Mental health profiles of depressive symptoms and personal well-being among active-duty military families
Abstract
Although some research has examined the mental health of individual family members in military families, additional research is needed that considers mental health among multiple members of the family system simultaneously and that characterizes subsets of families with distinct patterns. Mental health patterns of depressive symptoms and well-being in and among families were identified using latent profile analysis with a community sample of 236 military families with a service member (SM) parent, civilian partner, and adolescent. Drawing from the Family Adjustment and Adaptation Response model, we examined several military-related family demands (e.g., relocations, deployments) and capabilities (e.g., family cohesion, social support outside the family) as correlates of the family profiles. Three profiles emerged: thriving families (62.3% of the sample where all three family members reported relatively low depressive symptoms and high personal well-being), families with a relatively distressed SM (24.2%), and families with a relatively distressed adolescent (13.5%). Overall, there were no differences between the groups of families regarding military-related demands, yet there were differences between the groups regarding their capabilities, namely family cohesion and social support. In general, families in the thriving profile tended to have higher family cohesion and social support as reported by multiple family members compared to the other two profiles. Findings can inform the development of family needs assessments and tailored interventions (and intervention points) based on family profiles and current capabilities.
INTRODUCTION
A significant number of families in the United States (U.S.) are connected to military institutions. As of Fall 2023, there were over 1.3 million active-duty service members (SMs), with over 440,000 SMs associated with the Army in particular (U.S. Department of Defense, 2024). Given this prevalence, there is a need for helping professionals, policymakers, and researchers to understand the needs and experiences of SMs and their families and to cultivate cultural competence as it relates to military-connected families (Atuel & Castro, 2018; Meyer et al., 2016). Growing empirical attention has been devoted to the mental health of members of military families and the factors that promote and sustain it. Studies have focused on the well-being of SMs or Veterans (Vogt et al., 2021; Williamson et al., 2019), their civilian partners (Donoho et al., 2018; Green et al., 2013), and/or their children (Chandra et al., 2010; Cramm et al., 2019). However, there has been less attention to the well-being of multiple members of military families simultaneously, which would offer novel insights regarding the health and well-being of military families.
The purpose of the present study is to identify mental health patterns among Army-connected SMs, their civilian partners, and their adolescent children. Specifically, we apply a profile-based approach to identify distinct subgroups of military families based on family members' reports of their depressive symptoms and personal well-being. That is, the profiles consider all three family members simultaneously to identify different subgroups of families based on the individual mental health of the members. Our use of triadic data represents a significant strength of the study and a notable contribution to the literature regarding mental health among military-connected families. To further characterize the groups, we also assessed associations between particular family mental health patterns and theoretically plausible correlates, namely military-related family demands (e.g., relocations, deployments) and social relationship quality in and outside the family (e.g., family cohesion and social support).
Variability in mental health among members of the military and their families
Several studies have investigated the mental health of members of the military, with growing attention to documenting and understanding variability in functioning. Profile-based analyses (also referred to as person-centered analyses), commonly conducted with latent profile analysis (LPA), are well-suited for capturing such variability because they identify latent (unobserved) subgroups with distinct patterns across multiple indicators (Wickrama et al., 2021). LPAs with SM and/or Veteran samples have examined patterns related to multiple characteristics of mental health, such as intra-individual patterns of suicidal ideation, alcohol/drug use, and depression (Allan et al., 2020; Smigelsky et al., 2019). One such LPA with 1266 National Guard soldiers identified groups of soldiers with distinct mental health patterns based on symptoms of posttraumatic stress disorder, depression, and generalized anxiety disorder (Contractor et al., 2015). Three subgroups of soldiers were identified with differing levels of symptom severity, rather than symptom type. The groups represented soldiers with mild (62%), moderate (27%), and severe (11%) mental health symptoms.
LPAs have also been utilized to understand risk and protective factors for mental health among military spouses and military-connected youth, with results generally demonstrating fewer mental health diagnoses for groups with more protective factors (e.g., higher levels of family cohesion, positive coping) and fewer risk factors (e.g., recent relocation, cumulative deployment length; Sullivan et al., 2021a, 2021b). Studies focused on military-connected adolescents have also examined patterns of coping and anxiety in connection to their well-being (Okafor et al., 2016; O'Neal et al., 2017). Together, these studies reveal important variability in mental health among SMs, their civilian partners, and their children, with some doing quite well and others struggling.
Understanding mental health at the family level
Most studies examining the mental health of individuals in military families include data from individual reporters, focusing exclusively on SMs, their civilian partners, or their children. Yet when research on mental health among military families is focused solely on an individual family member, potential connections between family members' mental health are overlooked. Indeed, family systems theory emphasizes the interconnections between family-system members and highlights the value of assessing the well-being of multiple family members concurrently and from each individual's perspective (Carr, 2016). Although not as prevalent as studies focused on individual family members, military family research supports the emphasis on interconnections between family members (Meadows et al., 2017), including studies noting that military fathers' mental health (depression and post-traumatic stress disorder) is associated with behavioral problems in their children as rated by parents, such that fathers with more mental health difficulties had children who were rated as having more behavioral problems (Farero et al., 2020; Parsons et al., 2018). Other research with data from multiple military family members has shown associations between the level of depressive symptoms of both active-duty and civilian parents with their adolescent's symptoms (O'Neal et al., 2016). Research has also found associations between SM reports of pre-deployment conditions (e.g., stressors, social support) and partner self-reports of psychological adjustment over the course of deployment (Erbes et al., 2017). Consequently, measuring the well-being of multiple family members provides valuable insights into the potential linkages between the mental health of family members (e.g., SMs, their romantic partner, and their child).
