ABSTRACT
The U.S. Army has a vested interest in retaining the skilled personnel necessary to achieve its mission and strategic goals. A wealth of research has investigated the retention process and what influences service member decisions to stay in the military. While families are an important influence on soldier retention decisions, research on the mechanism by which this happens is lacking. This report explores the relationship between spouse attitudes and perceptions, resource use, and soldier retention almost two years later, using a proposed theoretical model. Our results generally support our model, with the important change that resource use and unmet needs and stress were not directly associated with specific attitudes toward staying in the military as we had expected. Instead, the association was accounted for by relationship with general attitudes toward the military. Spouses whose needs were unmet after seeking help from available resources experienced greater stress, and spouse unmet needs and reports of greater stress were associated with worse general attitudes toward the military; worse general attitudes toward the military were associated with less inclination to stay a military family; which in turn predicted soldier turnover almost two years later. As the research in this report shows, providing benefits to military spouses is also associated with a tangible and important outcome for the military: improved service member retention.
KEYWORDS: Retention, turnover, military families, soldiers, attitudes
What is the public significance of this article?—Limited research establishes the importance of spouse attitudes and perceptions for their soldiers’ retention, and even less speaks to the underlying mechanisms of how these attitudes and perceptions influence soldiers’ retention. By exploring the provision and more importantly use of resources by spouses to address the challenges of military life, this research helps researchers and policy makers understand the importance of these resources to key service outcomes.
Introduction
Retention of soldiers has long been of interest to the Army. The skilled personnel and “knowledge workers” required to meet the strategic goals of the Army (see, e.g., 2019 Army People Strategy1) are much in demand throughout the general U.S. economy. The Army invests heavily in training its recruits. As this investment in the workforce increases, so does the Army’s interest in retaining these members to benefit from a return on that investment. Hence, ongoing research examines the process of retention, or attrition/turnover – retention’s flip side – and what factors influence soldiers to leave the Army. One potential factor in soldier retention that has been understudied is the experience of Army spouses.
Overview of retention and turnover
A commonly accepted process model of turnover is described in a Hom et al. (2017) review; a simplified framing of this model is one in which dissatisfaction with the current job leads to consideration of job alternatives which, when appealing, lead to intentions to quit, and turnover ensues. Similarly, a simplified framing of the Theory of Reasoned Action (TRA) and its extension, the Theory of Planned Behavior (TPB), specifies that attitudes toward a behavior and perceived norms regarding a behavior predict the intention to execute a behavior, which, when a person has perceived control over the behavior, in turn predicts actual behavior execution (Fishbein & Ajzen, 1975; Ajzen, 1985; see, e.g., review in Hale et al., 2002). Over the years, job satisfaction and organizational commitment in particular gained quite a bit of prominence as the most powerful attitudinal predictor of turnover, including among military populations (see, e.g., Fafara & Westhuis, 2007; Griffeth et al., 2000; Heffner & Gade, 2003; Lytell & Drasgow, 2009). However, researchers were troubled that even the best models explained a relatively low proportion of the variance.
In the 1990’s different theoretical approaches were proposed to address this. One such, the unfolding model, suggested that not all employees follow a path to turnover mediated by job satisfaction but rather some encounter shocks that provide additional information or meaning about a job and lead them to evaluate alternatives or even leave abruptly (Holtom et al., 2005). As another example, researchers examined what makes employees stay at work; that is, their job embeddedness (Lee et al., 2014). As reviewed by Lee et al. (2014), job embeddedness includes two prongs: job-related factors and community-related factors. Links to others at work (in the military, this would include constructs such as cohesion), sacrifices in terms of perks or other benefits that might be lost by leaving the job (in the military this would include the pension benefit), and fit of the person to the demands of the job itself are the job-related factors. Similar equivalents for community-related factors, such as links with community members (potentially both soldier and family unit ties with the military community) exist, as well. Both job-related and community-related factors have been found to influence retention. Smith et al. (2011) posited that for military members, a relevant community may be the military itself, although they did not explore this particular aspect of embeddedness.
Role of the family in embeddedness and retention
In addition to community-related factors, family is part of the relevant context in which decisions to stay or leave employment are made, and Army spouses may themselves have ties to the community that affect soldiers’ retention. Investigation into the role of the spouse and family in soldier retention showed that spouse attitudes and factors such as spouse employment were relevant (Orthner, 1990). The research also took into consideration community ties provided by military support services, as well as factors such as spouse satisfaction with and adaptation to the military way of life (Orthner & Bowen, 1990). Although not a test of community embeddedness per se, Burrell et al. (2003) examined whether integration into the military community (indexed by involvement in various Army activities and social connections) affected spouse preferences that their soldier remain in the military, and found that it did.
Research consistently finds that service members’ perceptions of their spouses’ attitudes, and sometimes spouse attitudes themselves, toward the military are highly correlated with their soldiers’ attitudes (Bowen, 1986; Heilmann et al., 2009; Segal & Harris, 1993) and actual servicemember retention – that is, whether or not soldiers stay or leave (Huffman et al., 2014). More recent research has linked military spouses’ own attitudes toward their service members’ retention to actual retention behavior (Campbell et al., 2017; Woodall et al., 2022). Campbell et al. looked at the relationship between spouse support for their members’ retention and servicemember retention two years later, but did not explore the underlying mechanisms through which this happened. Woodall et al. (2022) found spouses’ perceived social support and conflicts between military demands and family life influenced servicemember retention through satisfaction with military life. They suggest that support provided by military programs could improve retention by helping reduce the stresses of military life, though they did not test this empirically. Similarly, although Burrell et al. (2003) did not find that spouse use of Army resources was associated with spouse health outcomes, they speculated that use of resources when they are needed to help with problems could improve retention by alleviating spouse stress.
