Abstract
Despite the expansion of absentee voting protections as recently as 2016, Service member spouses have not enjoyed the same level of voting protections as Service members. Active duty military spouses, uniquely positioned between military service and civilian life, are arguably as important to the election process as their Service member counterparts. Thus, we examine the voting behaviors of this underserved and seldom studied subpopulation. Matsusaka’s information theory–based economic model of voter turnout provides our framework for identifying determinants that shape the voting interest and participation of active duty military spouses. We analyze the Federal Voting Assistance Program’s 2010 Post-Election Voting Survey of Active Duty Military Spouses utilizing logistic regression models. We found that voting interest and participation were increased among respondents who planned to vote, received more election information, voted within the previous 6 years, and who were older; however, voting interest and participation were not diminished by absentee status.
Keywords
In a 2011 publication, supported by a coalition of all 15 executive branch departments and the Environmental Protection Agency, Barack Obama pledged his administration’s commitment to members of the U.S. Military Services and their families. The Obama administration not only pledged to review the resources, leadership, and support necessary for those Service members to accomplish their mission and come home safely, it also committed to improve support for military families (Obama, 2011). Michelle Obama, First Lady at the time, offered what could be considered the mission statement of the administration’s commitment, that “this government-wide review will bring together the resources of the Federal Government, identify new opportunities across the public and private sectors, and lay the foundation for a coordinated approach to supporting and engaging military families for years to come” (Obama, 2011, p. 1).
In part, this commitment encourages voting interest and participation among Service members and their families. The commitment is also illustrated in the 1986 Uniformed and Overseas Citizens Absentee Voting Act, the 2009 Military Voting Protection Act, the 2010 Military and Overseas Voter Empowerment Act, and other legislation regarding access to voting for military families. We suggest that military spouse voting is just as important a concern as Service member voting. Academic research regarding both Service member voting behaviors and the effectiveness of military voting programs is scarce. Teigen (2006) and Smith (2014) find that veterans and active duty Service members tend to vote at higher rates than nonveterans, but that little is known about the voting patterns of military spouses. There is, however, an inroad to developing a greater understanding of an active duty military spouse’s voting behavior.
Of all the subpopulations of American voters, why examine military spouses? First, it is imprudent to assume that active duty military spouses do not want to be engaged in political activities, and second, to assume that their votes could not affect election outcomes (Smith, 2014). As of 2017, there were 612,127 active duty military spouses (U.S. Department of Defense [DoD], 2017, p. 136), down from 725,877 in 2010 (DoD, 2010, p. 55). Although a small subpopulation of American voters, active duty military spouses are experienced in both military and civilian lifestyles and therefore may have a unique political worldview. We suggest that, especially as elections become more competitive, military spouse voting may be just as important a concern as the voting behaviors of any other population group. Further, many active duty military spouses may vote absentee, adding another layer of complexity for this unique subpopulation.
Additional support for the influence of absentee voters in recent decades has been documented. First, there have been instances in which absentee voters have tipped the voting balance away from the Election Day front-runner (Joyce, 2018). For example, the 1982 California gubernatorial race experienced a reversal of fortunes for the two leading candidates. Although Tom Bradley led by 20,000 votes on Election Day, absentee ballots reversed the race, officially giving the victory to “Duke” Deukmejian (Patterson & Calderia, 1985, p. 766). Second, the 2000 presidential election results for the state of Florida had immense consequences for the eventual winner and for the nation. In the state of Florida, George W. Bush defeated Al Gore by 537 votes; however, post-election research points to the inclusion of 680 absentee ballots that may have violated Florida election law (Imai & King, 2004, pp. 537–538). Had the absentee votes been counted differently, Gore could possibly have carried the state of Florida, and thus the presidency. Finally, several historical American elections have been won by just one vote or by tiebreaker (Bohrer & Baumann, 2018; Joyce, 2018; Montanaro, 2018). Effectively, every vote is important, and active duty military spouse votes are no exception.
Not all military spouses are absentee or overseas voters. Some may, due to their place of residence, be able to vote in person. Nevertheless, being a military spouse often entails moving and many are temporarily living in jurisdictions that require their vote be cast by absentee ballot. The Federal Voting Assistance Program (FVAP) was created to ensure that Service members and their eligible family members are aware of their right to vote and have the tools and resources to do so successfully from anywhere in the world (About FVAP, 2019). The program is designed to be easy, quick, and intuitive (DoD/Federal Information & News Dispatch [FIND], 2010) and includes a 2010 survey of voting behavior among active duty military spouses. We draw on the 2010 Post-Election Voting Survey of Active Duty Military Spouses to examine the perceptions of active duty military spouses concerning their ability to vote when not located in their formal place of residence. We derive variables from this 2010 FVAP survey to analyze the voting behaviors of active duty military spouses. Among the questions asked in the survey, an active duty military spouse’s level of voting interest and voting participation are of primary interest to this research.