To date, the limited research that has used data from multiple members of military families primarily has examined within-family associations using variable-focused models. These types of models assume a homogenous sample comprised of a single group with the same pattern of associations, such as SMs with higher levels of mental health difficulties typically corresponding to a higher level of mental health difficulty in their child. Profile-based analyses, including LPAs, provide an opportunity to uncover potential sub-groups of families with distinct patterns of associations (Wickrama et al., 2021). Although some groups of military families might be characterized by similarities, such that in some families all members are experiencing higher levels of mental health and in other families, all members report relatively lower levels of mental health, other patterns are possible. For example, because each family member has such a unique experience of the military context (e.g., during a deployment there is a “leaving” parent and a “home front” parent, with very different experiences of the same deployment), there might be some families in which SMs could be faring well while their partners or child(ren) are not, or vice versa. LPA using mental health data from multiple members of military families can shed light on these different subgroups, thereby filling a gap that is important for our understanding of military families' mental health. The practical value of capturing multiple perspectives is also particularly pronounced in the military context given the notable possible leverage points, such as family mental health interventions and family life education, including military family programming.
To address this gap, the first aim of the present study is to identify distinct mental health profiles (or groups) of military families based on SMs', civilian partners', and adolescents' reports of their depressive symptoms and personal well-being (Research Question [RQ] 1) using LPA. This analysis provides insight into the unique patterns of mental health among military family members that are not captured in previous variable-centered association studies. Our inclusion of both depressive symptoms and personal well-being is consistent with the dual-factor model of mental health (Suldo & Shaffer, 2008), recognizing the importance of both subjective indicators of well-being (i.e., personal well-being) and more traditional, negative indicators of psychopathology (i.e., depressive symptoms). Building on previous LPAs focused on individuals in military families (e.g., Contractor et al., 2015; Okafor et al., 2016; O'Neal et al., 2017), we expected that there would be distinct profiles differing in levels and configurations of mental health, with some profiles characterized by higher mental health (i.e., lower depressive symptoms and higher personal well-being) than others. Based on dyadic research with military and civilian couples, we also expected that profiles may exist that include different patterns for distinct family members (e.g., more distress for the SM relative to their other family members; e.g., Pflieger et al., 2020; Novak et al., 2022). However, substantially less is known about family patterns that incorporate data from both parents and an adolescent family member in a single LPA.
Correlates of family mental health profiles
In addition to identifying distinct profiles based on multiple family members' mental health, we sought to understand different characteristics associated with membership in these profiles. The Family Adjustment and Adaptation Response (FAAR) model posits that family adjustment and adaptation operate as functions of family demands, family capabilities, and the accompanying meaning-making processes (Patterson, 2002). Family demands can be normative or nonnormative stressors, ongoing strains, or daily hassles. Family capabilities can be conceptualized as tangible and psychosocial resources or coping behaviors. A family's probability of adjusting well over time in the face of family demands is increased when the family has sufficient capabilities. In families without favorable adjustment and adaptation, suboptimal levels of well-being may develop among various family members and across various dimensions, including mental health.
Informed by this perspective, in the present study we focus on several military-related family demands. In addition to stressors common to all families, military-connected families can face stressors that are unique to the military context; these demands are often associated with family transitions and increased risk of exposure to traumatic experiences. For example, it is typical for active-duty military families to relocate every few years to a new duty station. Thus, military families are regularly uprooted from their communities and must establish new connections and resources. Another unique, but relatively common, demand associated with military life is deployment, which sets in motion a general sequence of the SM's separation from their family system and later reintegration into the family system. Experiences around deployment—often referred to as the “deployment cycle”—can shift family roles and routines and disrupt family norms (Flittner O'Grady et al., 2018; Riggs & Cusimano, 2014). Many military families also experience shorter family separations for training and other temporary duty assignments. Taken together, these varying family demands can exert influence on a family's equilibrium and, if not met with sufficient capacity, can impact the well-being of individuals and families. To explore this possibility, we investigate how military-related demands, specifically relocation history and several dimensions of family separation (total months deployment, months since most recent deployment, and time away in the past year), are related to the identified profiles.
Regarding family capabilities, we focused on the quality of social relationships in and outside the family, including each member's ratings of family cohesion and their available social support outside the family. Research has long espoused the value of social support from high-quality relationships for promoting psychological health in the context of stressful experiences (e.g., Cutrona & Russell, 1987; Thoits, 1985). Previous work with military families also indicates that these psychosocial resources can promote positive outcomes and protect against the detrimental effect of military stress (e.g., Mancini et al., 2015; O'Neal & Mancini, 2021).
Aligning with this research, the second aim of the present study was to examine whether the profile groups differed concerning military-specific family demands (relocation history, deployment history) and in the quality of their social relationships (family cohesion, social support) (RQ2). We generally expected that there would be significant differences between the profiles in these domains, with profiles marked by higher levels of mental health among one or more family members averaging relatively fewer military-related family demands and relatively higher social relationship quality than profiles marked by lower levels of mental health.
