vote share for the nonveteran versus nonveteran race in each pooled data (49.67 for all contested races, 47.45 for open races). On average, Democratic veteran candidates received 4.99% more vote share in all contested races and 2.62% more vote share in open races from 1982 to 2016 than Democratic nonveterans. Republican veterans gained a 4.02% advantage in all contested races, but did 1.16% worse than Republican nonveterans when only open races are considered. Average veteran advantage in vote share, U.S. Senate elections (1982–2016).
Mean Two-Party Democratic Vote Share, U.S. Senate Elections, 1982–2016, Veteran as Binary Variable, All Contested Races (N = 591).
Standard errors in parentheses.
+p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.
Mean Two-Party Democratic Vote Share, U.S. Senate Elections, 1982–2016, Veteran as Common Vet versus War Vet, All Contested Races (N = 591).
Standard errors in parentheses
Mean Two-Party Democratic Vote Share, U.S. Senate Elections, 1982–2016, Veteran as Binary Variable, All Open Races (N = 140).
Standard errors in parentheses.
+p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.
Mean Two-Party Democratic Vote Share, U.S. Senate Elections, 1982–2016, Veteran as Common Vet versus War Vet, All Open Races (N = 140).
Standard errors in parentheses.
The control variables throughout the model perform overall as expected in the full models. Candidate quality, the presidential vote (overall state partisanship), and spending are all statistically significant with the signs of coefficients in the expected direction (positive for Democratic quality, negative for Republican quality, positive for Democratic spending and presidential vote share). Gender of the candidate is not statistically significant in any of the full models. The previous Democratic vote share for each Senate race was statistically significant in the models that include all races from 1982 to 2016, but this significance does not translate to the open race models. This suggests that open races for the Senate do not hinge on the results from 6 years prior. Notably, the coefficients for latest presidential vote share are slightly larger in the open races compared to the coefficients from the models that account for all 591 races (approximately +0.2% increase for presidential vote share in the open races).
Overall, the results indicate support for the Democratic trespass hypothesis (H2) over the Republican ownership hypothesis (H1). Considering the full dataset of Senate races over three and a half decades (1982–2016) and veteran as a binary variable, Democratic veterans receive a veteran bonus of slightly over 1.2 % points in vote share at a statistically significant level after controlling for other major electoral outcome factors (candidate quality, gender of the candidates, previous vote share, partisanship of the state, campaign spending, and the year). When this full model considers the distinction between common veterans and war veterans, the Democratic war veteran bonus increases to approximately 1.8 percentage points in vote share and remains statistically significant. In the binary veteran model, Republican veteran candidates do not impact the outcome of the race at levels of statistical or substantive significance (coefficient −0.178 and not statistically significant) as measured by Democratic vote share. However, in the model that distinguishes between types of veteran service, Republican war veterans do impact the outcome of the race, with an average decrease of approximately 1.6 percentage points in Democratic vote share at a statistically significant level. This finding highlighted in the full model of Table 3 suggests strong support for the war veteran bonus hypothesis (H3).
In open races (N = 140), Democratic veteran bonus increases to over three percentage points in vote share (full model in Table 4, coefficient for Democratic veteran candidate = 3.488 and statistically significant at 0.01 level). When type of military service is considered (Table 5), the bonus for Democratic veterans is exclusive to war veterans and increases to over five percentage points (coefficient = 5.234 and statistically significant at 0.01 level). This finding is notable in that it exceeds the electoral impact of incumbency in the full Senate dataset, where incumbency impacted Democratic vote share by approximately +4.8 percentage points for Democratic incumbents and −4.7 percentage points for Republican incumbents over the reference category of no political experience for a candidate.
However, these findings with respect to open races require more data and analysis races above and beyond the 34-year period of the data in this study. The fact remains that the overall numbers of veterans by party in open Senate races as outlined in Table 1 is relatively small in number (40 Republican veterans in open races, 38 Democratic veterans in open races). The findings suggest that war veteran status matters more for Democratic candidates in open races, but clearly more data is required to further evaluate this finding and distinction between war veteran and common veteran.
