Abstract
The appropriate format for services supporting military families depends on how vulnerabilities and resources are distributed across and within those families. If different types of vulnerabilities cluster together, then programs supporting families should combine multiple services rather than targeting specific concerns. Yet scant data exist about how vulnerabilities and resources covary within military families. The current study addressed this issue through a latent class analysis of data on a wide range of domains obtained from a stratified random sample of 1,981 deployable, active component, married servicemembers and their spouses. Within married deployable servicemembers, results indicated that vulnerabilities and resources cluster together within individuals; servicemembers at high risk in one domain are likely to be high risk in multiple domains. This is less the case for spouses. One or both spouses are vulnerable in 39% of couples. These results support programs that provide vulnerable military families with more comprehensive services.
Over a decade since operations began in Afghanistan in 2001, the military force of the United States is largely composed of servicemembers (SMs) who have experienced at least one overseas deployment. According to the Department of Defense, almost 1.8 million active component SMs have deployed in support of military operations overseas (Defense Manpower Data Center, 2012), and roughly one third of those deployed between 2001 and 2011 have experienced a cumulative deployment time of 2 years or more (Baiocchi, 2013). Moreover, as military families have confronted the stress associated with deployments, they also confront the additional stressors that many nonmilitary families face, such as economic strain, stressful life events, and relationship problems (Karney & Bradbury, 1995). Several theories of marital relationships (e.g., Karney & Bradbury, 1995; McCubbin & Patterson, 1983) and mental health (e.g., Billings & Moos, 1982; Hobfoll & Walfisch, 1986) propose that individuals and families with greater levels of current stress or poorer mental health are less equipped to deal effectively with future stressful experiences. In other words, these individuals and families are vulnerable, and future stressors are more likely to push them toward declining mental health and family crisis than those who do not possess these vulnerabilities.
Yet military families are also likely to possess resources that may buffer them from stress, such as social support networks of other military families who are experiencing similar challenges (Hobfoll et al., 1991). Because individuals and families with more vulnerabilities and fewer resources are more likely to have negative outcomes when exposed to stress (Maisel & Karney, 2012; Randall & Bodenmann, 2009; Rauer, Karney, Garvan, & Hou, 2008), the balance between vulnerabilities and resources within military families may determine how much they are at risk for future negative outcomes (e.g., mental and physical health problems, divorce) in the face of the continuing demands of military life.
To ameliorate the effects of these demands, the military has a strong interest in providing support services to SMs and their families, but the most appropriate format for those services is likely to depend on how different vulnerabilities and resources are distributed across the population of military families. For example, if distinct vulnerabilities are relatively independent from each other, then support programs would be best advised to target specific vulnerabilities, as individuals or families with one vulnerability would be unlikely to have other concurrent vulnerabilities. In contrast, if different vulnerabilities generally cluster together, efforts to support military families would be best advised to provide more comprehensive services to families, rather than focusing on specific vulnerabilities.
Currently, military programs are not guided by this information because scant data exist about the extent to which military families possess different vulnerabilities and resources at the same time. Mental health problems such as posttraumatic stress disorder (PTSD) are the most publicly discussed vulnerabilities that SMs face, but it is not clear that SMs with more mental health problems also possess other family vulnerabilities, such as low marital quality, poor parent–child relationships, or financial stress. Moreover, there have been few attempts to document the levels of different vulnerabilities and resources among SMs and their families. The goal of the current study is to address these issues and in so doing help the military appropriately design and target programs to promote resilience in military families.
Domains of Vulnerability and Resources in Military Families
Family stress models have long proposed that how a family reacts to stressful events and circumstances is shaped by preexisting conditions that constrain or facilitate a family’s ability to cope with stress effectively (McCubbin & Patterson, 1983). Although preexisting vulnerabilities do not guarantee that a family will experience stress-related adverse outcomes, it does make them more likely to have difficulty dealing with negative life events that place stress on the family. Having fewer vulnerabilities and greater levels of resources, in contrast, makes couples and families more likely to deal with negative events and marital stress effectively (e.g., Karney & Bradbury, 1995). Existing models (e.g., Karney & Crown, 2007) suggest a set of specific domains that are likely to be important in identifying families who might benefit from military support programs and interventions. Below, we briefly review the literature on each of these domains: enduring traits, family dynamics, military experiences, and nonmilitary circumstances.
