Reliability and Validation of U.S. Army-Oriented Brief Work-to-Family and Family-to-Work Conflict Scales: An Email Sample of 262 Army Career Officers

Abstract

Work–family conflicts (WFCs) and family–work conflicts (FWCs) have been found to be important to worker morale and retention as well as family (as defined by the respondents) well-being, with particular importance within the military, as indicated by a number of studies in the United States, Canada, and European countries. However, few studies have focused on the impact of WFC and FWC for high ranking officers and their families. This study involved two samples of officers attending an advanced leadership course at Fort Leavenworth, Kansas, in 2007, featuring a total of 262 Army officers and another 45 officers from other services and countries. The sample of Army officers included 239 men (85.4% of whom were a parent) and 22 women (63.6% of whom were a parent; Fisher’s Exact Test, p < 0.02), with one missing case for sex. Of the men and women, respectively, most were married for the first time (190/11), with some never married (14/5), married but divorced (10/1), married/divorced/remarried (20/4), married/spouse died/remarried (1/0), married/divorced/remarried/divorced/remarried (4/0), and married/divorced/remarried/divorced/remarried (0/1). Measures for cohabitation or same-sex partnerships were not used. Two modified measures of WFC and FWC of four items each were tested and found to represent different factors and to have high internal consistency reliability. In general we found few sex differences, but female officers seemed to be more influenced by family–work conflict than male officers. One of our most substantial findings was that work–family conflict was more prevalent than family–work conflict within our sample. Also, we found that marital satisfaction tended to be higher than parental satisfaction and that officers usually found their own retention intentions to be higher than their perception of that of their spouses. Satisfaction with the military was consistently and strongly related to lower levels of work–family conflict while similar but weaker trends were found for family–work conflict. Most of our results were found to cross-validate with a subsample of non-Army officers in our sample. We found bias from marital social desirability to be lower for our marital process scale and for FWC than for marital satisfaction and WFC. In sum, our results confirm adverse effects of deployments and other stressors on military families and a continuing need for military support for families, even among higher ranking officer families.

1. Introduction

Role strain is a concept that is defined by excessive role demands on a person. Interrole conflict defines the situation when a person has excessive and/or conflicting/incompatible role demands from more than one source, such as employment and family. Several theories have been used with these concepts, recently comprehensively reviewed by Powell et al. (2019) and Day et al. (2025). Symbolic interaction theory could explain how role strain or interrole conflict could change a person’s perceptions of work or family life as well as the balance or imbalance between them. Conservation of resources (COR) theory (Hobfoll et al. 2012; Frone and Blais 2019; Pickering 2017) posits that when a person uses up too much of their physical or psychological resources, dealing with either high role demands or high levels of role conflicts could reduce their capability and motivation to manage those strains well. Conflicting demands from either work or family, common among military families, can deplete resources otherwise useful for one or both of those roles. When work roles interfere with family roles/responsibilities, that situation is identified as work-to-family conflict (WFC); the reverse situation is identified as family-to-work conflict (FWC). WFC and FWC can occur at the same time (Borowski et al. 2021). Such conflicts may have adverse impacts on mental health and many other outcomes (Pickering 2017). As Huffman et al. (2014) have stated, “The need to simultaneously fulfill both work and family demands is a stark reality for many military personnel, making WFC an important issue for military decision makers” (p. 26). That “stark reality” is associated with the concept of military duty and families as both conflicting and both high demand, “greedy institutions” (Segal 1986; De Angelis and Segal 2015; Vuga and Juvan 2013). One aspect of this greediness is that “Most military organizations require their personnel to view their occupational demands as superceding any competing family demands” (Skomorovsky et al. 2015, p. 810), or as Wadsworth and Southwell said, “military service generates substantial structural, energy, psychological, and behavioral tensions with family life” (p. 163). A popular saying in the US military has been “If the Army had wanted you to have a family, it would have issued you one” (Moelker et al. 2019, p. 10). A television show in the United States, Army Wives, ran for seven years (2007–2013), featuring scenes in which spouses battled over these competing work–family demands, a spin-off of a book on those same issues (Biank 2006). Family systems theory has also been used to study WFC and FWC but is less useful for our study since we only surveyed soldiers, not their spouses (Huffman et al. 2017). We did not take advantage of recent models of work–family enrichment (Cavaleiro et al. 2019; Silveira-Rodrigues et al. 2021; Carvalho and Chambel 2018; Greenhaus and Powell 2006; McNall et al. 2010; Carlson et al. 2006) in the design of our research, however.