In addition, we explored between-profile differences in select demographic characteristics linked to mental health outcomes, including age, education, family relatedness, residential location, SM rank (a proxy for pay grade), SM sex, and civilian partner income. For instance, research has shown that women and those with less income and education are at greater risk for depressive symptoms (Cohen et al., 2020; Debreczeni & Bailey, 2021). Similarly, there are documented developmental differences in mental health across adolescence and adulthood with declining mental health in older ages (for both adolescents and adults; Substance Abuse and Mental Health Services Administration, 2023). Family relatedness and residential location (on or off the installation) were included given their implications for instrumental and social support available to the family and, subsequently, mental health (Gariepy et al., 2016).
METHOD
Respondents and procedure
The present study leverages secondary data from a larger study of 273 military families on an active-duty Army installation in the continental United States conducted between 2012 and 2013. Eligible families were those with at least one active-duty SM and at least one adolescent between the ages of 11 and 18. This age range was chosen for the original study to maximize the breadth of child ages included while limiting the sample to ages that could realistically comprehend and answer the questions asked. All families had experienced one or more deployments. In families with more than one participating adolescent, we utilized data from the older adolescent, to provide maximum age variability, given that the sample was skewed toward younger adolescents. Given our focus on families with a SM, a civilian partner, and an adolescent, seven dual-military couples were removed along with 30 families without data from both partners. The final analytic sample consisted of 236 families with data from a SM, civilian partner, and adolescent.
Almost all SMs reported that they were married to their partner (n = 234; 99.2%), and the other two couples were engaged. All couples were mixed-gender, and most (n = 223; 94.5%) of the SMs were men. SMs and civilian partners reported their age category. The most common age ranges were “31 to 35 years” (n = 69, 29.2% for SMs; n = 92, 39.0% for civilian partners) and “36 to 40 years” (n = 87, 36.9% and n = 75, 31.8% for SMs and civilian partners, respectively). Adolescents ranged in age from 11 to 18 (M = 14.0, SD = 2.1). Consistent with the general military population of this age range, the majority of SMs were enlisted with the rank of “Sergeant (E5) to Sergeant Major (E9)” (n = 168, 71.2%). SM and civilian partners most commonly reported “some college” education (n = 105; 44.5% and n = 85; 36.0%, respectively). Approximately one-fifth of the families (n = 50; 21.2%) had more than three children aged 18 or under living with them. Regarding family relatedness, the majority (n = 139; 58.9%) of households had two biological parents. Other family configurations included biological child of the SM only (n = 13; 5.5%), biological child of the civilian partner only (n = 79; 33.5%), child of a relative (n = 2; 0.8%), and adopted child (n = 1; 0.4%). Just over half of the families lived on the installation (n = 132, 55.9%). Approximately half (n = 134; 56.8%) of the civilian partners did not report earning an income in the past year. Of those who were employed, annual income ranged from “less than $10k” (n = 26; 11.0%) to “$100k or more” (n = 3; 1.3%).
Families were recruited to participate through multiple methods, including print and radio advertisements and signs/flyers at youth centers and military and community stores and restaurants. Surveys were administered at on-installation computer labs. All eligible family members came to the lab to take the survey concurrently in separate locations. Participation was voluntary. Consent was obtained from SMs and civilian parents. Assent and parental permission were obtained for minor children. Respondents were compensated for their time. The research protocol was executed as approved by the University of Georgia Institutional Review Board for Human Subjects (2012108633) and the U.S. Army Research Institute.
Measures
Descriptive statistics and correlations among study variables are shown in Table 1. Rates of missing data were minimal, averaging less than 1%.
| Variables | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. SM depressive sym. | — | |||||||||||||||
| 2. CP depressive sym. | 0.21*** | — | ||||||||||||||
| 3. Ad. depressive sym. | 0.04 | 0.14* | — | |||||||||||||
| 4. SM personal WB | −0.54*** | −0.14* | −0.09 | — | ||||||||||||
| 5. CP personal WB | −0.14* | −0.51*** | −0.16* | 0.18** | — | |||||||||||
| 6. Ad. personal WB | −0.09 | −0.10 | −0.56*** | 0.11 | 0.15* | — | ||||||||||
| 7. Time deployed | 0.09 | 0.02 | 0.09 | −0.10 | −0.06 | −0.07 | — | |||||||||
| 8. Time away past year | −0.03 | −0.08 | 0.05 | −0.09 | −0.02 | −0.03 | 0.10 | — | ||||||||
| 9. Months since deploy | −0.04 | −0.04 | −0.05 | 0.05 | 0.14* | 0.09 | −0.10 | −0.50*** | — | |||||||
| 10. Relocations | 0.10 | 0.09 | 0.13* | −0.08 | −0.06 | −0.10 | 0.04 | 0.01 | 0.01 | — | ||||||
| 11. SM family cohesion | −0.35*** | −0.17** | −0.12 | 0.40*** | 0.13* | 0.06 | −0.06 | −0.06 | 0.12 | 0.02 | — | |||||
| 12. CP family cohesion | −0.17** | −0.34*** | −0.13* | 0.13* | 0.41*** | 0.19** | −0.05 | −0.03 | 0.17* | −0.05 | 0.29*** | — | ||||
| 13. Ad. family cohesion | −0.13* | −0.13 | −0.49*** | 0.09 | 0.19** | 0.50*** | −0.06 | 0.02 | 0.05 | −0.04 | 0.22*** | 0.27*** | — | |||
| 14. SM social support | −0.45*** | 0.03 | −0.07 | 0.43*** | −0.04 | 0.12 | 0.02 | 0.03 | 0.02 | 0.06 | 0.39*** | 0.01 | 0.04 | — | ||
| 15. CP social support | −0.06 | −0.23*** | −0.01 | 0.12 | 0.36*** | 0.07 | 0.06 | −0.03 | 0.09 | 0.05 | 0.09 | 0.30*** | 0.08 | 0.11 | — | |
| 16. Ad. social support | −0.08 | −0.18** | −0.43*** | 0.02 | 0.17** | 0.56*** | −0.06 | 0.03 | 0.10 | −0.08 | 0.14* | 0.22*** | 0.51*** | 0.08 | 0.15* | — |
| M | 1.54 | 1.58 | 1.47 | 3.08 | 3.09 | 3.25 | 29.76 | 2.74 | 3.59 | 3.99 | 3.97 | 4.04 | 4.02 | 3.07 | 3.21 | 3.36 |
| SD | 0.45 | 0.47 | 0.36 | 0.60 | 0.57 | 0.47 | 15.32 | 1.42 | 0.71 | 1.81 | 0.54 | 0.59 | 0.76 | 0.47 | 0.49 | 0.42 |
- Note: Time deployed, time away past year, months since deployment, and relocations were reported by the service members.