Enduring Traits
Enduring traits are qualities of family members that tend to persist over time. These traits include previous experiences (e.g., childhood neglect or abuse) and long-term mental and physical health status (e.g., depression, chronic illness). Prior research has demonstrated that these factors can shape how SMs and their spouses (SPs) react to stress. For example, SMs reporting greater numbers of adverse childhood experiences and more preexisting mental health problems prior to deployment developed more PTSD symptoms after deployment to Afghanistan than SMs who did not report preexisting vulnerabilities (Berntsen et al., 2012). After combat exposure, SMs with preexisting mental health conditions and a history of childhood adversity exhibit PTSD symptoms for a longer period of time than SMs with either preexisting vulnerabilities or combat exposure, but not both (Dohrenwend, Yager, Wall, & Adams, 2013). Together with consistent research on civilian populations (see Liu & Alloy, 2010, for a review), the accumulated literature offers strong evidence that SMs and SPs with prior mental health problems or adverse childhood experiences are more vulnerable for experiencing stress-related problems than are those without these enduring traits, and so may be suitable targets for additional support.
Family Dynamics
Family dynamics refer to the nature of the relationships among family members. When those relationships are close and positive, family members can support each other to cope effectively with stressful events. When their relationships are distressed or high in conflict, they drain psychological resources that might otherwise be directed elsewhere. Research on the implications of positive or negative family relationships has been conducted most frequently on civilian families, but this research consistently demonstrates the numerous physical and emotional benefits of positive relationships for spouses (e.g., Waltz, Badura, Pfaff, & Schott, 1988) and their children (e.g., David, Steele, Forehand, & Armistead, 1996). To the extent that these benefits accrue in military families, then the quality of the marital relationship would be an important domain to assess for identifying vulnerable families.
Military Experiences
The demands of military life may be borne more easily when SMs are more committed to and satisfied with military life. For these SMs, commitment to military life may serve as a resource, helping them to view specific challenges as necessary sacrifices in pursuit of a worthwhile goal. Indeed, both commitment and satisfaction are associated with increased probability of SMs retention (Farkas & Tetrick, 1989), and a meta-analysis of studies of organizational commitment in civilian populations found that increased commitment was associated with less job stress and with higher rates of job retention (Mathieu & Zajac, 1990). When commitment to and satisfaction with military life are low, however, it may be harder for SMs and their families to justify the sacrifices required by military life, and so more difficult to generate effective coping responses. Indeed, recent research has found that low commitment to one’s branch of service and low satisfaction with military leadership are associated with greater likelihood of psychiatric diagnosis (Booth-Kewley et al., 2013). Thus, the way SMs evaluate their military experiences is an important factor in predicting their outcomes.
Nonmilitary Circumstances
In addition to the unique demands and rewards of military life, military families also confront the same challenges as civilian families. One of those challenges is financial strain. Among civilian families, financial strain is associated with increased parent depression and poor adolescent child adjustment (Conger et al., 1992, 1993). Military families with adequate finances are better equipped to avoid these outcomes and may cope more effectively with the demands of military life. A second challenge is the absence of social support (Thoits, 1986). Among civilian couples, increased access to social support buffers the negative effects of financial strain on marital quality (Conger, Rueter, & Elder, 1999), whereas the lack of social support exacerbates those effects (Vinokur, Price, & Caplan, 1996). Within the military, isolated families may cope less effectively than families who are connected to a supportive network. A third challenge is the presence of stressful life events. Among civilians, stressful life events are associated with the onset of depression (Kessler, 1997) and negative relationship outcomes (e.g., Bolger, Foster, Vinokur, & Ng, 1996). In sum, military families who lack financial security, social support, and who experience stressful events may be appropriate targets for additional support.