The results from both the full dataset (Tables 2 and 3) and the open races (Tables 4 and 5) suggest that prior research overly homogenized the nature of military service, and that voters do distinguish between common veterans and those that serve in harm’s way. A few percent of Democratic vote share is not likely to decide most Senate elections, but the conditional nature of the veteran bonus uncovered in this study suggests that the veteran status of Senate candidates can be determinative in some contexts. Based on the initial support indicated in this study for the Democratic trespass (H2) and war veteran bonus hypotheses (H3), the election of James Webb in the 2006 Virginia Senate race—a Democratic war veteran candidate running against a Republican incumbent with no record of military service—may in fact represent an example of these findings in action.
Conclusion
Veteran candidates for the United States Senate do gain a veteran bonus at the polls under certain conditions. Utilizing a unique dataset of all Senate races from 1982 to 2016 that included the detailed military service history of each candidate, this study uncovers that veteran candidates for the United States Senate gained a veteran bonus over nonveterans that is conditioned by party, the type of race, and the type of military experience of the veteran candidate. Overall, from 1982 to 2016, military veteran Senate candidates from the Democratic Party gained an advantage of just over 1.25% in vote share at a statistically significant level when military experience is considered as a binary variable. Over this same time period when considering veteran status beyond a binary variable to include the distinction of service in a warzone, Democratic Party candidates with military experience in war gained an advantage of almost 2% in vote share. Ceteris paribus, war veterans from both major political parties impacted the election outcome at a statistically significant level between one and two percent of vote share. In open races without an incumbent, the veteran bonus in Senate races from 1982 to 2016 was exclusive to Democratic Party candidates. The veteran bonus for Democratic Party candidates in open Senate races remained statistically significant and increased to over 3% of vote share when veteran status is considered as a binary variable and increased to over 5% for war veterans. The most important finding of this study is that the war veteran bonus for Democratic Senate candidates running in open races rivals the statistical and substantive significance of incumbency (over 5% of vote share).
It is critical to note that this study utilized Senate election data exclusively in the analysis. Therefore, it is not suggested that these results immediately translate to House races without further inquiry. There are many trade-offs in the decision to study Senate versus House elections over time. House elections, because of increased numbers of candidates per election cycle, allow better leverage to evaluate hypotheses with additional data and to narrowly evaluate additional factors such as the salience of national security issues at the time of elections. However, this leverage is hindered by the issue of redistricting around ideal testing periods that include the election cycle immediately following the attacks of September 11, 2001. Additionally, House elections are often less visible, less competitive, and present challenges with the ability to verifiably code detailed military experience from publicly available biographical data. That said, the findings of this study do suggest that prior work should be reevaluated with the expanded coding of military service in House races to further test the findings of the war veteran bonus hypothesis. An obvious place to start would be the development of a new dataset for House elections around the first Gulf War (1988–1992), and the recoding of at least a few election cycles of Teigen’s existing House dataset (2000–2014) around 9/11 to include a distinction between common veterans and war veterans.
In conclusion, the study presented in this article suggests that a veteran bonus in congressional elections does exist under certain conditions and that all military service is not the same in the eyes of the electorate. Distinguishing between common veterans and war veterans is important for uncovering the nature of the veteran bonus in congressional elections. In recent years, the Democratic Party has encouraged Iraq and Afghanistan war veterans to seek election to Congress. Based on the empirical evidence presented in this study, and the ability of Democratic war veteran candidates to reap a veteran bonus in Senate races over the last three decades, this may be a wise strategy.
Supplemental Material
sj-pdf-1-afs-10.1177_0095327X211038032 – Supplemental Material for The Electoral Impact of Military Experience: Evidence From U.S. Senate Elections (1982–2016)
Supplemental Material, sj-pdf-1-afs-10.1177_0095327X211038032 for The Electoral Impact of Military Experience: Evidence From U.S. Senate Elections (1982–2016) by David K. Richardson in Armed Forces & Society
Footnotes
Acknowledgments
The author would like to thank the anonymous reviewers, as well as Sarah Binder, Danny Hayes, and Eric Lawrence at the George Washington University, for their insights and feedback.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author discloses receipt of the following financial support for the research, authorship, and/or publication of this article: The author received financial support in the form of a doctoral study fellowship from the United States Navy for conducting this research and authorship as part of his doctoral dissertation.
Author's Note
The views of the author are his own and do not reflect the position of the U.S. Naval Academy, the Department of the Navy, or the Department of Defense.
Supplemental Material
Supplemental material for this article is available online.
Notes
Author Biography
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