Use of resources to ameliorate the challenges of military life
Several studies have found that improved service member quality of life is related to improved service member satisfaction with the military and retention intentions (e.g., Hindelang et al., 2004; Kelley et al., 2006; Wilcove et al., 2003, 2009). Furthermore, research has demonstrated that spouse use of military resources for help with problems is associated with increased support for their soldier staying in the military (e.g., Burrell et al., 2003; Kelley et al., 2006; Schwerin et al., 2002) and increased servicemember retention (Knapp et al., 2019). Other research has used a coping and problem-solving framework to explore the relationship between spouses’ (and soldiers’) use of military programs and services to ameliorate the demands of military life and their attitudes toward the military and retention. Research by Sims et al. (2017) found that soldiers’ ability to use resources to resolve their problems was associated with increased retention intentions and more positive general attitudes toward the Army compared to those who used resources and did not have their needs met. Research using a similar problem-solving model found that Army spouses who used resources and had their needs met expressed a greater desire for their soldier to stay in the military, more positive general attitudes toward the Army, and lower levels of stress (see Trail et al., 2019). A similar relationship between use of resources and spouse attitudes toward retention has been found in the Air Force (Sims et al., 2019).
The current research builds on the problem-solving models used by Sims et al. (2017) and Trail et al. (2019). The model specifies that families may perceive some problems as solvable by relying on the resources of the family unit itself without need for assistance from outside the family unit. For other problems, families might want outside help but do not know how to navigate the resource landscape to access available resources. If resources are accessed, they may either help the family solve the problem (i.e., meet the needs) or not (i.e., unmet needs). Finally, the extent to which spouses have problems, need resources, and have their problems resolved by those resources will have implications for important outcomes, both psychological (such as attitudes toward the military, perceived stress, and attitudes toward retention) and behavioral (soldier retention in the Army). In a parallel to the employment retention literature in which stressful shocks can spark a reevaluation of an employee’s desire to stay with a company, use of resources, particularly when unsuccessful, may even serve as a sufficient shock to spouses’ perceptions of Army life that it directly affects their attitudes toward their soldier staying in the Army as well, which in turn affects soldier retention.
Model of spouse attitudes, resource use, and soldier retention
Based on this past research, we developed and tested a theoretical model of the relationship between spouse attitudes and resource use and soldier retention. In our review of the research on employee retention more generally and soldier retention more specifically, we find that spouses’ use of resources to alleviate the challenges posed by military life, the level of stress spouses experience, spouses’ general attitudes toward the military, and spouse specific attitudes toward retention are likely factors in soldiers’ retention in the Army. Based on this literature, we developed a model of the relationship between spouses’ use of resources and soldier retention with the following hypotheses (see Figure 1):
Hypothesis 1:
As demonstrated by previous research and specified in TRA and TPB theories, we expect that spouse positive attitudes toward retention should directly influence actual soldier retention in the Army.
Hypothesis 2:
We predict that spouses’ specific attitudes toward retention are directly influenced by spouse general attitudes toward the military, perceived stress, and the successful or unsuccessful use of resources to alleviate military life’s challenges. Positive attitudes, lower stress, and successful use of resources is associated with retention while negative attitudes, higher stress, and unsuccessful use of resources would be associated with turnover.
Hypothesis 3:
Finally, we predict that two additional pathways influence retention: the successful use of resources to alleviate military life’s challenges should predict lower stress and higher general attitudes toward the military, which themselves predict retention via specific attitudes toward retention (as predicted in Hypothesis 2). Thus, as suggested by Woodall et al. (2022) and Burrell et al. (2003), the successful use of resources to support Army spouses in relieving the stresses of military life should be associated with reduced perceptions of stress and improved attitudes toward the military.
Figure 1.
Theoretical model of relationships between spouse attitudes and experiences and soldier retention.
Method
We examined these relationships using data from three sources: 1) Today’s Army Spouse Survey (TASS); 2) Defense Manpower Data Center (DMDC) data (e.g., Defense Enrollment Eligibility Reporting System [DEERS]); and 3) Army personnel files (e.g., Total Army Personnel Database [TAPDB]). The TASS contains the key predictors of retention: self-reported attitudes and resource use. We matched responses from spouses who completed the survey, fielded January-April 2018, to military records of the soldiers they were married to at the time of the survey in TAPDB and DEERS and analyzed whether their soldier remained in the active component Army as of December 2019 .
Survey sampling and participants
As detailed in Trail et al. (2019), the TASS was designed to assess the problem-solving process of spouses of active component Army soldiers stationed in the continental United States (CONUS). It took a stratified random sample of 75,000 married soldiers, with surveys sent to the spouses of the selected soldiers. Stratification was used to ensure representation of sampled soldiers in terms of soldier paygrade (junior enlisted, E1-E4; senior enlisted, E5-E9; junior officers, O1-O3; and senior officers, O4 and above), presence of dependent children, housing location (on- or off-post), and the urbanicity of the soldier’s assigned garrison (urban, rural, or mid-size city). Spouses were invited to participate in an online survey via mailed postcards and were compensated with a $10 Amazon.com gift card for completing the survey.2
A total of 8,636 spouses accessed the survey and 8,275 provided enough information for our analysis (a response rate of 11.0%; Trail et al., 2019).3 We constructed weights to account for observed differences between responding and non-responding spouses. The current project reports the results of 7,510 participants for whom all data in the analysis models were available. Table 1 displays the weighted demographic characteristics of the project participants.