The Structure of Vulnerabilities and Resources
If a family possesses vulnerabilities in one of the above domains, are they also likely to possess vulnerabilities in other domains? Among civilian families, the answer appears to be “yes,” for two reasons. First, different domains of vulnerabilities frequently covary. Individuals who are disadvantaged in one way (e.g., lower income) are more likely to be disadvantaged in other ways (e.g., poorer mental health; Maisel & Karney, 2012). Second, vulnerabilities tend to accumulate. Established theories of family stress suggest that over time vulnerabilities in one domain have negative consequences for other domains, creating a downward spiral (Rutter, 1987) or a “pile-up of family demands” (McCubbin & Patterson, 1983). As a consequence, civilian families are generally characterized by distinct vulnerability profiles within and across family members, such that some have higher levels of vulnerability across multiple domains, whereas others are characterized by relatively low levels of vulnerability across multiple domains (e.g., Radke-Yarrow & Brown, 1993; Repetti, Taylor, & Seeman, 2002; Zweig, Phillips, & Lindberg, 2002).
Yet, despite such evidence, it is not clear that the same clustering should be observed among military families, for several reasons. First, although many people who join the military come from disadvantaged backgrounds (e.g., low socioeconomic status families; Kleykamp, 2006), they are generally physically and psychologically healthy when they enter the military since those recruits with serious mental illness, criminal histories, poor physical health, and addictions are explicitly excluded from service. In the absence of these more severe vulnerabilities, SMs may be less likely to accumulate other types of vulnerabilities (e.g., poor marital relationships). Second, many of the most severe stressors affecting military families are exogenous—that is, stressors such as military relocations and deployments are not products of the family system—and so are not inherently linked with vulnerabilities that are endogenous to individual family members. Third, whereas civilian couples typically experience similar mental health problems as a result of assortative mating and shared circumstances (Meadows, McLanahan, & Brooks-Gunn, 2007), some mental health problems among SMs may result from experiences they do not share with their SPs (e.g., combat), limiting the likely associations among vulnerabilities across family members. For these reasons, whether or not distinct domains of vulnerability or resources are likely to cluster together within military families remains an open question.
Overview of the Current Research
The goal of the current analysis was to address these issues by surveying a stratified random sample of deployable, active component, married SMs, and their SPs. For both spouses, we assessed four domains of vulnerabilities that current models of military families (e.g., Karney & Crown, 2007; Tanielian & Jaycox, 2008) suggest are relevant for understanding outcomes in military families: mental and physical health problems, adverse childhood experiences, family vulnerabilities (i.e., poor relationship quality and family environment), and nonmilitary circumstances (i.e., financial strain, lack of social support, and stressful events). For SMs, we also assessed evaluations relevant to prior military experiences (i.e., satisfaction with the military and commitment to the military).
We then used latent class analysis (LCA) to identify distinct profiles across these domains. Examining these profiles allowed us to address three specific questions. First, how do different domains of vulnerability covary within SMs and within SPs (i.e., does vulnerability in different domains cluster together within individuals or are vulnerabilities relatively independent)? Second, what is the prevalence of vulnerability to adverse outcomes in the deployable active component population of married SMs and SPs (i.e., what proportion of SMs and SPs are vulnerable)? Finally, what is the association between SPs’ vulnerability profiles within a couple (i.e., are more vulnerable SMs likely to be married to more vulnerable SPs)?
This study is the first of which we are aware to examine profiles of military couples based on multiple domains of vulnerability and resources, so there was no basis for strong a priori hypotheses about the composition of these profiles. Nevertheless, since most of the population of SMs has already experienced multiple deployments (Tanielian, Karney, Chandra, Meadows, & Deployment Life Study Team, 2014), theories that highlight the accumulation of vulnerabilities over time, and the strong evidence of covariance among vulnerability domains within the civilian population support the tentative prediction that different vulnerabilities should cluster within military couples as well.
Method
Sample
The data described here were drawn from the initial assessment of the Deployment Life Study in 2012. The Deployment Life Study is a longitudinal survey of the deployable, married population of the military, their SPs, and their children, stratified by service component. Married SMs and their SPs were randomly sampled within service components, and the final sample was weighted to reflect the population of married SMs eligible to deploy in 2012 (see Tanielian et al., 2014), for additional information about the study design and sampling frame). Data collection was ongoing at the time of this writing. The current study is meant to be a snapshot of the levels of vulnerabilities and resources among deployable SMs and their SPs. The profiles of vulnerabilities and resources revealed in the current analysis will eventually be used to assess family resilience to deployment stressors over time. Thus, complete longitudinal data are not necessary for the current analysis.