Although Frone and Blais (2019) assert that “most research on work-to-family conflict has focused on civilian workers” (p. 8), work–family conflict has been researched in a variety of employment contexts, including Islamic societies (Silveira-Rodrigues et al. 2021), civilian military employees (Moghaddam and Beydokhti 2023), Australian parents (Haslam et al. 2015), civilian employees in the United States (Kelly et al. 2014), Chinese civilians (Jung and Kim 2023), Italian civilians (Loscalzo et al. 2019), military spouses (Park et al. 2023; Pflieger et al. 2018), and military service members (Farnsworth and O’Neal 2025, US Army soldiers; Silveira-Rodrigues et al. 2021, Brazil; Hobfoll et al. 2012, US Air Force service members; Huffman et al. 2014, 2017, USA, Army, garrison duty; Frone and Blais 2019, Canadian non-deployed military personnel; Sachau et al. 2012, US Air Force service members; Skomorovsky et al. 2019, single mothers in the Canadian Armed Forces; Cavaleiro et al. 2019, Portuguese Navy; Carvalho and Chambel 2018, Portuguese Marine Corps; Seo et al. 2014, South Korean Air Force members; Vuga and Juvan 2013, Slovenian military).

Kelly et al. (2014) assert that work–family conflict is “increasingly common among U.S. workers” for up to 70% of such workers (p. 485). Research on work–family conflicts within military families is becoming more common (Ross 2014) and may even apply to non-deployed, garrison situations (Frone and Blais 2019; Huffman et al. 2017) due to a high “level of workload demands” (Frone and Blais 2019, p. 8). Silveira-Rodrigues et al. (2021) have noted that “the military profession always implies some withdrawal from family” (p. 1).

In the military, changes in roles of husbands and wives toward greater equality may lead to increasing conflict between “military work requirements and a member’s home responsibilities” (Dursun 2018, p. 1), or as Moelker et al. (2019) noted, “work-family conflict is sharper today where relations between partners are more equal” (p. 11).

With reference to military work–family conflicts, Pickering (2018) identifies four types of work–life conflict: (1) time-based where a lack of sufficient time occurs to fulfill multiple roles at the same time, (2) behavior-based where required activities in one role interfere with required activities in another role, (3) strain-based where role strain in one area (work) carries over to another area (e.g., family), and (4) energy-based where a person feels a lack of energy to perform their roles adequately across more than one role area (pp. 9–10). As Pickering notes, conflict can occur from work to family or family to work. She also reports that such conflicts may occur for up to half of all military personnel (p. 10). Increases in operational tempo over a decade (2007–2017) for military members of the United States, Canada, and the United Kingdom may have led to increases in these types of conflict (p. 11). Pflieger et al. (2018) extended that period to 15 years (p. 719).

A variety of measures have been used to measure work-to-family (WFC) and family-to-work (FWC) conflict, including Netemeyer et al.’s (1996) two five-item scales (Sachau et al. 2012; Kelly et al. 2014; Huffman et al. 2014, 2017; Frone and Blais 2019; Pluut and Andres 2019; Woodall et al. 2023; Sullivan et al. 2020), Haslam et al.’s (2015) two five-item scales—adapted to Chinese by Jung and Kim (2023), Matthews et al.’s (2010) two three-item scales (Wong et al. 2022), and Carlson et al.’s (2000) longer scales (Bruck and Allen 2003; Huffman et al. 2017; Cavaleiro et al. 2019; Loscalzo et al. 2019; Vieira et al. 2014). Thompson et al. (1999) used a four-item WFC scale from Thompson and Blau (1993). Hakhmigari and Diamant (2025) and Wayne et al. (2004) both used two four-item WFC and FWC scales from the US MIDUS survey. Frone et al. (1992) developed a two-item scale for WFC and FWC (Hobfoll et al. 2012). Grzywacz et al. (2006) used Frone et al.’s (1992) two-item scales with one item added to each scale from Bellavia and Frone (2005). Fisher et al. (2016) attempted to condense six items (four from Matthews et al. 2010) into one item each for WFC and FWC but did not strongly recommend the best single item obtained for each. Geurts et al. (2005) developed, as part of their SWING scale, a nine-item WFC scale and a six-item FWC scale; Pluut and Andres (2019) used five of the SWING scale items to satisfy their need for a five-item FWC scale as well as using Netemeyer et al. (1996) for a five-item WFC scale. Carlson et al. (2006) also have developed a work–family enrichment scale with nine items (Vieira et al. 2014); Geurts et al. (2005) also developed two similar scales with six items each as part of their SWING scale. Borowski et al. (2021) used two items to measure both WFC and FWC. Voydanoff (1988), with a civilian US sample, used a single item to measure WFC: “How much do your job and your family life interfere with each other—a lot, somewhat, not too much, or not at all?” Pflieger et al. (2018) used a single item to measure the military service member’s spouse’s report of work–family conflict, concerning their “difficulty balancing demands of family life and your spouse’s military duties”.

Research for the US Army, Navy, Marine Corps, and Air Force has found that higher levels of work–family conflict are related to lower job satisfaction and higher turnover levels, as reported by Sachau et al. (2012). Borowski et al. (2021) reported, for US veterans, significant correlations between WFC and FWC with depressive symptoms 1.5 years later.