Table 1.
Weighted participant characteristics.
| Variable | Source | Percent or mean | SD |
|---|---|---|---|
| Soldier’s Active Component status as of December 2019 | TAPDB | 78.8% | – |
| Spouse attitude toward retention (scale 1–5)ab | TASS | 3.78 | 1.30 |
| Spouse’s general attitudes toward the military ab | TASS | −.02 | 0.70 |
| Spouse perceived stress ab | TASS | 2.41 | 0.81 |
| Spouse resource use and unmet needs: Problem solving status categoryb | TASS | ||
| No problems | 4.9% | – | |
| Problems, no needs | 16.9% | – | |
| Problems, needs, no resource use | 7.6% | – | |
| Problems, needs, resource use, unmet needs | 22.6% | – | |
| Problems, needs, resource use, needs met | 48.1% | – | |
| One or more PCS moves since survey | TAPDB | 41.4% | – |
| One or more dependent children | DEERS | 78.3% | – |
| Soldier is Female | TAPDB | 6.3% | – |
| Soldier has deployed since survey | Deployment data | 15.0% | – |
| Soldier deployed in the year prior to the survey | Deployment data | 16.8% | – |
| Number of soldier deployments since Sept. 2001 | Deployment data | 1.73 | 1.97 |
| Soldier’s years of service as of the survey date | TAPDB | 8.98 | 6.77 |
| Soldier paygrade | TAPDB | ||
| Junior Enlisted (E1-E4) | 32.0% | – | |
| Senior Enlisted (E5-E9) | 49.6% | – | |
| Junior Officer (O1-O3) | 11.2% | – | |
| Senior Officer (O4 or higher) | 7.1% | – | |
| Spouse employment status at time of surveyb | TASS | ||
| Employed full-time | 27.3% | – | |
| Employed part-time | 14.2% | – | |
| Unemployed, looking for work | 15.1% | – | |
| Unemployed, not looking for work | 43.3% | – | |
| Spouse was a student at time of surveyb | TASS | 16.9% | – |
| Housing location relative to soldier’s military postb | TASS | ||
| On-post | 31.9% | – | |
| Less than 5 miles away | 8.9% | – | |
| 5–10 miles away | 2.9% | – | |
| 11–20 miles away | 21.8% | – | |
| 21–40 miles away | 8.6% | – | |
| More than 40 miles away | 7.9% | – |
aHigher values indicate more favorable attitudes and higher stress levels, respectively. b Percentages may differ slightly from Trail et al. (2019) because the current analyses restricted respondents to those for whom we have complete information on all the variables.
TASS: Today’s Army Spouse Survey.
Measures
Our main outcome of interest was each soldier’s status in the active component of the Army as of December 2019: i.e., their retention status.4 Retention is a binary variable taking value 1 if the soldier is listed in DEERS as being in the active component of the Army as of December 2019 and 0 otherwise.
Spouse specific attitudes toward retention
Spouse specific attitudes toward their soldier remaining in the military was measured by their rating of one item: “How much do you favor your soldier staying or leaving the military”, adapted from the 2015 Survey of Active Duty Spouses (Defense Manpower Data Center [DMDC], 2015). Responses were measured on a five-point scale ranging from 1 = “I strongly favor leaving” to 5 = “I strongly favor staying” (overall mean = 3.78). Analyses exclude respondents who endorsed that this question was not applicable to them because their service member was retiring soon (n = 467).
Perceived stress scale
We used the 4-item version of the Perceived Stress Scale, which is a reliable and valid measure of experienced stress (Cohen & Williamson, 1988). The four items ask participants to state how often during the past month they have felt a certain way, for example, “that you were unable to control the important things in your life.” Responses to each item ranged from 0 (Never) to 4 (Very Often). We computed the average score on the scale (overall mean = 2.41). Reliability was acceptable, Cronbach’s alpha = .75.
Attitudes toward the military
We used a nine-item scale to assess spouses’ general attitudes toward the military (Pittman et al., 2004), modified and used by Sims et al. (2017) for a survey of Army soldiers. Sample items include: “In general, how well have you adjusted to the demands of being in the Army community?” “How much of a problem, if at all, is … The demands the Army makes on my spouse’s personal time?” and “Overall, how satisfied are you with the military way of life?” Cronbach’s alpha for nine items was .87. As the items were on different scales, we transformed each item to have a mean of 0 and a standard deviation of 1 prior to calculating the average.
Spouse resource use and unmet needs: Problem-solving process
To operationalize the problem-solving process, the TASS assessed the most significant problems each spouse experienced in the past year and how they sought to resolve the problems. Respondents read a detailed list of 96 issues they might have experienced (e.g., difficulty paying debt or bills) and indicated which of these issues, if any, they had experienced in the past year. Issues were grouped into nine problem domains and included an “other” category. If spouses indicated they had experienced issues in more than two domains, they were asked to choose the two domains that contained “the most significant problems” they experienced in the past year.
Respondents experiencing no problems
Respondents who did not endorse any issues were asked to confirm that they had experienced no problems in the past year. About 5% of spouses indicated that they had not experienced any problems (Table 1) and skipped the remaining problem-solving questions.