Sampled couples were invited to participate in the survey, and only those couples where both the SM and SP completed the survey were included in the data analysis. A total of 2,724 SMs and their families were sampled, 26 of whom were already deployed at baseline and were deleted from the data, leaving 2,698 households. To reduce variance associated with heterogeneity across active and reserve/guard components, the current analyses were restricted to active component SMs in the Air Force, Army, Marines, and Navy and SPs. Of the remaining 1,981 households, 92.7% of SMs were male. With respect to race and ethnicity, 74.8% of the SMs were White (non-Hispanic), 9.1% were Black (non-Hispanic), 11.7% were Latino, and 4.3% indicated “Other.” With respect to current service branch, 43.7% served in the Navy, 38.1% in the Army, 12.7% in the Air Force, and 5.6% in the Marine Corps. Of the SPs, 7.3% were male, 73.4% were White (non-Hispanic), 7.7% were Black (non-Hispanic), 12.1% were Latino, and 6.7% indicated “Other.” At the time of the survey, couples had been married an average of 7.0 years. Just over 75% of couples reported having one or more children in the household, and among those households, the average number of children was two. On average, SMs had been in the military for 9.7 years and had experienced 3.1 years overseas deployments. For complete documentation of the study, see Tanielian et al. (2014).
Measures
All measures described below were administered via telephone or via an online survey. SMs and SPs provided their data separately. To simplify the presentation of the results, we coded all variables so that higher numbers reflected greater vulnerability.
Enduring Traits
SMs and SPs completed measures about their own mental and physical health and their adverse childhood experiences. Mental and physical health were assessed using four scales. The first was the 4-item anxiety subscale of the Mental Health Inventory–18 (Sherbourne, Hays, Ordway, DiMatteo, & Kravitz, 1992; α = .84 for SPs and .84 for SMs). The second was the Patient Health Questionnaire (PHQ-8), an 8-item measure of depressive symptoms based on the DSM-IV criteria for depressive disorders (Kroenke, Spitzer, & Williams, 2001; Kroenke et al., 2009; α = .82 for SPs and .81 for SMs). The third was a 4-item measure of physical health derived from the Short Form Survey (SF-12; Ware, Kosinski, & Keller, 1996) and the National Health and Interview Survey (Idler & Angel, 1990). The items focused on the degree to which the respondent is limited by his or her physical health, whether his or her health status limits the ability to work or attend school, how much physical health interferes with normal social activities, and the respondent’s general physical health (α = .80 for SPs and .78 for SMs). Finally, SMs who reported having experienced a traumatic event (58.8% of the sample) completed the PTSD checklist–Specific (PCL-S), a 17-item scale assessing how much respondents were bothered by PTSD symptoms in the past 30 days (α = .78; Weathers & Ford, 1996). In place of the PCL-S, SPs completed the Primary Care PTSD (Prins et al., 2004), which uses a screen item to “cue” respondents to traumatic events and then asks whether four specific symptoms have occurred in the past 30 days (α = .80). All four scales were coded so that higher numbers reflect poorer health. Each scale was standardized and averaged to form an index of poor mental/physical health (α = .76 for SPs and .76 for SMs).
We measured negative events in childhood using five items taken from the Adverse Childhood Experiences Scale (ACE; Anda et al., 1999). These items assessed whether respondents experienced abuse or neglect in childhood (e.g., feeling physically threatened, did not have enough to eat, had a parent who went to prison). Respondents indicated whether or not they had each experience in childhood; endorsements were summed to form a cumulative index.
Family Dynamics
SMs and SPs completed measures of their perceptions of marital satisfaction and, if they had children in the household, their perceptions of the quality of their family environment. Marital satisfaction was assessed using nine items adapted from a set of items used previously in the Florida Formation Survey (Rauer et al., 2008). These items measure different aspects of relationship quality including satisfaction with time spent together, communication, trust, and overall satisfaction with the relationship. Scale items were recoded so that higher scores equal more distressed relationships, standardized so that all items were on the same metric, and combined (α = .84 for SPs and .81 for SMs).