Utilizing the 2010 Post-Election Voting Survey of Active Duty Military Spouses, we developed two research questions in accordance with Matsusaka’s (1995) information theory–based economic model of voter turnout. This theory posits that voters are predisposed to vote even if their preferred candidate is expected to lose, and when enough information is received regarding the election and its candidates. We first ask what factors significantly influence an active duty military spouse’s voting interest? Second, what factors significantly influence an active duty military spouse’s voting participation? From these research questions, we derive four hypotheses:
Following our analyses, we share our findings and offer potential implications from this research and, lastly, we present our conclusion.
Background and Literature Review
Peer-reviewed literature regarding the voter mobilization of military families and overseas citizens, including military spouses, is in short supply (Smith, 2014). What we can say with certainty is that absentee votes have occasionally influenced the outcomes of elections, whether it be directly (Patterson & Caldeira, 1985; Smith, 2014) or indirectly (Imai & King, 2004). This research seeks to add to the literature by increasing our understanding of active duty military spouse voting behaviors that may result in increased rates of voter interest and participation. Theories addressing how an individual chooses one candidate over another are more readily available than theories about why a potential voter decides to cast a vote at all. When considering potential determinants, such as a voter’s predisposition to vote, or the desire to seek out information about candidates or election processes, Matsusaka’s (1995) information theory–based economic model of voter turnout provides a theoretical framework based on two underlying assumptions of voter utility.
Prior to a summary of Matsusaka’s theory, it is important to explain our conceptualization of utility as it pertains to voting behavior. Matsusaka’s information theory–based model of voter turnout views the political behavior of voting through an economic lens; however, the essence of voting remains political. Therefore, political utility, derived from an external stimulation occurring from a political activity such as a political process, differs from an economic conceptualization of utility (Kenaphoom, 2018). As Matsusaka explains, this includes the consumption of information toward a political process such as voting.
Theory
Drawing from Downs (1957) for insight into voting information and Stigler and Becker (1977) for the economic effects of decision-making (i.e., price and income effects as engines of economic analysis resulting in an observed behavior), Matsusaka (1995) seeks to reinforce previous models of individual preference and utilize the perceived outcome costs of voting to explain variations in voting behavior. Prior research (Piven & Cloward, 1988; Wolfinger & Rosenstone, 1980) examined various costs associated with voting. This cost variable can represent several factors that may alter the decision-making process, including weather, health, traffic, distance to the polls, opinions of neighbors, and so on. Once all known considerations are accounted for, Matsusaka posits that a voting decision can then be made.
Matsusaka (1995) introduces readers to early attempts at explaining voter turnout via social–psychology studies, suggesting that people vote if they “develop the appropriate mental inclinations” (p. 91). Eventually, this social–psychological approach was overtaken by the rational voter approach. This approach assumes an individual, rational voter will vote according to his or her own cost–benefit analysis of the voting choice. Matsusaka posits that the rational voter model has limited utility in explaining correlations not related to voting costs; therefore, the model has demonstrated “little progress in incorporating such correlations into a common theoretical structure” (p. 92). The rational voter model suggests that increased voter turnout positively correlates with (1) increased campaign spending, (2) contact by campaign workers prior to casting a ballot, (3) public employment and/or farm ownership, (4) being married, (5) higher levels of formal education, and (6) increased age. As well, the model suggests that people who have recently moved are less likely to vote. This negative correlate is noteworthy, as FVAP resources are designed to aid absentee active duty military spouses (and others), thus aiding to counter the negative influence of relocation. In a nod to Downsian voter rationality, Matsusaka (1995) asserts that his economic model addresses those correlations not related to voting costs due to the inclusion of voter information. Matsusaka’s model makes two key assumptions. First, it posits that most citizens are predisposed to vote, believing that one should vote even if their preferred candidate is expected to lose. Second, it posits that some citizens choose not to vote in one or more election due to a perceived lack of information about one or more candidates on the ballot.
Matsusaka (1995, p. 93) states that, despite a common perception of duty to vote, individuals need to know the candidates’ policy agendas and the subsequent consequences of proposed policies. Despite this, he believes that a voter’s information and knowledge will always be incomplete. Matsusaka (1995, p. 98) explains the acquisition or purchase of information in terms of the formula: y = qi + x, where y is a person’s available income or resources, i is information, x is a consumption good, and q is the price of i in terms of x. As well, Matsusaka (1995, p. 96) notes that a strength of the information approach is that it may be formulated as a consumption theory by assuming that people feel better about the decision to vote when they are confident of voting for the right candidate. He further explains that a person receives utility from the election and from the consumption process, which includes the consumption of political information. How much information a person requires may vary according to their individual utility function, which includes voter confidence in their level of knowledge.