Respondents experiencing problems but no needs
Respondents who indicated that they experienced problems were next asked what kinds of needs for help they had for each of their top two problems. The survey included a list of potential needs (e.g., emotional or social support; activities for fitness, stress relief and bonding) and respondents selected those they experienced. If respondents selected more than two needs for a problem, they were asked to select the top two types of needs.
Respondents were given the option to indicate that they had no need for assistance with the problem, and respondents who selected this option for both of their top problems were categorized as having problems, but no needs. As detailed in Table 1, 16.9% of spouses indicated that they had experienced problems in the past year but that they had no need for assistance with their problems. These respondents skipped the remaining questions on problem-solving.
Respondents experiencing problems and needs, but not using resources
Respondents who indicated that they experienced problems and had needs for help were next asked to indicate all the resources they reached out to for help with each need. The survey included a list of military and nonmilitary resources, including an “other” category, and respondents were asked to indicate if they had contacted each resource. For the list of military resources, respondents could also indicate that they did not contact any military resources for help with the need. Similarly, for nonmilitary resources, respondents could indicate that they did not contact any nonmilitary resources for help with the need. Respondents who selected both of these options for all of their needs were categorized as having problems, needs, but not using resources to help meet those needs. As detailed in Table 1, 7.6% of spouses indicated that they had experienced problems, had needs for assistance, but did not reach out to any resources for help. These respondents skipped the remaining questions on problem-solving.5
Whether respondents experiencing problems and needs and using resources had their needs met
Respondents who reported using one or more resources for help were asked whether they had received the help they needed by responding “yes,” “no,” or indicating they were not sure. Across problems and needs, we considered spouses to have an unmet need if they indicated “no” or they were “not sure” they had received the help they needed. Else, if respondents indicated that all their needs were met by the resources they used, they were counted as having their needs met. As detailed in Table 1, 22.6% of spouses indicated that they had unmet needs. Conversely, 48.1% of spouses reached out to resources for help and had all of their needs met.
Analysis plan
Our analysis strategy followed the theoretical model displayed in Figure 1, using a series of regressions. We first considered the relationships between spouse attitudes and stress perceptions and soldier retention. We estimated four models with logistic regression to determine whether three key variables – spouse specific attitudes toward retention, general attitudes toward the military, and perceived stress – predict retention while controlling for soldier and spouse demographic characteristics (described below).
A second set of models examined relationships with spouses’ specific attitudes toward retention using linear regression analysis. As shown in Figure 1, our key predictors in the set of models predicting specific attitudes toward retention were spouse general attitudes toward the military, perceived stress, and spouse problem-solving category.
Covariates: Characteristics known to be associated with soldier retention
In addition to primary predictors, our analysis controlled for spouse and soldier characteristics shown in previous research to be related to service member retention or spouse experiences. These covariates are shown in Table 1, and include family PCS moves since the 2018 TASS, soldier’s total number of deployments since September 2001 and deployment before or just prior to the TASS, soldier’s years of service, soldier paygrade, spouse employment status, and where spouses lived relative to their soldier’s military post at the time of the survey.
Results
Predicting soldier retention from spouse attitudes and stress
Table 2 presents results from our main models of interest: the effect of spouse attitudes and stress on soldier retention. Results are from logistic regression, and each of the models include the covariates.
Table 2.
Logistic regression results predicting soldier retention from key predictor variables (N = 7,510).
| Key Predictors’ Average partial effect |
||||
|---|---|---|---|---|
| Model 1: Specific Retention Attitudes | Model 2: General Attitudes | Model 3: Stress | Model 4: All Key Predictors | |
| Spouse’s specific retention attitudes | 0.055*** | 0.053*** | ||
| Spouse’s general attitudes toward the military | 0.064*** | 0.005 | ||
| Spouse perceived stress | −0.024*** | −0.007 | ||
| Model Pseudo R2 | 0.28 | 0.25 | 0.24 | 0.28 |
***p < .001, **p < .01, *p < .05. Each regression model included the covariates PCS moves since survey, presence of dependent children, distance living from post; spouse employment status and student status; soldier deployment since survey, deployment in the year immediately prior, and number of deployments since 2001; and soldier years of service at survey date, gender, and paygrade.
The columns displaying models 1, 2 and 3 are the results including the variables of interest one at a time, and the last column shows the results including all three variables simultaneously: spouse’s specific attitudes toward retention, general attitudes toward military life, and perceived stress level. The first three columns show that each variable is a significant predictor of soldier retention in the expected direction. These models provide evidence that spouse attitudes and experience of stress have a significant effect on soldier retention.
The results from model 4 support the theory that specific attitudes toward retention drive the relationship between general spouse attitudes toward the military and stress and soldier retention. When all three variables are entered into the model, the effect of specific retention attitudes hardly changes (from 5.5 to 5.3% points on average) and remains significant. In contrast, once specific attitudes are accounted for, the effect of general attitudes toward the military or stress are not significantly different from zero. Furthermore, the pseudo-R2 is unchanged between the model including only specific attitudes and the model with all three predictors, providing additional evidence that once specific attitudes are accounted for, spouse general attitudes and stress have little additional predictive power for soldier retention.
Predicting spouse specific attitudes toward retention
Our next set of models uses linear regression to predict specific attitudes toward retention from general attitudes toward the military, perceived stress, and problem-solving status. The results are presented in Table 3. We first consider each key variable individually as a predictor of spouse specific attitudes toward retention, and then enter all three variables together.
Table 3.