The present study was conducted in 2007, at a unique time in U.S. military history, and may open an important window into the mindset of Army and other military officers at that time. The Iraq war was at a high point with a very high operational tempo and tens of thousands of casualties among active and reserve component soldiers. Combat deployments were occurring at a high frequency, even compared to Pentagon policy (Ross 2014). While “A well-established body of literature indicates that military service members are subject to significant work-family stressors” (Wong et al. 2022, p. 425), as noted by Park et al. (2023, p. 1079), since September 2001, US military members and their families had been experiencing an increase in separation/stress associated with more frequent and longer overseas deployments, with some research finding overseas combat deployments having adverse effects on military families (Cigrang et al. 2014). Thus, some concluded that “conflict between the military and the family may be greater today than in the past” (Pluut and Andres 2019, p. 121). Our military sample was taken from officers in an educational context, a short tour of approximately a year or less, that most likely imposed restrictions on their often highly educated spouses’ ability to obtain satisfactory jobs during that time (Ross 2014, p. 892). On the other hand, at the same time, key researchers were reporting that the high tempo was not stressful on Army families—or at least not in terms of increasing divorce rates, except possibly for female soldiers or enlisted personnel (Ross 2014, p. 893; Allen et al. 2010; Karney and Crown 2007, 2011; Karney and Trail 2016); however, at the same time, the military was trying to provide substantial benefits designed to enhance soldier and family resilience (Wadsworth and Southwell 2011). Regardless, it could be argued that field grade military officers (mostly majors or higher in rank) on an educational tour in the United States would be relatively more privileged and less oppressed than perhaps a few percent of the military as a whole, being possibly well insulated from any effects of work–family issues. However, that would remain an empirical question until actual data could be brought to bear. We developed several hypotheses regarding work–family conflict issues in the military with our privileged sample.

2. Development of Hypotheses

H1. While many attempts have been made to measure work–family conflict (WFC) and family–work conflict (FWC), most of the scales or items, despite different wordings, appear to have had some merit in terms of reliability and validity. In this study, eight items were used to measure work–family conflict (WFC) and family–work conflict (FWC), with four items each. The hypothesis is that those eight items will factor into two separate sets of items, corresponding with the intended meanings, with loadings of 0.60 or higher on each subset and lower than 0.40 on those items not associated with a given scale. The percentage of residuals from the reproduced correlation matrix will be less than ten percent. The chi-square measure of goodness of fit associated with a maximum likelihood extraction and varimax rotation will not be significant (H1A). The results will be cross-validated against the non-Army officers in the sample (H1B).

H2. Each WFC and FWC scale, found with the factor analyses, will feature internal consistency reliability greater than 0.80 by Cronbach’s (1951) alpha (H2A). George and Mallery (2003) define alpha results as follows: unacceptable, <0.50; poor, between 0.50 and 0.60; questionable, 0.60 to 0.70; acceptable, 0.70 to 0.80; good, 0.80 to 0.90; and >excellent, >0.90. Tavakol and Dennick (2011) cite alphas from 0.70 to 0.95 as being acceptable. Generally, alphas of 0.80 or above are usually deemed adequate to good (Carmines and Zeller 1979, p. 51; Taber 2018), even though high alphas do not imply unidimensionality nor that all items in a scale are the best possible. The results will be cross-validated against the non-Army officers in the sample (H2B).

H3. Ng et al. (2025) state that “For researchers and practitioners who utilize self-report methodology, there are concerns that social desirability may bias people’s responses to a degree that threatens the reliability and validity of findings” (p. 394). Social desirability can occur with respect to individual self-evaluations or with respect to one’s evaluations of one’s marriage or parenting (Schumm 2015), usually measured by asking respondents if their evaluations are perfect (true vs. false) or at least very high, likely inflated in a socially desirable direction, to present one’s self, marriage, or parenting in a most positive light (Robinson and Anderson 1983). Preliminary work will be performed predicting marital satisfaction, marital process, parental satisfaction, marital stability, WFC, and FWC from marital social desirability, assessing heterogeneity of variance and potential linear, quadratic, and cubic trends. The expected pattern for heterogeneity of variance between marital social desirability and marital outcomes will be higher variation in marital outcomes at higher levels of marital social desirability. An expected pattern for trends will be stronger nonlinear trends for marital outcomes compared to non-marital outcomes as a function of marital social desirability. The expectation is that the strongest associations will be between marital satisfaction and marital social desirability, compared to that for the other variables, which may demonstrate discriminate validity of our measure of marital social desirability (i.e., marital social desirability influences marital outcome variables more than parental outcome variables).

H4. Based on recent research with the Kansas Marital Satisfaction Scale (Whisman et al. 2025), the expectation is that female officers would report lower marital satisfaction scores (H4A) and, by its association with related variables, lower scores on WFC (H4B), FWC (H4C), parental satisfaction (H4D), marital stability (H4E), and marital social desirability (H4F). In some studies, women have reported higher levels of WFC and FWC than men (Allen and Finkelstein 2014; Skomorovsky et al. 2025b), although Vuga and Juvan (2013) found the opposite in a study with the Slovenian military and some studies (Matthews et al. 2010; Carvalho and Chambel 2018; Borowski et al. 2021; Plummer 2023) and some meta-analytic results have not supported this hypothesis (Allen and French 2023, p. 445). However, Hobfoll et al. (2012) did find small sex differences (0.09 < r < 0.17).