- Abbreviations: Ad, adolescent; CP, civilian partner; SM, service member; Sym, symptoms; WB, well-being.
- * p < 0.05.
- ** p < 0.01.
- *** p < 0.001.
Mental health
Depressive symptoms
SMs' and civilian partners' depressive symptoms were measured using the 7-item Abbreviated Center for Epidemiologic Studies Depression Scale (Radloff, 1977; sample item: “I felt sad” over the previous week). Items were rated on a 3-point scale ranging from none of the time (1) to most of the time (3). Following the scale scoring instructions, mean scores were computed, and the scores were utilized as a continuous variable, with higher scores indicating greater levels of depressive symptoms. The measure had good internal reliability (α = 0.83 for SMs and 0.85 for civilian partners).
For adolescents, depressive symptoms were measured using the 20-item Center for Epidemiological Studies Depression Scale for Children (Faulstich et al., 1986). Adolescents rated their agreement using response options ranging from not at all (1) to a lot (4). Sample items include: “I was bothered by things that usually don't bother me” and “I felt sad.” Similar to the adult measure, a mean score was computed with higher scores indicating more depressive symptoms, and the measure had good internal reliability (α = 0.91). To facilitate comparisons with SMs' and civilian partners' scores, we transformed adolescents' scores to a 1–3 scale using a linear transformation (1 = 1; 2 = 1.67; 3 = 2.33, 4 = 3).
Personal well-being
SMs' and civilian partners' well-being was measured using the eight-item Personal Well-being Index (International Wellbeing Group, 2013), which examines satisfaction in various domains such as standard of living, personal relationships, and future security (e.g., “I was satisfied with my life as a whole” and “I was satisfied with what I was currently achieving in life”). Participants responded on a 4-point scale ranging from strongly disagree (1) to strongly agree (4). Mean scores were computed, with higher scores indicating greater well-being. Reliability was high (α = 0.89 for SMs and 0.88 for civilian partners).
Adolescents' well-being was assessed using six items from the Personal Well-being Index—School Children (PWI-SC; Cummins & Lau, 2005) to examine quality of life on a 4-point scale ranging from very sad (1) to very happy (4). Adolescents indicated how they felt about different domains of life (e.g., “Your life as a whole” and “What may happen to you later on in your life”). A mean score was computed with higher scores indicating greater well-being. The scale had good internal reliability (α = 0.79).
Military-related family demands
Relocations
SMs indicated how many permanent change of station relocations they experienced since joining the Army on a 6-point scale from zero (1) to 5 or more (6).
Total time away for deployment
SMs indicated the total number of months they were deployed since 2001 as a continuous variable.
Months since the most recent deployment
SMs indicated if they returned from their most recent deployment less than a month ago (1), 1–3 months ago (2), 4–12 months ago (3), or more than 1 year ago (4).
Time away in the past year
SMs indicated their total time away from home in the past year for military duty (deployment, training, or other military requirement). Response options ranged from no time away (1) to 10–12 months away (5).
Quality of social relationships
Family cohesion
Each family member rated their perceptions of family cohesion using the 7-item Balanced Cohesion subscale of the Family Adaptability and Cohesion Scale (FACES IV; Olson et al., 2006; sample item: “My family members feel very close to each other”). Responses ranged from strongly disagree (1) to strongly agree (5). Mean scores were computed for each family member, with higher scores indicating more family cohesion. The scales had good internal reliability (α = 0.74 for SMs, 0.80 for civilian partners, and 0.91 for adolescents).
Social support
Each family member rated the quality of their social connections to people outside the family (including relationships with friends and community members) using the 24-item Social Provisions Scale (Cutrona & Russell, 1987; sample item: “I have other people I can turn to in times of stress”). Participants rated their agreement with each item on a 4-point scale from strongly disagree (1) to strongly agree (4). Mean scores were computed; higher scores indicated more social support (α = 0.93 for SMs, 0.93 for civilian partners, and 0.91 for adolescents).