Linear regression results predicting spouses’ attitudes toward retention from key predictor variables (N = 7,510).
| Key Predictors’ Coefficients |
||||
|---|---|---|---|---|
| Model 1: General Attitudes | Model 2: Stress | Model 3: Problem Solving Status | Model 4: All Key Predictors | |
| Spouse’s general attitudes toward the military | 0.946*** | 1.040*** | ||
| Spouse perceived stress | −0.274*** | 0.103*** | ||
| Spouse resource use and unmet needs: Problem solving status category, compared to those who used resources and had their needs met | ||||
| No problems | 0.312*** | −0.272*** | ||
| Problems, no needs | 0.132** | −0.076* | ||
| Problems, needs, no resource use | −0.046 | −0.026 | ||
| Unmet needs | −0.265*** | 0.084* | ||
| Model R2 | 0.32 | 0.10 | 0.09 | 0.32 |
***p < .001, **p < .01, *p < .05. Each regression model included the covariates PCS moves since survey, presence of dependent children, distance living from post; spouse employment status and student status; solder deployment since survey, deployment in the year immediately prior, and number of deployments since 2001; and soldier years of service at survey date, gender, and paygrade.
When entered individually, spouses’ general attitudes toward the military and spouse perceived stress each have a significant relationship with specific attitudes toward retention with the effect in the expected direction. The relationship between the problem-solving status categories and specific attitudes is shown in the next column. Here the results are relative to a reference group of spouses with problems and needs, who use resources, and whose needs are met. As shown in Trail et al. (2019), spouses with no problems had relatively more positive specific attitudes toward retention than those who had problems but whose needs were met, and spouses with unmet needs had lower specific attitudes toward retention than spouses whose needs were met. The results show that of spouses who had needs, those who were successfully able to use resources had more positive specific attitudes toward retention than those who either didn’t use resources or for whom the resources used did not meet their needs.
The final column contains the results of a model predicting specific attitudes toward retention using all three variables: spouse general attitudes toward the military, perceived stress, and the categories of problem-solving status. The association between general attitudes toward the military and specific attitudes toward retention are consistent with model 1, which included only general attitudes toward the military. On the other hand, the results for both stress and problem-solving category change substantially in the combined model, changing in both direction and magnitude. When general attitudes toward the military are controlled for, stress has a positive relationship with specific attitudes toward retention.
We see similar changes in the problem-solving status categories. When entered alone, spouses with no problems have more positive specific attitudes toward retention. Once general attitudes toward the military are controlled for, spouses with no problems have, on average, attitudes toward retention that are 0.272 points lower than spouses who had problems and whose needs were met through resource use. Note that the R2 is the same for the model including only general attitudes toward the military and the model with all three variables. The fact that the R2 does not increase indicates that the association between the three variables and specific attitudes toward retention is driven by general attitudes toward the military. We discuss some potential explanations for this finding in the discussion, but first consider the relationship of problem-solving status with other variables in the model.
Predicting general attitudes and stress from problem-solving status
To follow-up on this observation and to flesh out the theoretical model proposed in Figure 1, we present two additional analyses: one planned linear regression model predicting perceived stress from problem-solving status, and one post-hoc model predicting general attitudes toward the military from problem-solving status and stress. Replicating the results from Trail et al. (2019), problem-solving status significantly predicted perceived stress. Compared to spouses who used resources and had their needs met, spouses who had unmet needs reported higher levels of perceived stress (b = 0.278 [SE = 0.024], t[7,488] = 11.69, p < .001), while those with no problems (b = −0.688 [SE = 0.042], t[7,488] = −16.41, p < .001) or problems and no needs reported lower levels of stress (b = −0.340 [SE = 0.026], t[7,488] = −13.24, p < .001). Spouses with problems and needs who did not use resources did not significantly differ from spouses who used resources and had their needs met, (b = −0.027 [SE = 0.036], t[7,488] = −0.75, p = .45).
In the post-hoc model, both perceived stress and problem-solving status category are significant predictors of general attitudes toward the military, both when entered into separate models (stress: b = −0.391 [SE = 0.010], t[7,491] = 39.52, p < .001; unmet needs: b = −0.364 [SE = 0.021], t[7,488] = 17.36, p < .001) and when entered together into the same model predicting general attitudes (stress: b = −0.331 [SE = 0.010], t[7,487] = 32.76, p < .001; unmet needs: b = −0.272 [SE = 0.091], t[7,487] = 14.01, p < .001).
Discussion
The current report adds to a small body of work demonstrating that providing benefits to address needs of military spouses is associated with a tangible and important military outcome: improved servicemember retention. It suggests that studying the utility that spouses gain from these programs is of value for its potential to enable families, as well as soldiers, to accommodate military life. We tested a model detailing the mechanisms by which using family support programs contributes to servicemember retention. Drawing from the general literature on retention and turnover as well as the TRA, we theorized that military spouses’ ability to get help from available programs to resolve problems would be associated with improved retention. This would occur because successful problem resolution would decrease stress and improve general attitudes toward the military. All three of these factors would, in turn, improve spouses’ specific attitudes toward retention. Our results should also generalize to other Services given their basis in general theories of the human condition.