The Influence of Military Service on Active Duty Military Spouses
Service in the Armed Forces, for both the Service member and his or her family, includes inherent risks. According to the DoD (2019), its mission is to “provide the military forces needed to deter war and to protect the security of our country.” Even during peacetime operations, military families may serve and live overseas, change duty stations multiple times, or endure periods in which the Service member is separated from his or her family (Park, 2011, pp. 66–67). Additionally, military families may have to endure one or more deployments during times of conflict or war (Faulk et al., 2012; Spera, 2009; Van Winkle & Lipari, 2015). These examples of separation and risk create tangible stresses for military families including the potential marginalization of military spouses due to the demands of military service. One example of the marginalization of military spouses involves the negative effects of military life on spousal employment (Castaneda & Harrell, 2008; Meadows et al., 2016; Schwartz et al., 1991). However, Burrell et al. (2006) posit that being an active duty military spouse has broader impacts than just difficulties in employment.
Factors Affecting Military Voting
Within the broad influence of military life on active duty military spouses and families, several factors affect military voting. We discuss three here. First, Kohn (2002) notes that, prior to the present generation, military officers operated within a culture that favored voting abstention. This was to avoid perceptions of political partisanship among officers that might foster concerns among enlisted soldiers and thus negatively impact esprit de corps. Nix (2012) suggests that this apolitical military culture extends to enlisted soldiers as well. Second, for many military families, the ability to develop a sufficiently informed and confident voting decision may be hindered by the necessity for many military families to live away from their hometowns or overseas. This may result in reduced access to voting resources and insufficient candidate information for an individual to be confident in making an informed voting decision. Third, there is a strong history of legislation supporting Service members and their families, including voting, as demonstrated by the FVAP and other voter assistance program.
Still, questions may be raised asking if such programs are doing enough to support military family voting participation and if the programs remain vigilant in closing program gaps for those who tend to be marginalized by the status quo. A recent survey from the Military Officer’s Association of America (MOAA) finds that active duty military spouses tend to vote at lower rates than their active duty counterparts (Bushatz, 2018; Harris, 2018). The first factor is outside the scope of this research. The second and third factors are linked and are the topic of this research. One of the first steps, when exploring the impact of policy on active duty military spouse voting is to “recognize that, for these citizens, the act of casting a ballot does not take place in a voting booth, but rather in nontraditional locations and in a nontraditional fashion” (Smith, 2014, p. 9). The legislation governing absentee voting for active duty military and their spouses and families has evolved over time and is relevant to this research.
Military Voting Legislation
The first nationwide protection for Service member absentee voting came in 1942 under President Roosevelt. The 1942 Soldier Voting Act ensured that World War II Service members serving abroad could vote by absentee ballot. However, in 1944, Congress amended the act to “recommend” (rather than ensure) that states permit absentee balloting for all Service members (Leval, 2008). The Federal Voting Assistance Act of 1955 still did not guarantee Service members and their dependents the right to vote absentee, instead it simply revised its recommendations to the states (Reagan, 2016). However, the Overseas Citizens Voting Rights Act of 1975 granted absentee voting rights to Service members and their families. Next, the Uniformed and Overseas Citizens Absentee Voting Act of 1986 provided overseas voters with backup federal write-in absentee ballots (Reagan, 2016).
Since 1986, a number of federal acts have extended absentee voting protections and created federal programs to facilitate these protections (Reagan, 2016; Smith, 2014). The Help America Vote Act of (2002) directed a study of best practices for facilitating absentee uniformed service voters and ensured that Service members and overseas voters “receive timely ballots that will be properly and expeditiously handled and counted.” The Military and Overseas Voter Empowerment Act (2010) ensures expedited absentee ballot delivery for overseas Service members and establishes procedures for absentee voter registration. Nevertheless, numerous reports, corroborated by national survey data, reveal low absentee voter turnout rates and a frequent perception by absentee voters that they are disenfranchised due to inadequate ballot delivery methods (Smith, 2014). To combat this, 2016 saw the enactment of additional legislation designed to protect Service member voting rights. The Service Members Civil Relief Act (SMCRA, 2003) was amended to extend protections for Service members living away from their hometowns and included the assurance that home state residency may not be lost or that residency or resident status may not be determined by military orders.