Sex differences for FWC, WFC, marital satisfaction, parental satisfaction, marital stability, and marital social desirability will be tested against the null hypothesis, based on the variety of sex effects reported across several studies. These hypotheses could not be cross-validated with the non-Army sample because of limited data (only four female officers).

H5. Dual-military (full-time military employment of both spouses or partners) career respondents have reported higher levels of WFC or FWC (Ford et al. 2007; Park et al. 2023; Woodall et al. 2020). Although we did not have the responses of spouses, it could be argued that both spouses or partners in a dual-military family would also experience higher levels of WFC (H5A) or FWC (H5B). If so, similar patterns might be observed for outcomes of WFC or FWC to include lower marital satisfaction (H5C), lower parental satisfaction (H5D), higher marital instability (H5E), and lower levels of marital social desirability (H5F). Since only a few studies found significant results for dual-military families, the data will be tested against the null hypothesis. These hypotheses could not be cross-validated with the non-Army sample because of limited data (only two dual-military career families).

H6. In three different samples, Netemeyer et al. (1996) found that levels of WFC exceeded those of FWC, as well as did both Durand et al. (2003) and Plummer (2023). Therefore, we expected similar results with our sample (H6A). In terms of comparing marital and parental satisfaction, unpublished research by the authors had found that marital satisfaction was rated higher by parents than their parental satisfaction, probably because stresses associated with child-rearing are felt to be more related to parenting than to marriage (Kurdek 1996, p. 336). Therefore, we expected marital satisfaction to exceed parental satisfaction (H6B). Retention intention measures were available for the officer and for the officer’s perception of his or her spouse’s retention intentions. Given that the officers were of high rank and most likely “lifers”, we expected that their own retention intention perceptions would exceed those of their spouses, as the officers perceived them (H6C). We planned to cross-validate these hypotheses with non-Army officers in our sample (H6D, H6E, H6F). Reviewers raised concerns about never married officers—would they rate WFC and FWC despite having never been married? Therefore we also compared WFC and FWC for the 18 never married Army officers who indicated that they were not now and had never been married and who also answered the questions for both WFC and FWC.

H7A. Recent levels of deployment (here, total quarters deployed between 2001 and 2006) have been associated with higher levels of WFC and FWC (Skomorovsky 2014; Park et al. 2023). To the extent that recent high levels of military deployments lead to dissatisfaction with the amount of time spent with family, that dissatisfaction with the amount of time spent with family in recent years will be positively related to WFC and FWC (Skomorovsky 2014; Park et al. 2023). Past research has found respondent age to be negatively related to WFC and FWC (Allen and Finkelstein 2014; Matthews et al. 2010) but may be curvilinearly related to WFC or FWC (per Huffman et al. 2013). Military officers with more children or with younger children (including number of preschool children and number of adolescent children), who represent a drain on parental resources, are expected to report higher levels of WFC or FWC (Allen and Finkelstein 2014; Huffman et al. 2013, 2014; Park et al. 2023; Wong et al. 2022; Voydanoff 1988; Vuga Bersnak et al. 2023; Woodall et al. 2023). FWC may be more related to the number of children than WFC (Matthews et al. 2010; Michel et al. 2011). Thus, to test the construct validity of our measures of WFC and FWC, demographic variables were correlated with WFC and FWC against the null hypothesis, including age of the officer, years served as an officer, education, marital status, number of children, number of preschool children, number of adolescent children, age of youngest child, adopted status, officer’s military rank, and number of quarters deployed between 2001 and 2006.

H7B. To the extent that religion may act as a personal resource to buffer stress (Day et al. 2025), higher levels of religiosity may be associated with lower levels of WFC and FWC. Therefore, we expected that measures of intrinsic religiosity and religious practice would be correlated with reduced levels of WFC and FWC (H7B1). Since having higher levels of teamwork in marriage and higher levels of love and respect could be expected to reduce WFC and FWC, we also expected our measure of positive marital process to also be negatively related to WFC and FWC (H7B2).

H7C. Higher levels of work–family or family–work conflict will be related to lower levels of family satisfaction (including marital satisfaction, parental satisfaction, marital process, and higher levels of marital stability) (Kopelman et al. 1983; Ford et al. 2007; Allen et al. 2019, 2020, p. 547; Park et al. 2023; Tidwell and Lucier-Greer 2025; Woodall et al. 2020; Skomorovsky et al. 2015; Amstad et al. 2011; Fisher et al. 2016; Matthews et al. 2010). Some research has found that WFC is more strongly related to family satisfaction than is FWC (Wayne et al. 2004). Higher levels of work–family or family–work conflict have been related to lower levels of Army job satisfaction (including satisfaction with the Army as a way of life) (Kopelman et al. 1983; Allen et al. 2019, 2020; Skomorovsky et al. 2025a, 2025b; Pearce et al. 2025; Park et al. 2023; Tidwell and Lucier-Greer 2025; Amstad et al. 2011; Pluut and Andres 2019; Sachau et al. 2012; Fisher et al. 2016; Matthews et al. 2010). Some studies have found FWC more strongly related to job satisfaction than WFC (Wayne et al. 2004). Higher levels of work–family or family–work conflict have been related to lower levels of Army career intentions (Allen et al. 2019; Skomorovsky et al. 2025a; Sachau et al. 2012; Pearce et al. 2025; Park et al. 2023; Amstad et al. 2011). Recent levels of stress associated with the level of recent deployments have been associated with higher levels of WFC and FWC (Skomorovsky 2014; Park et al. 2023). To the extent that recent high levels of military deployments lead to dissatisfaction with the amount of time spent with family, that dissatisfaction with the amount of time spent with family in recent years will be positively related to WFC and FWC (Skomorovsky 2014; Park et al. 2023). WFC and FWC will likely be correlated with overall life satisfaction (Kopelman et al. 1983; Park et al. 2023; Amstad et al. 2011; Fisher et al. 2016). Wayne et al. (2004) reported that WFC may be more strongly related to life satisfaction than FWC. Correlations between WFC and FWC and other variables may differ between men and women (Allen and Finkelstein 2014).