We assessed the quality of the family environment using six items taken from the Conflict and Cohesion subscales of the Family Environment Scale (FES; Moos & Moos, 1994). The cohesiveness items were reverse-scored and combined with the conflict items (α = .71 for SPs and .66 for SMs). For families with children (n = 1,491), the FES and relationship quality scales were combined to form one indicator of family stress (α = .71 for SPs and .69 for SMs). As the FES is constructed for families rather than couples, marital satisfaction was the sole indicator of family dynamics for couples with no children (n = 490).
Military Experiences
SMs (but not SPs) reported on their commitment to the military and their satisfaction with military life. Commitment to the military was measured with a 5-item scale used in the Defense Manpower Data Center’s Status of Forces Survey, itself adapted from O’Reilly and Chatman’s (1996) organizational commitment scale. The scale includes items such as willingness to “make sacrifices to help” their branch of service, the extent that serving inspires them to do the best job they can, and the extent that they like being in their military branch because their values are similar to those of the branch (α = .78). Satisfaction with military life was measured using a single item taken from the Defense Manpower Data Center’s Status of Forces Survey: “Generally, on a day-to-day basis, how satisfied are you with the military way of life?” Responses to both scales were combined (α = .69), with higher scores indicating more negative evaluations of prior military experiences.
Nonmilitary Circumstances
SMs and SPs completed questions assessing economic strain, nonmilitary stress, and social support. Economic strain was measured with four items from the Gutman and Eccles (1999) financial strain scale: whether the family could make ends meet, whether there has been enough money to pay bills, whether there was any money left over at the end of the month, and the level of worry caused by the family’s financial situation (α = .87 for SPs and .86 for SMs). We measured nonmilitary stress using 15 items adapted from the List of Threatening Experiences (Brugha, Bebbington, Tennant, & Hurry, 1985). For this measure, respondents indicated whether they had experienced each stressor (e.g., the death of a parent) and we treated the number of affirmative responses as a cumulative index. We measured social support using three measures, each addressing a different aspect of support: number of different sources of support, instrumental support, and expressive support. The number of different sources of support available to SMs and SPs was measured using nine items asking respondents how much they felt they could depend on different sources of support (e.g., one’s own family, SP’s family, other military families; α = .71 for SPs and .79 for SMs). We assessed instrumental support using four items, taken from the Fragile Families and Child Wellbeing Study (Reichman, Teitler, Garfinkel, & McLanahan, 2001), asking whether the respondent had enough people they could count on to help with housing, medical care, child care, and finances (α = .74 for SPs and .77 for SMs). Expressive support was assessed with a single item asking whether respondents have someone to listen to their problems. Responses to each of these scales were recoded when necessary so that higher numbers represent greater isolation (i.e., less support) and then standardized and combined to form a single index of stressful nonmilitary conditions (α = .64 for SPs and .62 for SMs).
Demographics
SMs and SPs also provided basic demographic information, including service, gender, race/ethnicity, parental status, and age.
Analytic Strategy
To identify distinct profiles of vulnerabilities and resources that characterize SMs and SPs, we used LCA, an exploratory procedure that groups individuals into classes with similar patterns of scores across the variables assessed in the model. In contrast to factor analysis, LCA allows for a person-centered rather than a variable-centered approach to estimating latent constructs (Marsh, Lüdtke, Trautwein, & Morin, 2009). Since higher levels of vulnerabilities are associated with higher risk for future negative outcomes (e.g., mental health problems, divorce), the latent constructs estimated in the current analysis are risk propensities of SMs and SPs. In addition, we extended this person-centered approach to families. LCA models have been recommended for use in family research to “characterize parents and families along multiple dimensions of theoretical interest, producing types of families” (Henry, Tolan, & Gorman-Smith, 2005, p. 126). The number of classes is determined through measures of fit (Akaike information criterion [AIC], Bayesian information criterion [BIC], sample size adjusted BIC [SSBIC], and entropy) and interpretability.