Demographic characteristics
Age, education, family relatedness, residential location, SM rank, SM sex, and civilian partner income were also examined. Regarding age, SMs and civilian partners separately indicated their age using categorical variables ranging from “25 or younger” to “over 50 years.” Adolescents reported their age as a continuous variable. SMs and civilian partners indicated their educational attainment with six response options (1 = less than high school to 6 = graduate degree or higher). Family relatedness was determined based on both parents' reports of the adolescent's relationship with them (1 = living with two biological parents; 0 = other family configurations). SMs reported their residential location (1 = living on the installation; 2 = living off the installation), rank [1 = Private (E1) to Corporal (E4), 2 = Sergeant (E5) to Sergeant Major (E9), 3 = Second Lieutenant (O1) to Major (O4) including Warrant Officers, and 4 = Lieutenant Colonel (O5) to Lieutenant General (O9)], and sex (1 = man; 2 = woman). Civilian partners reported their annual income in the past 12 months using ten- and twenty-thousand dollar increments categories ranging from 0 = no income to 9 = $100k or more.
Data analytic strategy
First, to identify distinct profiles of families (RQ1) based on the mental health of the SM, civilian partner, and adolescent, we conducted an LPA (Asparouhov & Muthén, 2013; Wickrama et al., 2021) in Mplus Version 8 (Muthén & Muthén, 1998–2017) using Full Information Maximum Likelihood. In this procedure, unobserved data patterns are identified using an SEM-based classification technique. The LPA assumption of local independence (i.e., that the items used to identify the profiles are independent of each other) was assessed by adding all possible correlations between the variables in the LPA individually to identify any significant associations. Several statistically significant correlations were identified and were thus retained in the LPA, namely, the intra-individual correlations (i.e., SMs' depressive symptoms with their well-being, civilian partners' depressive symptoms with their well-being, adolescents' depressive symptoms with their well-being). Following the recommendations of others (e.g., Nylund-Gibson & Choi, 2018), class enumeration (i.e., deciding how many classes to retain) was based on different model fit indices, class differentiation, group size, and theoretical considerations. Fit indices examined include the Akaike Information Criterion (AIC) and Sample-size adjusted Bayesian Information Criterion (SABIC), where smaller numbers indicate better model fit, and the bootstrapped likelihood ratio test (BLRT), where a statistically significant p-value indicates the k − 1 model is a better fit than the k model. Entropy values provide an indicator of class differentiation, and values greater than 0.80 indicate “good” classification (Clark & Muthén, 2009). Along with these fit indices, we also considered group size (as it is recommended the smallest group contain at least 5% of the sample) and theoretical considerations (e.g., if the groups represent theoretically meaningful and plausible classes; Wickrama et al., 2021).
To test whether the identified profiles differed in military-related family demands and social relationship quality (RQ2), we examined mean differences between the groups in these domains by estimating auxiliary variables using the “BCH” option in the LPA. Demographic characteristics were also assessed in this manner. One of the advantages of this auxiliary option is that profile membership does not change with the addition of covariates. In addition, this modeling technique acknowledges and accounts for the variance, or possible uncertainty, of group membership (because the latent profiles are inferred from the data rather than known a priori).
Due to the small number of women SMs, sensitivity analyses were conducted to determine if their exclusion resulted in different findings. No significant differences were noted and, thus, women SMs were retained in the analytic sample.
RESULTS
Family mental health profiles (research question 1)
Panel A of Table 2 summarizes multiple model fit indices used to evaluate the competing LPA solutions. Based on fit indices and the conceptual and practical analytic strategies described above, the 3-group solution was selected as the optimal solution. The BLRT for this model was statistically significant, indicating that this model represented an improvement over the 2-group model. The entropy value was acceptable for the 3-group model (0.82), and AIC and SABIC statistics for the 3-group model improved over the 2-group model. In the 3-group model, each group size was adequate for analysis (13.5% for the smallest group). The 4- and 5-group solutions were not considered due to their inclusion of a small group (<4%). The results of the 4-group solution were also deemed inadmissible due to a potential local maxima error. Figure S1 illustrates the mean plots for the 2- through 5-class solutions.
| Panel A: Model fit statistics for family mental health profiles | ||||||
|---|---|---|---|---|---|---|
| Solution | BLRT | p | AIC | SABIC | Entropy | Group sizes |
| Two-group solution | 71.57 | 0.000 | 1655.35 | 1661.82 | 0.90 | 13.7%/86.3% |
| Three-group solution | 40.43 | 0.000 | 1628.92 | 1637.45 | 0.82 | 13.5%/24.2%/62.3% |
| Four-group solution | local maxima error | |||||
| 26.57 | 0.020 | 1616.35 | 1626.94 | 0.88 | 1.9%/2.7%/12.7%/58.3% | |
| Five-group solution | 25.78 | 0.06 | 1604.57 | 1617.22 | 0.85 | 3.3%/8.5%/10.0%/25.8%/52.4% |
| Panel B: Descriptive statistics for family mental health profiles and post-hoc analyses of between-group differences1 | |||
|---|---|---|---|
| Mental health domains | Thriving (n = 147) Mean (SD) | Relatively distressed service member (n = 57) Mean (SD) | Relatively distressed adolescent (n = 32) Mean (SD) |
| Depressive symptoms | |||
| SMs | 1.32 (0.25)a | 2.12 (0.26)b | 1.51 (0.38)c |
| Civilian partners | 1.49 (0.45)a | 1.73 (0.45)b | 1.72 (0.49)b |
| Adolescents | 1.34 (0.31)a | 1.42 (0.37)a | 2.17 (0.42)b |
| Personal well-being | |||
| SMs | 3.27 (0.54)a | 2.60 (0.49)b | 3.03 (0.53)c |
| Civilian partners | 3.18 (0.58)b | 2.97 (0.56)ab | 2.90 (0.50)a |
| Adolescents | 3.34 (0.43)a | 3.21 (0.49)a | 2.91 (0.47)b |
- Abbreviations: AIC, Akaike Information Criterion; BLRT, Bootstrapped Likelihood Ratio Test; SABIC, Sample-size adjusted Bayesian Information Criterion.