Thus, to further test the construct validity of our measures of WFC and FWC, potential outcomes of WFC and FWC, including marital satisfaction, parental satisfaction, marital instability, retention intentions, satisfaction with the amount of time spent with the officer’s family since 2001, satisfaction with the Army as a way of life, satisfaction with the officer’s own life overall, the family’s fit with Army life, satisfaction with the Army’s support for the families of deployed soldiers, and the difficulty of maintaining the current frequency of deployments and a high quality of family life will be assessed against the null hypothesis, even though most of our variables should correlate negatively with WFC and FWC.

3. Methods

3.1. Measures

When our measures included the term “family”, we allowed our respondents to define that for themselves. Our projected measures included a four-item work-to-family conflict (WFC) scale, and a four-item family-to-work (FWC) scale, both modified versions of selected items in Netemeyer et al.’s (1996) WFC and FWC scales [whose scales have been used in other research with the US Army (Huffman et al. 2008, 2014) and more recently with Canadian military members (Skomorovsky et al. 2025b)], the three-item Kansas Marital Satisfaction Scale (Antunes et al. 2021; Cahill et al. 2010; Whisman et al. 2025), the three-item Kansas Parental Satisfaction Scale (Bunch et al. 2007; Edwards et al. 2018; Yopp et al. 2015; Power et al. 2009; Cahill et al. 2010; Wynne et al. 2016; Kyriazos et al. 2021; Carpenter and Donohue 2006), and a one-item measure of marital social desirability (developed for this study). The characteristics of these scales are reported in Tables 3 (for Army respondents) and 10 (for non-Army respondents). All four scales featured Cronbach (1951) alphas greater than 0.80, indicating high levels of internal consistency reliability. Borowski et al. (2021) by comparison found alphas of 0.82 to 0.85 for WFC for men and women, respectively, and of 0.76 to 0.80, respectively, for FWC. Cavaleiro et al. (2019) obtained an alpha of 0.89 using Carlson et al.’s (2000) 18-item measure of WFC. Jung and Kim (2023) obtained alphas of 0.87 for both WFC and FWC. Thompson et al. (1999) reported an alpha of 0.86 for four WFC items adopted from Thompson and Blau (1993). Sullivan et al. (2020) reported an alpha of 0.92 for WFC. Pluut and Andres (2019) found an alpha of 0.88 for Netermeyer et al.’s WFC scale and an alpha of 0.86 for their FWC scale adapted from Geurts et al. (2005). With Netemeyer et al.’s (1996) scales, Skomorovsky et al. (2015) obtained alphas of 0.87 for WFC and 0.89 for FWC, while Park et al. (2023) found 0.92 for WFC. Woodall et al. (2020) used a four-item modification of Netemeyer et al.’s WFC, obtaining an alpha of 0.90. Using three-item modifications of Netemeyer et al.’s (1996) WFC and FWC scales, Skomorovsky et al. (2025b) obtained alphas of 0.92 for WFC and 0.87 for FWC. Using Matthews et al.’s three-item scales, Wong et al. (2022) obtained alphas of 0.75 and 0.64 for WFC and FWC, respectively. Correlations between WFC and FWC were 0.343 (N = 257) and 0.224 (N = 44) in our analyses (Tables 8 and 9). Other researchers have found correlations of 0.32 between WFC for husbands and wives (Huffman et al. 2017) or 0.47 between WFC and FWC (Allen and Finkelstein 2014) as well as 0.31 between WFC and FWC for US Air Force service members (Sachau et al. 2012); 0.65 among Canadian Armed Forces personnel (Skomorovsky et al. 2015) and, later, 0.41 for men and 0.30 for women among Canadian Armed Forces personnel (Skomorovsky et al. 2025b); 0.30 (Fisher et al. 2016); between 0.28 and 0.61 for several WFC and FWC scales in a US-based online survey (Min et al. 2021); 0.28 for male and 0.45 for female U.S. veterans responding to a survey in 2014–2015 (Borowski et al. 2021); 0.57 for Chinese civilians (Jung and Kim 2023); 0.42 for Portuguese parents (Vieira et al. 2014); 0.55 for Israeli citizens (Hakhmigari and Diamant 2025); 0.33 for a sample of New York civilians (Frone et al. 1992); 0.46 in the MIDUS sample (Allen et al. 2023); 0.40 and 0.52 for two US samples of civilians (Huffman et al. 2013); 0.52 for a sample of Army National Guard personnel in the northwest US (Wong et al. 2022); 0.50 in a large US sample (Wayne et al. 2004); 0.43 from a Dutch Navy sample (Pluut and Andres 2019); and 0.52 for a US sample (Matthews et al. 2010). Mesmer-Magnus and Viswesvaran (2005) found a weighted mean observed correlation between WFC and FWC of 0.38 (noted by Amstad et al. 2011). Kopelman et al. (1983) found correlations of 0.14 and 0.30 between WFC and FWC. Huffman et al. (2014) used the Netemeyer et al. (1996) WFC/FWC scales three times, three months apart in 1999, with test–retest correlations of 0.56 (T1T2), 0.43 (T2T3), and 0.49 (T1T3).