We first developed separate LCA models for SMs and SPs, using Mplus 6.11 (Muthén & Muthén, 2011). If different vulnerabilities and resources are independent within individuals, then classes for SMs and SPs will be composed of individuals with both high and low levels of vulnerability across different domains. If different vulnerabilities and resources are not independent, then classes will be composed of individuals who systematically score high or low across domains. In the second stage of the analysis, we examined the degree to which partners within a couple had similar or different vulnerability profiles. If the vulnerabilities of SMs and SPs are independent, then SMs with low levels of vulnerability should be no more likely to be married to SPs with low vulnerabilities than to SPs with greater vulnerabilities.
Results
Latent Class Analysis for SMs
The LCA for SMs resulted in a five-class model that demonstrated the best fit with the data relative to other class solutions: AIC = 17,499, BIC = 17,795, SSBIC = 17,627, entropy = .84. A six-class solution offered higher entropy (.93), but the other fit indices were poorer (AIC = 21,633, BIC = 21,856, SSBIC = 21,729). The complete set of fit indices is available on request.
Overall, the five classes were distinguished by SMs’ level of vulnerabilities or resources across all indicators. As shown in Figure 1, Class 1 (“Low Risk”) included SMs whose scores were relatively low on all five indicators, and are thus at low risk for future negative outcomes (e.g., mental and physical health problems, divorce) in the face of the continuing demands of military life. Over half of SMs (50.3%) were classified into this group. Class 2 (“Low Current Risk—Moderate Childhood Adversity”) included SMs who reported more adverse childhood experiences than SMs in the low-risk class, but who otherwise were less vulnerable on indicators reflecting current circumstances and experiences. The weighted percentage of SMs in Class 2 was 13.0%. SMs classified into Class 3 (“Low Current Risk—High Childhood Adversity”) reported very high levels of adverse childhood experiences, but their scores on the other domains were about average for the sample. The weighted percentage of SMs in Class 3 was 14.5%. The combined percentage of married, active component SMs demonstrating low risk on current indicators of vulnerability (Classes 1, 2, and 3) was 77.9%.
SM standardized vulnerability variables by class and weighted percentage of SMs in each class (N = 1,981).
Class 4 (“Moderate Risk”) includes SMs who scored moderately high on four of the five vulnerability domains but low on adverse childhood experiences. Finally, Class 5 (“High Risk”) includes SMs who were at relatively more vulnerable across all five indicators. The weighted percentages of SMs of these classes were 16.3% and 5.8%, respectively.
Latent Class Analysis for SPs
The LCA for SPs resulted in a six-class model (see Figure 2). The six-class model demonstrated a good fit to the data relative to other class solutions (AIC = 12,114, BIC = 12,337, SSBIC = 12,210, and entropy = .89). A five-class solution offered slightly better entropy (.90), but did not offer as good a fit to the data as the six-class model. Fit statistics and entropy were not as good for the seven-class model as the six-class model, so the six-class model was retained.
SP standardized vulnerability variables by class and weighted percentage of SPs in each class (N = 1,981).
SP classes shared some similarities with SM classes, but there were also important differences. Similar to the LCA results for SMs, the six classes were generally distinguished by SPs’ levels of vulnerability and resources across all indicators. As shown in Figure 2, Class 1 (“Low Risk”) included SPs with lower vulnerability on all four indicators, suggesting that they are at lower risk for future negative outcomes in the face of the continuing demands of military life. Over half of SPs (56.2%) were classified into this group. Class 2 (“Low Current Risk—Moderate Childhood Adversity”) was similar to Class 2 for SMs in that it included SPs who were less vulnerable across all indicators except for adverse childhood experiences, which were moderately higher for SPs in this group. The weighted percentage of SPs in this class was 11.3%. Class 3 (“Low Current Risk—High Childhood Adversity”) was also similar to SMs’ Class 3. SPs included in Class 3 were less vulnerable across all indicators except that adverse childhood experiences were higher for SPs in this group. The weighted percentage of SPs in Class 3 was 4.5%. Combining across Classes 1, 2, and 3, we estimate that 72% of military SPs were low risk on current indicators.