- 1 Groups with different subscripts differed significantly on that variable at the p < 0.05 level.
The three profile groups are shown in Figure 1, and group means for the different mental health variables are presented in Table 2, Panel B. Group 1 was the largest (n = 147 families; 62.3%), and these families were characterized as thriving because the SMs, civilian partners, and adolescents in these families all averaged relatively low depressive symptoms (M = 1.32, 1.49, and 1.34, respectively; possible range: 1–3) and relatively high personal well-being (M = 3.27, 3.18, and 3.34, respectively; possible range: 1–4).1 Families in Group 2 (n = 57 families; 24.2%) were characterized as having a relatively distressed SM, including SMs averaging moderate depressive symptoms (M = 2.12 out of 3) and moderate personal well-being (M = 2.60 out of 4). Civilian partners in these families averaged moderately low depressive symptoms (M = 1.73) and moderately high personal well-being (M = 2.97), and adolescents averaged low depressive symptoms (M = 1.42) and high personal well-being (M = 3.21).2 Last, Group 3 (n = 32; 13.5%) was the smallest and consisted of families with a relatively distressed adolescent, including adolescents with moderate depressive symptoms (M = 2.17 out of 3) but moderately high personal well-being (M = 2.91 out of 4). The SMs in these families reported moderately lower depressive symptoms (M = 1.51) and moderately higher personal well-being (M = 3.03), and civilian partners with moderately lower depressive symptoms (M = 1.72) and moderately higher personal well-being (2.90, respectively).3
As a post-hoc analysis to better understand the groups, in the LPA model, we examined between-group mean differences in depressive symptoms and personal well-being scores using the model constraint command (see Table 2, Panel B). Collectively, these results supported the group names: the thriving group was comprised of families where all three members reported the lowest depressive symptoms and the highest personal well-being relative to the other groups; the families comprising the relatively distressed SM group had SMs with the highest depressive symptoms and lowest personal well-being relative to the other groups; and the families in the relatively distressed adolescent group had adolescents with the highest depressive symptoms and lowest personal well-being relative to the other groups.
Correlates of family mental health profiles (research question 2)
Next, we examined between-group differences for the military-related family demands (number of relocations, total time away for deployment, months since most recent deployment, and time away in the past year) and two indicators of social relationship quality (family cohesion and social support, reported separately by each family member). Results are reported in Table 3. For military-related family demands, there were no overall differences across the three groups of families. For indicators of social relationship quality, there were significant omnibus differences in perceptions of family cohesion across the three groups of families. Pairwise comparisons indicated that in the group of families with a relatively distressed SM, SMs reported less family cohesion than SMs in the thriving family group (p < 0.001) and SMs in the relatively distressed adolescent group (p = 0.036). Civilian partners in families with a relatively distressed SM reported less family cohesion relative to the civilian partners in the thriving group (p = 0.001). In addition, adolescents in the relatively distressed adolescent group perceived that their families had significantly less cohesion compared to adolescents in the thriving and relatively distressed SM groups (p < 0.001 and p = 0.020, respectively).
| Variables | Thriving mean (SE) | Relatively dis. SM mean (SE) | Relatively dis. Adolescent mean (SE) | Overall comparison (Χ2) | p-value |
|---|---|---|---|---|---|
| Military-related family demands | |||||
| Relocations | 3.83 (0.16) | 4.25 (0.26) | 4.28 (0.34) | 2.38 | 0.304 |
| Total months deployed | 28.51 (1.28) | 31.08 (2.64) | 33.27 (3.99) | 1.66 | 0.437 |
| Months since deployment | 3.61 (0.06) | 3.57 (0.12) | 3.49 (0.16) | 0.52 | 0.773 |
| Time away in the past year | 2.77 (0.14) | 2.59 (0.22) | 2.83 (0.28) | 0.55 | 0.760 |
| Family cohesion | |||||
| SMs | 4.14 (0.04)a | 3.57 (0.08)b | 3.90 (0.11)c | 36.52*** | 0.000 |
| Civilian partners | 4.16 (0.05)a | 3.74 (0.11)b | 4.01 (0.11)ab | 12.34** | 0.002 |
| Adolescents | 4.18 (0.06)a | 3.95 (0.12)a | 3.46 (0.17)b | 15.82*** | 0.000 |
| Social support | |||||
| SMs | 3.22 (0.04)a | 2.66 (0.07)b | 3.08 (0.08)a | 42.04*** | 0.000 |
| Civilian partners | 3.22 (0.04) | 3.11 (0.07) | 3.31 (0.08) | 3.09 | 0.213 |
| Adolescents | 3.43 (0.03)a | 3.31 (0.07)ab | 3.13 (0.09)b | 9.99** | 0.007 |
| Demographic characteristics | |||||
| SM age | 3.83 (0.10) | 3.87 (0.14) | 3.57 (0.18) | 1.90 | 0.387 |
| Civilian partner age | 3.79 (0.09) | 3.60 (0.15) | 3.72 (0.21) | 1.10 | 0.576 |
| Adolescent age | 14.15 (0.20) | 13.84 (0.30) | 13.89 (0.36) | 0.83 | 0.660 |
| SM edu. attainment | 3.60 (0.10) | 3.40 (0.16) | 3.22 (0.19) | 3.17 | 0.205 |
| Civilian partner edu. attainment | 3.57 (0.12) | 3.50 (0.19) | 3.36 (0.23) | 0.65 | 0.724 |
| Family relatedness (two bio. parents) | 0.62 (0.04) | 0.46 (0.08) | 0.73 (0.09) | 5.39 | 0.067 |
| Residential location (off installation) | 1.46 (0.05) | 1.40 (0.08) | 1.40 (0.10) | 0.58 | 0.747 |
| SM rank | 1.95 (0.06) | 1.99 (0.08) | 1.97 (0.09) | 0.12 | 0.940 |
| SM sex | 1.09 (0.02)a | 0.99 (0.00)b | 1.03 (0.03)ab | 14.05*** | 0.001 |
| Civilian income | 1.50 (0.20) | 1.13 (0.26) | 1.37 (0.44) | 1.09 | 0.580 |
- Note: Classes with different subscripts differ significantly on that variable.