A variety of other single items were used to measure sample demographics (

Table 1. Sample demographics of the Army respondents (n = 262).

Variable	Mean (SD)/Range (N)	Options	N	Percentage
Age	36.95 (3.44)/30–50 (260)
Years/Officer	12.72 (1.92)/3–20 (260)
Children	1.92 (1.27)/0–5 (262)
Number of Preschool Children	0.67 (0.79)/0–3 (255)
Number of Adolescent Children	0.31 (0.63)/0–4 (304)
Quarter Years	5.53 (3.46)/0–17 (262)
Deployed since 2001
Age of youngest Child	6.84 (5.24)/1–20 (218)
Sex		Male	239	91.60%
		Female	22	8.40%
Marital Status		Not Married	15	6.20%
		First Marriage	202	83.10%
		Remarried after Divorce	25	10.30%
		Remarried after Spouse Died	1	0.40%
Current Rank		Captain (O-3)	1	0.40%
		Major (O-4)	256	98.80%
		Lieutenant Colonel (O-5)	2	0.80%
Education		Bachelor’s Degree	126	48.50%
		Master’s Degree	119	45.80%
		Doctoral/Professional Degree	15	5.80%
Dual-Career Status		Spouse in the Military	24	9.20%
Parental Status		Had Child(ren)	219	83.60%
Race/Ethnicity		White	210	80.3%
		Black	23	8.9%
		Native Americans or Alaskans	3	1.2%
		Pacific Islanders	3	1.2%
		Other	7	2.7%
		Hispanic	9	3.5%
Current Instability		Not stressed	158	69.3%
		Very stressed	42	18.4%
		Marriage in Trouble	6	2.6%
		Thinking about Divorce	22	9.6%
		Adopted as a Child Yes	5	2.0%
Number of Times Married		None	19	7.3%
		One	213	81.3%
		Two	28	10.7%
		Three or more	2	0.8%

Note: There were only 7 cases where the respondent was not married and had no children.

) that might be associated with our two measures of conflict between work and family (Table 8). In addition to those single-item demographic measures, we used three scales as potential predictors of WFC and FWC, including intrinsic religiosity, religious practice, and internal marital processes.

Intrinsic religiosity was measured by the sum of three items that asked for five levels of agreement/disagreement with “I have often had a strong sense of God’s presence”, “My religion is important to me because it answers many questions about life’s meaning”, and “My relationship with God is a vitally important part of my life.” This scale’s characteristics are presented in Table 3, including a Cronbach’s (1951) alpha of 0.905. A religious practice scale was formed by the sum of ten items, including one set asking to what extent your religion or religious beliefs helped you with finding “a sense of meaning and purpose”, “a feeling of hope about the future”, “a sense of peace and comfort”, and “help in solving my problems” and a second set concerning past help from religion or religious beliefs with respect to how you “sought God’s love and care”, “trusted that God wouldn’t let anything terrible happen to my spouse or me”, “used my faith to help me cope with deployments”, “asked God to make things turn out ok”, or engaged in “prayer” and “read scripture”. This scale’s characteristics are presented in Table 3, including an alpha of 0.921. These two scales were highly correlated, r = 0.738 (p < 0.001; N = 222). A single item involved satisfaction with one’s own spiritual life, as defined by each respondent for themselves; that item was correlated 0.144 with the Intrinsic Religiosity Scale (p = 0.021; N = 257) and 0.168 with the religious practice scale (p = 0.012; N = 221).