Classes 4 and 5 for SPs had no clear parallels to classes obtained for SMs. SPs in Class 4 experienced moderately high levels of both adverse childhood experiences and current stressful nonmilitary experiences. SPs in this class scored about average on mental/physical health and family stress. We labeled this class “Enduring Stressful Experiences,” and the weighted percentage of SPs in this class was 5.5%. Class 5 included SPs who were average or below average on all measures except family stress, which averaged 1.74 standard deviations above the mean. Thus, we labeled this class “High Family Stress.” The weighted percentage of SPs in this class was 4.2%. Finally, Class 6 was similar to SM Class 5 in that SPs in this class scored relatively high across all vulnerabilities. As with SMs, we labeled this class “High Risk,” and the weighted percentage of SPs in this class was 18.3%.
Congruency of Vulnerability Profiles Between SMs and SPs
Our final research question was whether SMs and SPs who are married to one another are more likely to share the same risk propensity. Because SM and SP class solutions differed, we combined latent risk classes to facilitate this comparison. For SMs and SPs, we combined the Moderate and High Childhood Adversity classes into one Low Current Risk—Childhood Adversity group. Because SPs in the Enduring Stressful Experiences class had more adverse childhood experiences and more current stressful nonmilitary experiences, we grouped them with the SP High Family Stress class to form a group that parallels the SM Moderate Risk class. We compared membership in these risk groups using a chi-square test of association and examined expected SM by SP cell memberships with actual cell memberships to determine whether couples were overrepresented or underrepresented in each cell.
The cross-classification matrix is shown in Table 1. For this matrix, the chi-square test of association was significant, χ2(9) = 181.7, p < .001. Examination of Table 1 reveals that, overall, couples were 18.3% more likely to be in congruent risk groups (observed n = 949) than would be expected by chance (expected n = 802.1). However, there was also a sizable percentage of couples in incongruent risk groups. To examine the overlap in risk for these couples more closely, we considered the Low Risk and Low Current Risk—Adverse Childhood classes to be low-risk groups for both SMs and SPs, and the Moderate Risk and High Risk classes to be high-risk groups. From this perspective, about 61.1% of couples were composed of two low-risk partners, and 11.2% of couples were composed of two moderate-to-high-risk partners. Thus, 27.7% of couples were composed of “mixed-risk” dyads, with 10.9% of couples where the SM was at higher risk, and 16.8% of couples where the SP was at higher risk.
Weighted Cross-Classification Matrix of Observed and Expected Membership in Risk Groups for SMs and SPs and Percent of Couples Observed Each Cell (N = 1,981).
Note. SM = servicemember; SP = spouse. Congruent categories are highlighted in grey.
Discussion
Rationale and Summary of Results
The goals of the current study were to estimate the proportion of deployable U.S. military SMs and SPs who are at risk for negative outcomes resulting from exposure to military stress and to examine the extent to which different vulnerabilities and resources overlap within individuals and families. The impetus for addressing these issues was an important policy question: should the military identify specific problems faced by SMs and their families and develop programs to target each problem independently, or should the military develop multifaceted programs to support individual and family problems across multiple domains? Within married deployable SMs, these results reveal a clear answer: vulnerabilities and resources cluster together within individuals. Those who are low risk in one domain are likely to be at low risk in multiple domains, and those at high risk in one domain are likely to be high risk in multiple domains. With respect to policy, these results support identifying vulnerable SMs and targeting them for multiple services, rather than targeting specific problems. Indeed, the military has resources that allow SMs and their families to call in with any problem and be routed to a specific service for that problem (e.g., Military OneSource). However, the programs to which callers are routed often target specific, individual problems (e.g., PTSD). SMs and their families who seek help outside of the military system are faced with similar compartmentalization of services (Horvitz-Lennon, Kilbourne, & Pincus, 2006). The current results suggest that SMs be offered a more holistic set of services to address interconnected individual and family needs (e.g., financial counseling along with treatment for depression, stress management, and family counseling).
Data from SPs painted a more nuanced picture. Like SMs, the dominant pattern among SPs was a clustering of indicators, such that SPs who were vulnerable in one domain were likely to be vulnerable in multiple domains. But this generalization was complicated by the fact that vulnerabilities are not as closely tied among military SPs as they are among their mates. The Enduring Stressful Experiences class was composed of SPs who had moderately high adverse childhood experiences and moderately high stressful experiences, but average scores on mental/physical health and family stress. The High Family Stress class had very high family stress scores, but average or below average scores on the other indicators. These classes were not large (5.5% and 4.2% of SPs, respectively), but they do suggest that, although military SPs could also benefit from programs that package multiple services for family members, some may also benefit from programs that offer support designed to relieve family-oriented stress specifically.