The results of our analysis generally support this theory, with one important change: problem-solving and stress were not directly associated with spouses’ specific attitudes toward their soldier staying in the military, but the association was accounted for by relationship with general attitudes toward the military. Worse problem-solving outcomes and greater stress were associated with worse general attitudes toward the military. Worse general attitudes toward the military were associated with less inclination to stay a military family, which in turn predicted retention: worse attitudes were associated with turnover. The revised model based on these results is shown in Figure 2. The finding that family influences affect turnover is not novel but rather is rising in importance in the turnover literature (Hom et al., 2017). Indeed, some empirical work even suggests that the role of families and their satisfaction with military life may play a larger role in military turnover than military members’ own job attitudes (Heilmann et al., 2009). Ramesh and Gelfand (2010) suggest that one organizational intervention to decrease turnover may be to appeal to families and develop their embeddedness with the organization. Here, we discuss a mechanism by which provision of resources and programs may serve to keep soldiers and their families in the military. We anticipated direct effects of stress and program use upon specific attitudes toward retention but found instead that these were mediated by general attitudes toward the military. This suggests that challenges in successfully using the programs intended to help manage military life may initiate a general evaluative process, similar to the unfolding model framework for turnover in which an event or shock precipitates a reevaluation of desire to stay with the organization.
Figure 2.
Model of relationships between spouse attitudes and experiences and soldier retention (modified).
One way of interpreting the revised model is provided by attribution theory (Heider, 1958). Attribution theory describes how individuals make judgments about the causality of phenomena, such as behaviors, thoughts, and feelings. In the current study, we consider the example of having a problem and a need for help, using a resource in an effort to meet that need, but not having the need met. A spouse would want to explain why they had an unmet need – they would want to attribute their experience of having unmet needs to some cause, be it something about themselves (internal attribution), or something about the situation, environment, or some other person (external attribution). Internal attributions could include that they used the wrong resource, had a problem that needed more intense help than they sought out, or that their problem is not easily solved. External attributions for having unmet needs could include explanations like someone gave them bad advice on which resource to use, the resource they used was of poor quality, or that the resource was not willing to do what was needed to help them.
To the extent that spouses attribute their unmet needs or experience of stress to some characteristic of the Army – for example, that the Army makes it difficult for families to find resources, doesn’t provide good resources to families, or doesn’t care about family needs – then their attributions might lead them to have more negative attitudes toward the military, generally. To the extent that spouses attribute their experiences to an internal source or to some other external source (e.g., thinking “Army resources are generally good, I just happened to use a less effective resource”), then their experiences would not be seen as caused by the Army and their attitudes would not be negatively affected by these experiences. To the extent that attributions to the Army can lead spouses to have negative attitudes toward the Army, which leads them to want their soldier to leave the military and then to actually leave the service, it is possible that addressing spouses’ attributions of their experiences of stress and having unmet needs to the Army – whether correct or incorrect – could be important for improving soldier retention.
Study limitations
Although this study has several strengths, including a large representative sample of CONUS Army spouses, controlling for known influences on soldier retention, and measuring actual soldier retention two years after the spouse survey, the study is also limited in important ways that constrain the strength of the conclusions that can be drawn from the results. One issue is that our sample was limited to CONUS Army spouses rather than including OCONUS spouses. Thus, we can only claim to be representative of CONUS Army spouses rather than all Active Component Army spouses.
Additionally, two years is not a long time to track soldier retention. It is possible that the specific relationships we measured in our model would become stronger or weaker if retention was measured at a point more distant in the future. In addition, the models did not account for whether soldiers had a decision point to remain in the military during our timeframe. Contracts for enlisted soldiers vary in length and can be as short as two years6 and as long as six, with the modal length of three years (Asch et al., 2010). Officer contracts are not as definitive as enlistment contracts, and aside from the initial obligation there is not a general timeframe for making decisions about remaining in the Army. This means that there was likely variance in the factors associated with continuing in the service that we did not account for, particularly between enlisted soldiers and officers. However, we did take steps to address this issue. We controlled for paygrade, and also removed from the analyses those spouses who said that their soldier was planning on retiring soon. We also examined separate models with spouses of officers and enlisted soldiers, and the results from these models were similar. While still possible that accounting for the occurrence of a decision point for remaining in the Army during the analysis period would have changed the results, soldiers without a decision point during the project would essentially be “noise” in the data and our results are actually more conservative than if we had limited the analysis to spouses whose soldier had to make a decision to stay or leave the Army.
Although we assessed soldier retention two years following the survey, all the other variables in the model were measured at the time of the survey. This means that the causal relationship between variables in the model, particularly the attitudinal ones, cannot be definitively determined. Thus, although the model assumes that, for example, unmet needs lead to worse attitudes toward the military, the fact that both attitudes and unmet needs were measured at the same time point means that it is possible that spouses with worse attitudes toward the military are also more likely to have unmet needs.
Finally, although we offer an interpretation of our results based on attribution theory, we did not measure actual attributions. Future research should measure actual causal attributions to the military, as well as a broader array of spouse well-being measures (e.g., quality of life, depression, anxiety) to replicate and confirm our interpretation of the results.
Conclusions
Given the resources the Army invests in its soldiers, it has a continued vested interest in ensuring that it is able to retain the best soldiers. It is clear that spouses are relevant to this goal: this report lays out the theoretical rationale tying the spouse experience to the larger turnover literature and articulates a process by which this influence is felt. It is notable that the experience of solving problems and navigating the resources provided by the military to help spouses alleviate challenges posed by Army life affects spouse attitudes. This suggests that continued attention to integrating spouses into the military community and alleviating the challenges they face is beneficial from both a moral and a practical perspective.
Supplementary Material
Funding Statement
This research was funded by the Deputy Chief of Staff, G-9 (formerly known as Assistant Chief of Staff for Installation Management), United States Army, and was conducted within the RAND Arroyo Center under Contract [W91CRB-15-D-0022].