Internal marital process was measured by the new Kansas Marital Process Scale, which included three items about the respondent’s level of satisfaction (in five levels from very dissatisfied to very satisfied) with the “ability to work with your spouse as a team”, “level of respect from your spouse”, and “love and care coming from your spouse”. This scale’s characteristics are presented in Table 3, including an alpha of 0.919. Most of our respondents reported both low love and respect (N = 16) or both high (N = 194) with very few reporting low love but high respect (N = 5) or low respect but high love (N = 3), contrary to popular ideas that the percentages of respondents might be more equal across those four types. This scale was correlated 0.887 with the Kansas Marital Satisfaction Scale (p < 0.001, N = 215), but the partial correlation was 0.800 after controlling for our single-item measure of marital social desirability (p < 0.001; N = 214). A single item was used as an index of marital instability, with four responses about the situation with one’s current marriage with a range from not stressed, very stressed, and in trouble to thinking about divorce. Correlations between this index of marital instability and the Kansas Marital Process Scale and the Kansas Marital Satisfaction Scale were correlated, respectively, −0.755 (p < 0.001; N = 215) and −0.674 (p < 0.001; N = 217), reduced to partial correlations of −0.585 (p < 0.001; N = 212) and −0.501 (p < 0.001; N = 212) after controlling for our single item for marital social desirability. That single item’s distribution (N = 260) from least to most was 11.9%, 17.7%, 18.8%, 39.6%, and 11.9%. These substantial differences in correlations may indicate some degree of discriminant validity between the two scales.

Parental satisfaction was measured in two ways, first by the Kansas Parental Satisfaction Scale, which includes three items concerning the respondent’s satisfaction (here, in five levels) with “yourself as a parent”, “your relationship with your child(ren)”, and “your child(ren)’s behavior towards you” (alpha = 0.813). Two additional items were added to those first three, for your “ability to connect well with your child(ren)” and “how well your child(ren) respond to you” for a longer version of the KPSS (increasing the scale’s alpha to 0.898). The shorter and longer versions were correlated, respectively, with the marital instability index (−0.183, p = 0.008, and N = 207; −0.151, p = 0.031, and N = 205), the Kansas Marital Process Scale (0.291, p < 0.001, and N = 209; 0.228, p < 0.001, and N = 207), and the Kansas Marital Satisfaction Scale (0.317, p < 0.001, and N = 198; 0.261, p < 0.001, and N = 196). The two versions of the KPSS were highly correlated, r = 0.966 (p < 0.001; N = 214). Non-parents usually did not complete the parental satisfaction items because of their irrelevance and that data was treated as missing.

Variables expected to be possible correlates or outcomes of WFC and FWC included several single items about satisfaction with “the amount of time you have had with your family since 2001”, “the Army as a way of life”, “your own life, overall”, “your family’s fit with Army life”, and “in general, your satisfaction with the Army’s support for the families of deployed soldiers”. Responses to each of these items ranged in five levels from very dissatisfied to very satisfied. Another item asked “How difficult is it for officers such as yourself to maintain both the current frequency of deployments and your quality of family life?” with four levels of response, including not difficult, somewhat difficult, very difficult, and extremely difficult. This item was designed as a measure of work–family balance.

Current marital satisfaction was measured by the three-item Kansas Marital Satisfaction Scale, featuring an alpha of 0.949, whose items pertain to, in this study, five levels of satisfaction from very dissatisfied to very satisfied, with respect to one’s marriage, one’s relationship with one’s spouse, and with one’s spouse as a partner. The scale most recently was used with over 7000 persons from 33 nations across five continents (Whisman et al. 2025).

Retention intentions were reported by single items with four response categories, leave before end of current tour, leave at end of current tour, leave some time after end of current tour, and stay until retirement (coded 1–4), using the officer’s opinion (0.0%, 0.8%, 5.4%, and 93.8%; N = 240) and a report by the officer about his or her perception of their spouse’s opinion (7.1%, 12.3%, 3.6%, and 77.1%; N = 253). The two items were only correlated 0.194 (p = 0.003; N = 231), which was of interest because of presumed common method variance since both answers came from the mind of the same person (the officer respondent) and were significantly different (Table 8) with a very large effect size (Hedges’ g = 0.955). The alpha was only 0.195 for the two items, insufficient for creating a scale from the two items.

3.2. Survey Implementation

Our surveys were emailed to all officer students enrolled in a full-time active duty basis in the Army’s Intermediate Level education program (formerly Command and General Staff College) at Fort Leavenworth, Kansas, including classes 7-01 and 8-01, with an introductory letter discussing the purpose of the survey and providing for informed consent, using procedures as approved by the commanding general of Fort Leavenworth and the Office of the Chief of Chaplains, U.S. Army, through the head chaplain at Fort Leavenworth, Colonel Glen Bloomstrom, and the Institutional Review Board of Kansas State University, which approved the study in May 2007 (email from M. DeHart, IRB administrator, Office of Research Integrity, Compliance, and Security, Kansas State University, 28 May 2025). Other researchers have used online surveys, entirely online or partly with mail surveys, for contacting potential respondents in Army garrison locations (Hobfoll et al. 2012; Huffman et al. 2017; Park et al. 2023). Despite our requests, the Army school and installation (to reduce interference with the educational experience and maintain consistency with the treatment of other surveys) did not allow us to meet in person in their classrooms with the officers to explain the purposes of the survey, to send mailed reminders (or an alternative options for handed-out initial or mailed follow-up surveys), or to provide any incentives for survey completion, methods which, for many decades, have been recommended and proven for increasing response rates in social science surveys (Dillman 1972, 1978, 1991, 2000, 2011, 2020, 2022, 2024).