At the couple level, most military couples in our study were composed of two low-risk partners, and partners were more likely to share classifications than would be expected by chance. That said, almost 39% of military couples in our study have one or both spouses who are higher risk, so it is important that military programs be prepared to handle multiple and overlapping problems within such families. Because problems that affect one spouse can crossover and affect the other spouse (Neff & Karney, 2007), it is possible that both spouses in mixed-risk couples will develop vulnerabilities over time. Military programs should therefore consider offering programs that address vulnerabilities in the family environment as well as within the individual. This kind of comprehensive “couples-based” approach to helping SMs and SPs would capitalize on the strengths of the low-risk partner to help their spouse, and may prevent mixed-risk couples from becoming dual-risk couples.
Analyses of SMs and SPs both revealed two groups that had higher than average adverse childhood experiences, but were otherwise average or below average on the other indicators of current vulnerability. Replicating past research (Reading, 2006), across respondents there was a strong correlation between adverse childhood experiences and current mental and physical health (r = .30 for both SMs and SPs), but the LCA identified groups of respondents who did not fit the overall correlational trend in the data. It is unclear why adverse childhood experiences did not covary more strongly with current vulnerabilities and resources in these two groups, but previous research has shown that some people recover from adverse experiences in childhood to have stable adult lives (e.g., Klika & Herrenkohl, 2013). In these groups, the pattern of childhood risk coupled with relative well-being in adulthood may be a sign of resilience. Because members of these groups appear to have overcome negative experiences in the past, they may be more resilient to future negative experiences, including deployment stressors, than will members of the other groups. If so, then understanding how these groups deal with deployment stressors may provide insight for promoting resiliency in others.
Strengths and Limitations
Several strengths of the current research lend confidence to the conclusions drawn here. First, the Deployment Life Study surveyed a large sample of SM households from across the services, supporting our ability to generalize to active component U.S. military families. Second, the analysis included variables encompassing a broad range of constructs assessing respondents’ well-being, family dynamics, financial and social capital, and (for SMs) military experiences. Third, the Deployment Life Study includes data from both SMs and SPs, which allowed us to examine the correspondence of risk indicators within couples as well as within individuals.
Despite these strengths, however, interpretations of the current findings must also be tempered by several important limitations. First, whereas we examined vulnerability and resource factors in terms of separate, distinct categories, these constructs—vulnerability on one side and resources on the other—are continuous and fluid variables. The advantage of using categories rather than continuous measures is that we could examine mean levels of risk between groups rather than linear associations. The disadvantage is that members of the same risk class may have very different scores on the same indicator of vulnerability, so there is unavoidable error involved in predicting class membership. Second, although we assessed a broad range of domains in this study, we could not be comprehensive. A different set of domains, or a different set of instruments, might have resulted in a different set of vulnerability profiles. Finally, our sample of SMs was overwhelmingly male, limiting our ability to generalize to female SMs.
Conclusion
Wartime deployments have strained military families, creating uncertainty about whether deployable SMs and their families are at risk for negative outcomes from future deployments. The current study used a LCA of survey data from SMs and their SPs to examine the covariance of vulnerabilities (e.g., poor mental health) and resources (e.g., social support) within and across families and individuals. Results demonstrated that SMs and SPs who are vulnerable in one domain are likely to be vulnerable in multiple domains, but that a large proportion of military families are composed of mixed-risk couples. This suggests that programs designed to help specific individual-level problems (e.g., self-management PTSD) should take a more holistic approach to treating the individual-family system rather than individual symptoms.
Footnotes
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(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the Office of the Surgeon General, U.S. Army, and the Office of the Assistant Secretary of Defense for Health Affairs, the Defense Centers for Excellence for Psychological Health and Traumatic Brain Injury and was conducted jointly within the RAND Arroyo Center and the National Defense Research Institute under Contracts W74V8H-06-C-0001 and W74V8H-06-C-0002.