Notes
While spouses surveyed in this way adhere to the DoD’s definition of dependent spouse, this omits nonmarried partners who may also be influencing Soldiers’ retention decisions.
Of the 75,000 spouses sampled, a match with United States Postal Service address records 74,509 valid addresses for spouses recruited to participate in the survey. See Trail et al. (2019) for a discussion of response rates and survey bias.
Although we use the term retention, our analyses did not account for whether soldiers had a decision point to remain in the military during the timeframe under study. Thus, we did not explicitly model the choice to remain in the military after their obligated term of service had ended, but rather the fact that their status was active duty in the most recent data, sometimes known as continuation.
There could be many reasons for not seeking help through resources. These range from not knowing who to contact for help to not wanting to ask for help to being in the process of resolving the issue themselves. Prior work demonstrated that spouses who had problems and needs but did not seek resources were largely similar on a variety of outcomes to spouses who had problems, needs, reached out to resources, and had their needs met. See Trail et al. (2019) for exceptions and more detail.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
The participants of this study did not give written consent for their data to be shared publicly. Moreover, the data that support the findings of this study are available from Defense Manpower Data Center and Human Resource Command. Restrictions apply to the availability of these data.
IRB
RAND operates under a “Federal-Wide Assurance” (FWA00003425) and complies with the Code of Federal Regulations for the Protection of Human Subjects Under United States Law (45 CFR 46), also known as “the Common Rule,” as well as with the implementation guidance set forth in DoD Instruction 3216.02. As applicable, this compliance includes reviews and approvals by RAND’s Institutional Review Board (the Human Subjects Protection Committee) and by the U.S. Army. The views of sources utilized in this study are solely their own and do not represent the official policy or position of DoD or the U.S. Government.
Supplementary material
Supplemental data for this article can be accessed online at https://doi.org/10.1080/08995605.2024.2319014.
References
- Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In Kuhl J. & Beckmann J. (Eds.), Action control: From cognition to behavior (pp. 11–39). Springer-Verlag. [Google Scholar]
- Asch, B. J., Heaton, P., Hosek, J., Martorell, P., Simon, C., & Warner, J. T. (2010). Cash incentives and military enlistment, attrition, and reenlistment (Report No. MG-950-OSD). RAND Corporation. https://www.rand.org/pubs/monographs/MG950.html [Google Scholar]
- Bowen, G. L. (1986). Spouse support and the retention intentions of air force members. Evaluation and Program Planning, 9(3), 209–220. 10.1016/0149-7189(86)90018-2 [DOI] [Google Scholar]
- Burrell, L., Durand, D. B., & Fortado, J. (2003). Military community integration and its effect on well-being and retention. Armed Forces & Society, 30(1), 7–24. 10.1177/0095327X0303000101 [DOI] [Google Scholar]
- Campbell, A., Luchman, J., & Kuhn, J. (2017). Spousal support to stay as a predictor of actual retention behavior: A logistic regression analysis (Survey Note N. 2017-009). Defense Research, Surveys, and Statistics Center, Office of People Analytics. [Google Scholar]
- Cohen, S., & Williamson, G. (1988). Perceived stress in a probability sample of the United States. In Spacapan S. & Oskamp S. (Eds.), The social psychology of health: Claremont symposium on applied social psychology (pp. 31–67). SAGE Publishing. [Google Scholar]
- Defense Manpower Data Center . (2015). 2015 survey of active duty spouses: Tabulations of responses (Report 2015-028).
- Fafara, R., & Westhuis, D. (2007). Morale, welfare, and recreation programs and their effect on readiness and retention. US Army Journal of Installation Management, 2, 10–14. [Google Scholar]
- Fishbein, M., & Ajzen, I. (1975). Beliefs, attitude, intention, and behavior: An introduction to theory and research. Addison-Wesley. [Google Scholar]
- Griffeth, R. W., Hom, P. W., & Gaertner, S. (2000). A meta-analysis of antecedents and correlates of employee turnover: Update, moderator tests, and research implications for the next millennium. Journal of Management, 26(3), 463–488. 10.1177/014920630002600305 [DOI] [Google Scholar]
- Hale, J. L., Householder, B. J., & Greene, K. L. (2002). The theory of reasoned action. The persuasion handbook: Developments in theory and practice, 14(2002), 259–286. [Google Scholar]
- Heffner, T. S., & Gade, P. A. (2003). Commitment to nested collectives in special operations forces. Military Psychology, 15(3), 209–224. 10.1207/S15327876MP1503_04 [DOI] [Google Scholar]
- Heider, F. (1958). The psychology of interpersonal relations. Wiley. [Google Scholar]
- Heilmann, S. G., Bell, J. E., & McDondald, G. K. (2009). Work-home conflict: A study of the effects of role conflict on military officer turnover intention. Journal of Leadership and Organizational Studies, 16(1), 85–96. 10.1177/1548051809334194 [DOI] [Google Scholar]