Despite the expansion of absentee voting protections as recently as 2016, Service member spouses have not enjoyed the same level of political participation protection as their significant others. It took the Military Spouses Residency Relief Act of 2009 to afford active duty military spouses the same residency rights provided to Service members by the SMCRA, directly protecting a military spouse’s right to vote in the place they consider to be home (New Law Allows Military Spouses to Retain Original Residency, 2010; Pennsylvania Office of the Governor; Governor Rendell Expands Ballot Access for Military Voters, 2008). As well, literature regarding access to voting resources posits that the absentee voting system tends to marginalize target populations, such as minorities (Prevost & Schaffner, 2008; Treleven, 2016), the elderly (Fay, 2005; Simpson & Perez, 2000), and women (Dade, 2012). Focusing on the passage of legislation to protect Service members and eventually their spouses, other research highlights program successes (Huefner, 2013; Karp & Banducci, 2001; West Virginia, 2012) and shortcomings (Military Voting, 2007; von Spakovsky, 2005). This research expands upon those findings.
Exploration of absentee voting legislation reveals varied considerations for Service members and their spouses. Research demonstrates that absentee voting programs have had mixed results in providing adequate support for their target populations (Alvarez et al., 2007). Given the paucity of literature examining absentee voting behavior in general, and specifically that of active duty military spouses, enhancing our knowledge of absentee voting behavior may inform policy makers of new ways to improve absentee voter program efficacy. By applying Matsusaka’s information theory–based economic model of voter turnout to the 2010 Post-Election Voting Survey of Active Duty Military Spouses, we hypothesize that (1) active duty military spouses will hold greater voting interest with a more active recent voting history, (2) the receipt of election information and FVAP/DoD resources as consumption goods will enhance an active duty military spouse’s likelihood of voting participation, (3) an active duty military spouse’s voting interest will remain high despite his or her absentee status, and (4) older active duty military spouses will be more likely to vote than their younger peers.
Data and Variables
Data
This research utilizes data from the Defense Manpower Data Center’s 2010 Post-Election Voting Survey of Active Duty Military Spouses available on the FVAP website (fvap.gov). Data were obtained via web-based and handwritten surveying by the Human Resources Strategic Assessment Program (FVAP, 2010). The target population consisted of active duty military spouses of the Army, Navy, Marine Corps, Air Force, and Coast Guard. The survey’s sampling frame 1 consisted of “799,606 records drawn from the June 2010 Active Duty Master Edit File” (FVAP, 2010, p. 9). The sample size for the 2010 survey is 50,132 active duty military spouses, with 47,301 sample members deemed to be eligible. Among eligible members, 6,491 responses were determined to be usable (a 13.7% response rate); therefore, our unit of analysis for this research is the individual active duty military spouse. Qualifying active duty military spouses had to meet the following criteria—the active duty military member must have 6 months of service at the time the survey was first fielded and be below flag rank, must not be spouses of National Guard and Military Reserve members, and must be U.S. citizens (FVAP, 2011a, pp. 8–9).
While the newest active duty military spouse survey was released in 2012, only 1,370 usable surveys were returned. With some listwise deletion to be expected from partially completed surveys, use of the 2012 national survey is not advisable for analysis. Also, the 2012 data set does not include demographic characteristics of the survey population, such as age, race, and gender. In using the 2010 data set, consideration was given for the use of survey-provided weights. The 2010 survey used a single-stage stratified design in which stratification was defined according to five population characteristics—service, pay grade, race/ethnicity, age, and duty station (FVAP, 2011b, pp. 2–3). Among these five population characteristics, pay grade, (exact) age, and duty state were not provided in the published data set. Weights were “created to account for unequal probabilities of selection and varying response rates among population subgroups,” and the adjusted weights were “poststratified to match population totals and to reduce bias unaccounted for by the previous weighting steps” (FVAP, 2011b, pp. 4–5). Utilizing the “survey” package version 3.6.2 in R, the provided weights were incorporated into our analyses.
Dependent Variables
Given Matsusaka’s two key assumptions of voters’ predisposition to vote and information-based voting participation, this research seeks to illuminate determinants of active duty military spouse voting interest and participation. First, the “Interest” dependent variable is derived from the survey question asking respondents how interested or uninterested they were in the 2010 U.S. elections. Survey respondents were asked to rate their level of interest in the elections on a 5-point scale ranging from 1 (= very uninterested) to 5 (= very interested). Second, “Participation” is a binary dependent variable generated from the survey question asking respondents if they voted in the November 2, 2010, elections. Respondents were provided four possible responses—1 (= definitely voted in person), 2 (= definitely voted by mail, e-mail, fax, or online), 3 (= definitely did not vote), and 4 (= not sure). Based on these four possible responses, response values of one and two were recoded to 1 (= definitely voted in person, by mail, e-mail, fax, or online). As well, values of three were recoded to 0 (= definitely did not vote), and “not sure” responses were not included in the analyses. Table 1 provides variable descriptions for Interest and Participation.
Dependent Variables and Measures.