- Abbreviations: Bio., biological; Dis., distressed; Edu., educational; SM, service member.
- ** p < 0.01.
- *** p < 0.001.
There was also evidence of significant between-group differences in SMs' and adolescents' ratings of social support, though not civilian partners' ratings. In the group of families with a relatively distressed SM, SMs reported significantly lower levels of social support than the thriving and relatively distressed adolescent groups (both p < 0.001), who did not differ from one another. In the relatively distressed adolescent group, adolescents reported less social support than those in the thriving group (p = 0.001); the other comparisons were not significant.
As a final step, we examined differences between groups for age, educational attainment, family relatedness, residential location, SM rank and sex, and civilian partner's income. The only statistically significant group difference was for SM sex, where families with a female SM were more likely to be classified in the thriving profile compared to the profile of families with a relatively distressed SM (p < 0.001).
DISCUSSION
Past research has identified important variability in the individual mental health of SMs, their civilian partners, and their children. Trends across studies suggest that, in general, members of military families are faring well and adaptive in the context of change, yet a meaningful minority exhibit adversity across a range of mental health, welfare, and coping indicators (Contractor et al., 2015; Okafor et al., 2016; O'Neal & Lavner, 2022; Sullivan et al., 2021a, 2021b). The current study advances this literature by using a family systems lens to identify mental health patterns in active-duty families (considering the depressive symptoms and personal well-being of SMs, their civilian partners, and their adolescent children). Employing the lens of the FAAR model (Patterson, 2002), family demands and capabilities were also examined in connection to profile membership. Findings add to our understanding of mental health in military families and have important practical implications.
Mental health of military families
The current study used data from multiple family members to provide a triadic analysis of the mental health of SMs, their civilian partners, and their adolescent children. Mental health was operationalized across two indicators, depressive symptoms and personal well-being, as reported by each family member. This approach acknowledges that mental health includes positive and negative elements and is more than just the absence of disorder but is also the state of being satisfied, healthy, or comfortable (Suldo & Shaffer, 2008).
Three family profiles emerged based on family members' reports of their mental health: thriving (62.3%), relatively distressed SMs (24.2%), and relatively distressed adolescents (13.5%). Similar to other profile analyses examining the well-being of individual members of military families (e.g., Contractor et al., 2015; Sullivan et al., 2021a), we found that there was a sizeable portion of our sample who were functioning well: the thriving profile was characterized by all three family members reporting comparatively fewer depressive symptoms and higher rates of personal well-being than the other groups. The high percentage of families characterized as thriving may not be surprising to those who regularly work with military families, but this is an important insight for new military service providers as the findings stand in contrast to commonly held perceptions of military families as being “in crisis.” For researchers and the broader scientific community, this percentage emphasizes the importance of strength-based research recognizing, and further enhancing, the many strengths military families often draw from to thrive in challenging situations.
Further, as their names convey, the other two profiles were characterized by relative distress primarily reported by one family member, the SM, or the adolescent. Of note, these members were labeled relatively distressed because a given indicator of their mental health (e.g., depressive symptoms) was worse within (e.g., the SM compared to their partner and adolescent) and across (e.g., the SM compared to SMs in the other profiles) families, not because their score on the measure was in the distressed range. These findings align with family research from the civilian literature demonstrating that individual family members can sometimes deviate from each other in terms of their mental health (Hawkins et al., 2018). About the relatively distressed SM profile, we know that SM mental health struggles can compound with other military stressors to exacerbate vulnerability (Wright et al., 2012). Families in this profile may be encountered more frequently by helping professionals and military providers. In contrast, families in this distressed adolescent profile may be harder to identify among providers serving SMs as it requires a systemic perspective of wellness that considers multiple family members' mental health. This may signal the need for training in systemic assessments and the development of tools that incorporate multiple family members.