- Hindelang, R. L., Schwerin, M. J., & Farmer, W. L. (2004). Quality of life (QOL) in the U.S. Marine Corps: The validation of a QOL model for predicting reenlistment intentions. Military Psychology, 16(2), 115–134. 10.1207/S15327876MP1602_3 [DOI] [Google Scholar]
- Holtom, B. C., Mitchell, T. R., Lee, T. W., & Inderrieden, E. J. (2005). Shocks as causes of turnover: What they are and how organizations can manage them. Human Resource Management, 44(3), 337–352. 10.1002/hrm.20074 [DOI] [Google Scholar]
- Hom, P. W., Lee, T. W., Shaw, J. D., & Hausknecht, J. P. (2017). One hundred years of employee turnover theory and research. Journal of Applied Psychology, 102(3), 530–545. 10.1037/apl0000103 [DOI] [PubMed] [Google Scholar]
- Huffman, A. H., Casper, W. J., & Payne, S. C. (2014). How does spouse career support relate to employee turnover? Work interfering with family and job satisfaction as mediators. Journal of Organizational Behavior, 35(2), 194–212. 10.1002/job.1862 [DOI] [Google Scholar]
- Kelley, M. L., Schwerin, M. J., Farrar, K. L., & Lane, M. E. (2006). Evaluation of the Navy new parent support program. Journal of Family Violence, 21(5), 301–310. 10.1007/s10896-006-9031-5 [DOI] [Google Scholar]
- Knapp, D., Marrone, J. V., Miller, L. L., & Trail, T. E. (2019). The impact of a spouse incentive on employee retention: Evidence from a military spouse scholarship (Report No. WR-1295-OSD). RAND Corporation. 10.7249/WR1295 [DOI] [Google Scholar]
- Lee, T. W., Burch, T. C., & Mitchell, T. R. (2014). The story of why we stay: A review of job embeddedness. Annual Review of Organizational Psychology and Organizational Behavior, 1(1), 199–216. 10.1146/annurev-orgpsych-031413-091244 [DOI] [Google Scholar]
- Lytell, M. C., & Drasgow, F. (2009). “Timely’ methods: Examining turnover rates in the U.S. Military. Military Psychology, 21(3), 334–350. 10.1080/08995600902914693 [DOI] [Google Scholar]
- Orthner, D. K. (1990). Family impacts on the retention of military personnel (Research Report 1556). U.S. Army Research Institute for the Behavioral and Social Sciences. [Google Scholar]
- Orthner, D. K., & Bowen, G. L. (1990). Family adaptation in the military (ARI Research Report 1559, AD A225 085). U.S. Army Research Institute for the Behavioral and Social Sciences. [Google Scholar]
- Pittman, J. F., Kerpelman, J. L., & McFadyen, J. M. (2004). Internal and external adaptation in army families: Lessons from operations desert shield and desert storm. Family Relations, 53(3), 249–260. 10.1111/j.0197-6664.2004.0001.x [DOI] [Google Scholar]
- Ramesh, A., & Gelfand, M. J. (2010). Will they stay or will they go? The role of job embeddedness in predicting turnover in individualistic and collectivistic cultures. Journal of Applied Psychology, 95(5), 807–823. 10.1037/a0019464 [DOI] [PubMed] [Google Scholar]
- Schwerin, M. J., Michael, P. G., Glaser, D. N., & Farrar, K. L. (2002). A cluster evaluation of Navy quality of life programs. Evaluation and Program Planning, 25(3), 303–312. 10.1016/S0149-7189(02)00024-1 [DOI] [Google Scholar]
- Segal, M. W., & Harris, J. J. (1993). What we know about army families (Special Report 21). U.S. Army Research Institute. [Google Scholar]
- Sims, C. S., Miller, L. L., Trail, T. E., Woods, D., Kofner, A., Rutter, C. M., Posard, M. N., Hall, O., & Kleykamp, M. (2019). 2017 U.S. Air Force Community’F eedback Tool: Key results report for Air Force headquarters (Report No. RR-3084-AF). RAND Corporation; 10.7249/RR3084 [DOI] [Google Scholar]
- Sims, C. S., Trail, T. E., Chen, E. K., & Miller, L. L. (2017). Today’s soldier: Assessing the needs of soldiers and their families (Report No. RR-1893-A). RAND Corporation; 10.7249/RR1893 [DOI] [Google Scholar]
- Smith, D. R., Holtom, B. C., & Mitchell, T. R. (2011). Enhancing precision in the prediction of voluntary turnover and retirement. Journal of Vocational Behavior, 79(1), 290–302. 10.1016/j.jvb.2010.11.003 [DOI] [Google Scholar]
- Trail, T. E., Sims, C. S., & Tankard, M. (2019). Today’s Army Spouse Survey: How Army families address life’s challenges (Report No. RR-3224). RAND Corporation; 10.7249/RR3224 [DOI] [Google Scholar]
- Wilcove, G. L., Schwerin, M. J., & Kline, T. L. (2009). Quality of life in the U.S. Navy: Impact on performance and career continuance. Military Psychology, 21(4), 445–60. 10.1080/08995600903206362 [DOI] [Google Scholar]
- Wilcove, G. L., Schwerin, M. J., & Wolosin, D. G. (2003). An exploratory model of quality of life in the U.S. Navy. Military Psychology, 15(2), 133–152. 10.1207/S15327876MP1502_3 [DOI] [Google Scholar]
- Woodall, K. A., Esquivel, A. P., Powell, T. M., Riviere, L. A., Amoroso, P. J., Standler, V. A., & Millennium Cohort Family Study Team. (2022). Influence of family factors on service members’ decisions to leave the military. Family Relations, 72(3), 1138–1157. 10.1111/fare.12757 [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The participants of this study did not give written consent for their data to be shared publicly. Moreover, the data that support the findings of this study are available from Defense Manpower Data Center and Human Resource Command. Restrictions apply to the availability of these data.