Primary Independent Variables
Once again, when considering Matsusaka’s two key assumptions of voters’ predisposition to vote and information-based voting participation, two of our three primary independent variables are derived from survey questions asking active duty military spouses about their plans to vote in the 2010 election and what type of information they received leading up to the election. A brief description of each primary independent is provided in Table 2. First, respondents were asked if they had “Planned to Vote” in the months leading up to the 2010 election, providing either an answer of yes or no. Second, the “Received Voting Info” index is derived from a nine-part survey question asking respondents if they received voting information from various sources in 2010. A spouse’s responses were summed then divided by nine to reflect a percentage of information sources received. Cronbach’s α for this measure was 0.76, indicating strong reliability for the items in this index. Finally, the nature of military service often requires that an active duty Service member and his or her family move away from home to serve either oversees or at their assigned U.S. duty station. Given the mission of FVAP, the “Absentee” status of an active duty military spouse is included in both analyses as a primary independent variable. An active duty military spouse’s absentee status was provided in the survey as a calculated variable as either being absentee status or not.
Primary Independent Variables and Measures.
Secondary Independent Variables
Our analyses include 12 secondary independent variables. Eleven of these variables regard voting behaviors, such as the awareness and usage of FVAP/DoD resources and recent voting history. The remaining secondary independent variable regards perceptions of online ballot safety. While both analyses share some of these secondary independent variables, they do not share all twelve. The following variable descriptions are organized according to their usage—in both analyses, in the voting interest analysis, and in the voting participation analysis. A brief description of each secondary independent variable is provided in Table 3. First, survey respondents were asked if they were referred to the DoD 2010–2011 Voting Assistance Guide for information about registering to vote or requesting an absentee ballot (“Voter Guide Referral”). Second, the “Voting Tools Access” index is derived from a seven-part survey question asking respondents if they had access to the information technology tools FVAP supports for voting participation. This index is also calculated as a percentage. Cronbach’s α for this measure was 0.86, indicating high reliability of the items in this index. Next, active duty military spouses were asked if they usually voted in national, state, and local elections (“Voted Last 6 Years”) and if they “Voted in 2008.” Last among the secondary independent variables included in both analyses, “Online Ballot Safety” is derived from a survey question asking respondents if they felt confident that completing and sending a ballot online via the Internet is safe, secure, and reliable.
Secondary Independent Variables and Measures.
Note. FVAP = Federal Voting Assistance Program; VAO = voting assistance officer; IVAO = Installation Voting Assistance Office; UVAO = Core Unit Voting Assistance Officer.
For the voting interest analysis, an active duty military spouse’s awareness of available resource, especially if the resources are perceived to improve convenience, may influence his or her level of voting interest. These four variables include an active duty military spouse’s awareness of the FVAP website (“Aware: FVAP website”), the FVAP toll-free telephone service (“Aware: FVAP Hotline”), the DoD voting guide (“Aware: Voting Guide”), and the availability of information or assistance from a DoD or Department of State voting assistance officer (“Aware: VAO”). For the voting participation analysis, three variables are included regarding respondents’ utilization of available voting resources. When considering voting participation, the ability to use available resources may influence a spouse’s decision to participate. These three variables include respondent’s use of the FVAP website (“Used: FVAP Website”), the FVAP toll-free hotline (“Used: FVAP Hotline”), and the use of information or assistance from a DoD or Department of State voting assistance officer (“Used: VAO”).
Six demographic and military characteristic variables were included in each analysis. First, among the demographic variables, “Age” was provided according to five categories. These categories, along with brief descriptions of all demographic and military characteristic variables, are provided in Table 4. Next, a survey calculated categorization of respondents’ “Race” was provided according to two categories—White and non-White. Third, “Gender” was categorized as either male or female. Despite the male dominance of military service, the sizable pool of 6,491 respondents warrants the inclusion of this variable. Regarding the military characteristic variables, the “Deployment” status of a spouse’s Service member during the 2010 was requested with a simple yes-or-no response. Next, two survey calculated variables regarding military characteristics were included. We use the active duty military commissioned/enlisted status (“Commission”) as a proxy for socioeconomic status. This variable is recoded as a binary variable where zero equals enlisted and one equals officer. Finally, “Branch” of service is included as a factor variable using the Coast Guard as the reference category. Among the five branches, the Coast Guard was chosen as the reference category solely on its service in the Department of Homeland Security, while the remaining four branches reside within the DoD.
Demographic and Military Characteristic Variables and Measures.