Although a profile of relatively distressed civilian partners did not emerge, there was some significant variation in civilian partners' mental health across the profiles. Most notably, civilian partners in families with a relatively distressed SM or adolescent generally experienced higher levels of depressive symptoms than their civilian partner counterparts in the “thriving” families (where all three family members demonstrated favorable mental health). This variation in civilian partners' depressive symptoms across the groups point to interdependence between family members' mental health, such that when one of their family members was relatively high in depressive symptoms, the civilian partner was also relatively high in depressive symptoms (though less so than the identified member). Examples of this type of interdependence were evident in other family members as well. For example, as shown in Panel B of Table 2, SMs averaged more depressive symptoms in the group of families with a distressed adolescent than when all three family members were thriving. These findings align with systems theory's contention that there are inter-individual connections between members' mental health (Carr, 2016).
Family demands and capabilities
Guided by the FAAR model, we considered family demands across these family profiles, specifically military-related demands, including relocation history and deployment history. No significant differences in military-related characteristics were identified across the profiles, suggesting that the families comprising the three profiles may have faced similar demands in terms of relocation and deployment.
We also considered family capabilities, namely the level of support within (family cohesion) and outside (social support) the family. The profiles did differ significantly concerning family cohesion and social support. On average, family cohesion and social support were highest for family members in the thriving profile, suggesting support is a “hallmark” that distinguishes thriving families from other families. This aligns with what we know about the importance of social provisions within and external to the family for the health and adjustment of SMs, partners, and children (Mancini et al., 2015). In contrast, those in the distressed groups reported relatively lower levels of support—SMs in the relatively distressed SM group reported relatively lower levels of family cohesion and social support compared to SMs in the other profiles, as did adolescents in the relatively distressed adolescent group (compared to adolescents in the thriving profile). These patterns have meaningful practical implications, as they suggest that programs and policies should seek to bolster psychosocial resources to position families to meet the demands of military life and promote the mental health of all military family members. Fortunately, there are evidence-based programming options designed for military families (see the National Academies of Sciences, Engineering, and Medicine (2019) report focused on strengthening the military family readiness system for more information on programs). Programs such as Families OverComing Under Stress (FOCUS) that include multiple family members, provide practical skills training, and build on family strengths are positioned to enhance family resilience and promote healthier parent–child interactions as well as reduce mental health difficulties.
We found no differences between the profiles in regards to age, educational attainment, family relatedness, residential location, SM rank (a proxy for pay grade), or the civilian partner's income. Okafor et al. (2016) similarly found no differences in the coping profiles of adolescents in military families for three military-related factors: number of parental separations, number of relocations, or the rank of the military parent. This lack of significant differences was somewhat surprising given previous research noting mental health differences based on these characteristics (e.g., developmental stage). However, it may reflect our focus on family profiles, whereas many of these variables are specific to the individual (e.g., individual family members' age). Interestingly, the one demographic difference that emerged was for SM sex. Women SMs were less likely to be in families classified as having a “relatively distressed SM” compared to the “thriving” mental health profile. However, more research is needed to confirm this finding, given the small number of women SMs in the sample.
Limitations and future directions
Study merits, including the use of triadic data from SMs, their civilian partner, and their adolescent child, should be considered alongside limitations. To begin, all data were self-reported, and our assessment of mental health was limited to depressive symptoms and personal well-being. Diverse assessment methods (e.g., clinical interviews, collateral reports) that allow for clinical diagnoses, alongside additional measures of mental health are warranted in future research. In particular, post-traumatic stress disorder would be valuable to include in future research, given it is a central risk for active duty military service and strongly linked to depressive symptoms. Because the data were collected at one-time point, we limited our analyses to examining correlates of the mental health groups rather than antecedents or consequences, and we note that the method of doing so required multiple individual statistical tests. Furthermore, although the cross-sectional data were adequate for identifying patterns of mental health within families, future research with longitudinal data is needed to identify patterns that account for directional dependencies (e.g., how one family member's mental health may impact another family member's mental health).
Regarding the timing of data collection, the families in the current study had a SM who was not currently deployed; examining the continuity and change of family profile membership across the deployment cycle and/or other times of meaningful transition (e.g., relocations) will expand our understanding of family mental health. Related to timing, while many aspects of military life (including family life and military service) have not changed considerably since the data were collected over a decade ago, results may vary with operational or policy changes that affect the force and their families. Concerning generalizability, examinations of family mental health are needed in more diverse samples with larger samples that include representation of diverse branches of service (e.g., Air Force, Navy), sex (e.g., women SMs), family structures (e.g., single caregiver families), and family developmental time points (e.g., families without children, families with young children). Similarly, the study utilized a community sample and is not representative of clinical samples, where we would expect higher levels of depressive symptoms and lower levels of personal well-being for family members. The sample characteristics are particularly important for the analyses used in the current study, where identified profiles are highly dependent on the sample from which they are estimated.
CONCLUSIONS
Family-focused investigations of mental health such as the present study help identify families in need of support as well as those that are functioning well. Identifying families in need of support is paramount for all families, but especially so for families that face unique challenges, such as military families. Understanding the differences that exist between families who are thriving and those in relative distress can inform efforts to identify families in need of services and intervening to provide appropriate services, such as building family cohesion and social support, more broadly. Of note in the current investigation, families who were thriving were able to pinpoint familial and social resources at their disposal. Thus, families may not be able to mitigate the demands that they face, but they, and the helping professionals they encounter, can take efforts to actively bolster their psychosocial resources.
ACKNOWLEDGEMENTS
This research was supported by the USDA National Institute of Food and Agriculture Award No. 2009-48680-06069 (PI: Jay A. Mancini).