Lastly, a summary of the representativeness of select demographic and military characteristic variables utilized provides additional context. To put the 2010 Post-Election Voting Survey of Active Duty Military Spouses survey data into context, it is compared against the 2010 Demographics Report: Profile of the Military Community and 2010 U.S. Census Bureau data. Beginning with gender, the 2010 FVAP survey of 6,491 usable responses is comprised of 656 (10.1%) male active duty military spouses. The 2010 demographic report, assuming predominately male–female marriages, states that 94,290 of the 798,921 (11.8%) married active duty Service members are female (DoD, 2010, p. 56). When examining the racial composition of the 2010 FVAP survey, 2,160 of the 6,491 (66.7%) respondents reported as being White or Caucasian. The U.S. Census Bureau (2011) reported that 72% of Americans identified as White alone in 2010. No data for the race of active duty military spouses was provided in the demographics report. Active duty Service members are similar, at 70% White (DoD, 2010, p. 20). Finally, different but similar age ranges were used by the FVAP survey and the demographics report, which is as follows: 29.8% are 25 or younger, 24.6% are 26–30, 18.3% are 31–35, 14.0% are 36–40, and 13.3% are 41 or over (DoD, 2010, p. 55). The 2010 FVAP survey is as follows: 11.7% are 18–24, 18.1% are 25–29, 18.7% are 30–34, 32.6% are 35–44, and 18.6 are 45 or over.
Results
Voting Interest
Our first dependent variable of voting “Interest” is derived from a 5-point ordered survey question; therefore, Table 5 presents the ordered logistic regression results that estimate associations between active duty military spouse voting interest and select voting behaviors, demographics, and military characteristics. We hypothesized that voting interest would be positively influenced by behaviors indicative of a preference for voting. With this in mind, a few significant voting behaviors stand out. First, receiving more voting information, B = 0.79, SE = 0.22, p < .001; planning to vote, B = 2.15, SE = 0.13, p < .001; absenteeism, B = 0.37, SE = 0.11, p < .001; and voting in the past 6 years, B = 0.42, SE = 0.14, p < .01, are positively associated with increased active duty military spouse voting interest. We found no significant association between any awareness of FVAP or DoD voting resources, access to voting tools, perceptions of online ballot safety, or participation in (only) the 2008 election. Regarding demographics, only greater age, B = 0.30, SE = 0.05, p < .001, is associated with increased active duty military spouse voting interest. We found no association between active duty military spouse voting interest and select military characteristics. Among the select voting behaviors, receiving voting information, planning to vote, and a history of voting within the previous 6 years would seem to be indicative of a predisposition to vote.
Ordered Logistic Regression of Voting Interest.
Note. SE = standard error; DV = dependent variable.
*p < .05. **p < .01. ***p < .001.
Voting Participation
Our second dependent variable of voting “Participation” is a binary, nonnormally distributed dependent variable. This analysis categorizes military spouses into two self-reported groups as either registered or unregistered voters. Given the binary nature of voting participation, our second analysis only includes respondents who indicated that they were registered to vote during the 2010 elections. To incorporate survey weights into this analysis, we conduct quasi-binomial logistic regression using the “survey” package verion 3.6.2 in R. Table 6 presents the quasi-binomial logistic regression results that estimate associations between active duty military spouse voting participation and select voting behaviors, demographics, and military characteristics. For this analysis, demographic and military characteristic variables remain the same; however, a different set of voting behaviors was selected to reflect actions suited to participation rather than interest (e.g., FVAP resource usage rather than awareness). Here, we hypothesized that factors indicative of individual perceived utility would influence an active duty military spouse’s decision to vote.
Quasi-Binomial Logistic Regression of Voting Participation Among Registered Active Duty Military Spouses.
Note. SE = standard error; AIC = Akaike information criterion.
*p < .05. **p < .01. ***p < .001.
Several voting behaviors returned significant and positive associations with increased active duty military spouse voting participation, including receiving more voting information, B = 0.27, SE = 0.05, p < .001; planning to vote, B = 0.27, SE = 0.02, p < .001; absenteeism, B = 0.07, SE = 0.02, p < .001; using the FVAP website, B = 0.14, SE = 0.03, p < .001; voting in the past 6 years, B = 0.07, SE = 0.02, p < .01; voting in 2008, B = 0.1, SE = 0.02, p < .001; and concerns about online ballot safety, B = −0.03, SE = 0.003, p < .01. Further, the negative relationship between concerns about online ballot safety indicates that active duty military spouses more likely to vote were unlikely to have concerns about online voting security. Among the demographic variables, only age, B = 0.04, SE = 0.01, p < .001, returned a significant association with active duty military spouse voting participation. Finally, commission, B = 0.1, SE = 0.02, p < .001, and active duty military service in the Army, B = −0.09, SE = 0.03, p < .01, returned significant associations with military spouse voting participation. As well, the negative relationship between active duty military service in the Army and voting participation indicates that, given the Coast Guard as the mediating variable, active duty Army spouses are less likely to vote.
Considering the significant association between active duty military spouses receiving voting information and their participation, we estimate the probability of turnout using two methods. First, Figure 1 presents the effects plot predicting active duty military spouse turnout according to the amount of voting information received. Generally, as more voting information is received, we can predict that an active duty military spouse is more likely to vote, which coincides with what we found in the regression results seen in Table 6. As well, the confidence band suggests that, as active duty military spouses receive less voting information, we can be more confident that they are unlikely to vote. For active duty military spouses receiving more voting information, the confidence band widens. This widening effect indicates that greater voting information is not as clear an indicator of participation as lesser voting information is for nonparticipation. Second, we use the “predict” function in R to predict the probability of specific types of active duty military spouse cases. Type 1 is a low-voting information active duty military spouse who has received information from 1 of the 10 possible sources. The probability that a low-voting information active duty military spouse will vote is 0.169 (SE = 0.011), meaning a low-voting information active duty military spouse will participate in voting 17% of the time. Type 2 is a high-voting information active duty military spouse who has received information from nine or ten possible sources. A high-voting information active duty military spouse will probably participate in voting 53% of the time or a probability of 0.525 (SE = 0.035). Therefore, while receiving high levels of voting information does not ensure participation, our analyses demonstrate that it significantly increases the likelihood of participation.

Participation among registered active duty military spouses by voting information received.
Discussion and Conclusions
Discussion
This research addresses a gap in our understanding of the motivating factors driving active duty military spouse voting interest and participation. Hypothesis 1 posits that an active duty military spouse’s recent voting activity will be linked to his or her level of voting interest. We find support for this hypothesis in the significant association of voting interest and having voted within the last 6 years. Respondents who had voted in the previous 6 years, though not necessarily in 2008, indicated a clear interest in voting in the 2010 election. Given this and the significance of respondents’ plans to vote, the apparent predisposition to vote lends support for Matsusaka’s first key assumption.
Hypothesis 2 contends that the receipt of election information and FVAP/DoD resources as consumption goods will enhance an active duty military spouse’s likelihood of voting. We find support for this hypothesis on two fronts. First, the receipt of voting information pertaining to the election and election processes significantly supports voting participation. Second, respondents were likely to utilize the FVAP website as an aid toward the eventual decision to vote in the 2010 election. As previously stated, Matsusaka asserts that the consumption of election information is a consumption good by which voters derive utility toward the decision to participate via voting. For active duty military spouses, this includes FVAP and DoD resources designed to assist and educate potential voters in the election and its processes. Given the support for our first two hypotheses, we find that the two key assumptions of Matsusaka’s information theory–based economic model of voter turnout hold up well when considering the results of our analyses.
Our Hypotheses 3 and 4 take a different approach to Matsusaka’s theory. Hypothesis 3 suggests that an active duty military spouse’s voting interest will remain high despite his or her absentee status. This strikes at the heart of Matsusaka’s first key assumption while considering FVAP efforts to combat the negative influences of relocation on voting participation. As active duty military spouses may often qualify for absentee status, FVAP resources may or may not influence their predisposition to vote. The significant association of voting interest and participation with absenteeism provides support for Hypothesis 3. While awareness of FVAP and DoD voting resources is not associated with voting interest among the survey’s population, we find that a predisposition to vote and the receipt of voting information are. A possible explanation for such results may be found within Matsusaka’s assertion that Americans tend to perceive voting as a duty. If this is the case, other motivations, such as interest, may take a back seat to duty for some active duty military spouse voters. Finally, we find support for Hypothesis 4, as older survey respondents were more likely to report voting interest and participation than their younger peers. This aligns with the traditional rational voter model as one of Matsusaka’s (1995) nonvoting cost-related correlates.
The lack of significant association for certain variables is also worth brief discussion. Many of the FVAP and DoD resources included in our analyses, whether it be awareness of or utilization of resources, did not appear to influence the voting interest or participation of active duty military spouses. These results raise questions about the efficacy or availability of the voting resources available to Service members and their families. Do these results suggest opportunities to realign program resources for greater convenience to the target population? As well, does the lack of awareness of these resources suggest an opportunity to improve current information and education campaigns? If the results of the MOAA survey are any indication, there may be support for such inferences. Further research and the revitalization of the FVAP’s Post-Election Voting Survey of Active Duty Military Spouses may aid in answering these questions.
Finally, much has been discussed regarding factors that influence voting interest and participation among active duty military spouses. With respect to those in this subpopulation who choose not to vote, there may be several reasons why an active duty military spouse decides not to vote. The 2010 survey asks respondents to identify their main reason for not voting in the 2010 election. Table 7 displays the results from active duty military spouses with usable survey responses (Question 50). The data show that 34.8% of active duty military spouses were either not interested in voting or they had no candidate preference. Excluding the generic response of “some other reason,” a lack of voting interest and candidate preference comprised the top two response categories. These results appear to lend further support for Matsusaka’s (1995) two key assumptions that some but not all citizens have a predisposition to vote and that some will elect to not to vote due to a perceived lack of information about one or more candidates.
Reasons Active Duty Military Spouses Did Not Vote in the 2010 Elections.
Limitations
The limitations of this research begin with the 2010 data set. While the 2010 FVAP survey is several years old, it is currently the most serviceable of any FVAP active duty military spouse survey. As well, the sample size of usable survey responses is not a representative one. Often, this concern may be addressed with the incorporation of post-stratification weights. While Losinger (2010) cautions against stratification of the sample within the 2006 survey, the 2010 Post-Election Voting Survey of Active Duty Military Spouses survey does report its stratification methods and includes the post-stratification weights in the data set. These weights were utilized in our analyses; however, the limitations of not using post-stratification weights are assumed to outweigh the limitations of using post-stratification weights.
Second, several independent variables common to political science and election research were not available within the published data set. Among those unavailable for testing were political party affiliation, political ideology, and education. As well, six of the seven traditional rational voter model correlates that are not voting cost-related were not testable using the 2010 survey. Only age was available within the data set. Given these limitations, we recommend future FVAP post-election voting surveys of active duty military spouses include such data.
Next, active duty military spouses, historically, are disproportionately female (Ziezulewicz, 2009). Given this phenomenon, the demographic composition of military spouses in 2010 was 93.1% female (DoD, 2010, p. 56). While this number has dropped to 91.5% as of 2017 (DoD, 2017, p. 136), the ratio of female active duty military spouses is much higher than the 2010 national ratio of 50.8% (U.S. Census Bureau, 2010, p. 2). Considering these figures, the proportion of female active duty military spouses in the 2010 survey (based on usable responses) is 89.9%. This distribution more closely represents the proportion of female active duty military spouses than it does the proportion of females in the United States.
Finally, though the 2010 FVAP survey’s response rate of 13.7% may be satisfactory for other survey collection efforts, the relatively low number of absentee active duty military spouses participating in the survey deserves consideration. If there are barriers to active duty military spouse participation in such a survey, such barriers would ideally be addressed to collect a greater and more representative proportion of survey respondents within the target population. This, as well, raises two questions. Is there a correlation between low survey participation and low voter turnout in this subpopulation? Do many active duty military spouses feel disenfranchised by a history of underserving voting protections? A more comprehensive survey, whether conducted by the FVAP or by others, and further research may provide additional insights for policy makers seeking to mitigate lingering participation barriers.
Conclusions
The American voting process is a time-honored tradition. When examining the voting behaviors of active duty military spouses through the lens of voting behaviors, demographics, and military characteristics, patterns emerge that may reinforce existing efforts to provide military families and overseas personnel with the cutting-edge resources needed to conveniently participate in this time-honored tradition. The need for innovative support and engagement policies and processes is clear; however, is it enough? The results suggest a few areas for further consideration. First, voting interest and participation among active duty military spouses tend to increase when voting behaviors are reported that could be considered proactive. How far should programs such as FVAP go in educating its target population on election processes and candidates? If the answer includes anything beyond what is currently done, there may be a correlating increase in military spouse voter participation, according to Matsusaka.
What do our conclusions mean for the process at large? Matsusaka’s information theory–based economic model of voter turnout does appear to hold up well when considering the 2010 Post-Election Voting Survey of Active Duty Military Spouses. However, Matsusaka’s theory that some individuals participate in elections because of a strong sense of civic duty presents a conundrum. Why does voter turnout remain so low? Why do some active duty military spouses choose not to vote? Again, more research on the voting behaviors of active duty military spouses is needed to answer these questions.
Due to the influence of proactive voting behaviors on voting interest and participation, there may be some support for a fresh perspective on existing absentee voter policies and processes. As we have shown, recent efforts from Congress and the Executive Branch have mandated support designed to engage military families and overseas personnel in a more modern and convenient voting process. The results of the MOAA survey suggest a need for enhanced active duty military spouse voter education (Bushatz, 2018). Our research agrees with the MOAA findings and suggests that this process has room for improvement. If the information theory–based economic model of voter turnout is at least partially representative of proactive voting behavior, continued efforts to provide innovative and diverse engagement should be targeted to enhance voter education, interest, and participation. Additionally, as voting interest and participation are generally found to not be dependent on FVAP and DoD voting resources, renewed efforts to better inform voters about best practices for successfully navigating the system may improve both program efficacy and the efficient allocation of voting resources.
Footnotes
Acknowledgment
This article is dedicated to Deborah J. Button, a 20-year military spouse.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