3.3. Sample

For the two ILE classes we surveyed, the population was all enrolled ILE students at the time of the surveys. The first sample was from ILE Class 7-01, with a 15% response rate (N = 119 overall, 97 Army); the second sample was from ILE Class 8-01, with a 25% response rate (N = 188 overall, 165 Army); and the combined response rate was 20% (N = 307, Army = 262). The lower response rate for the first sample was probably related to the start of data collection (end of May 2007), which occurred only a week before that class’s graduation; some of the officers had already departed for new assignments, many for combat zones, with many other priorities than completing a 13-page email survey from the Chaplain’s office. Because the survey was designed primarily to evaluate Army chaplain services as provided to the ILE students, it is likely that those with more familiarity with Army chaplain services/support for them were more likely to respond, which could bias the research results. Army documents indicate that some officers were being pulled early from ILE to go overseas while others had their enrollments cancelled before they could begin ILE (Hollis 2008). Eliminating potential respondents from the pool of ILE attendees may have biased the results as well as reducing the size of the available population. At the time of the surveys, we obtained the yearbook (The Bell) for ILE Class 07-01 and observed that only 142 Army officers took out time for individual or family pictures in that yearbook, either for security reasons or because they did not want to take time away from their studies or from their families, factors that could also have likewise reduced our surveys’ response rates. In other words, our response rates, though lower than desired, were comparable to those for the annual ILE yearbook for Class 07-1.

To give future researchers with email military surveys a basis of comparison, for the Army respondents, on the first day of response, 35 responses were obtained, with 18 on the second day, and then 4, 5, 4, 2, 1, 0, 0, and 1 for a total of 70 so far. On the 11th day (after an email reminder to respond), four responses occurred, then eight, then three, and then two, three, one, one, zero, and one. After a further email reminder, on the 27th day there were two more responses with another response on day 54 and one on day 82, which was all that were received from Class 07-01, and no further responses were permitted for that class, for a total of 97. The next day, responses from Class 08-01 began, with 35, then each following day, 14, 12, 6, 0, and 2, with no more until day 23 (after an email reminder) with 68 responses, then 10, 1, 1, 1, 0, 1, 0, 10, 3, 0, 0, 0, and 1, for a total of 165. After that the survey for Class 08-01 was concluded.

For Class 07-01, responses from non-Army services were received as follows for day 1 (8), 2 (6), 4 (1), 5 (1), 7 (1), 11 (1), 12 (3), and 15 (1), for a total of 22. For Class 08-01, responses from non-Army services were received as follows for day 1 (4), day 3 (4), and, after a reminder email, on day 22 (11), day 23 (2), ending at day 30 (2), for a total of 23. In addition to the 262 Army members, the non-Army sample included members of the U.S. Army Reserve (16), Army National Guard (14), International Military (9), U.S. Air Force (2), and the U.S. Navy (4) for a final total of 307 (262 + 45) participants.

Table 1 presents the demographic characteristics of the 262 Army officers who responded to the two surveys. The sample of Army officers included 239 men (85.4% of whom were a parent) and 22 women (63.6% of whom were a parent; Fisher’s Exact Test, p < 0.02), with n = 1 missing data on sex. Of the men and women, respectively, most were married for the first time (190/11), with some never married (14/5), married but divorced (10/1), married/divorced/remarried (20/4), married/spouse died/remarried (1/0), married/divorced/remarried/divorced/remarried (4/0), and married/divorced/remarried/divorced/remarried (0/1). Measures for cohabitation or same-sex partnerships were not used. In terms of race, 80.2% of the officers were White, with others being Black (8.9%), Native American (0.8%), Asian (2.3%), Native Alaskan (0.4%), Pacific Islander (1.2%), Hispanic (3.5%), or Other (2.7%); differences between men and women were not significant. Total numbers of children reported by officers were summed for ages < 1 to 5 years (0.58/0.77; 0 to 3), 6 to 12 years (0.69/0.80; 0 to 3), 13 to 17 years (0.30/0.62; 0 to 4), 18 to 20+ years (0.18/0.49; 0 to 3), and total of all children (1.76/1.20; 0 to 5). Some officers did not report having had any children (43/261; 16.5%). The number of preschool children was positively correlated with parental satisfaction (0.15; p < 0.05) while the number of teens was correlated negatively (−0.19; p < 0.01).

3.4. Analyses

H1. The eight WFC/FWC items will be factor analyzed for Army officers (H1A), using a maximum likelihood extraction with varimax rotation, further checked with factor analysis with a principal axis extraction; the same analyses (H1B) will be repeated for the non-Army officers for cross-validation.

H2. Assuming that the eight WFC/FWC items factor into the expected two scales, the internal consistency reliability of the two scales will be assessed with Cronbach’s (1951) alpha, for Army officers (H2A) and non-Army officers (H2B).

H3. Heterogeneity of variance (H3A) across the levels of marital social desirability predicting marital and parental outcome scales will be assessed as part of several one-way analyses of variance. Linear and nonlinear trends (H3B) for the marital and parental outcome scales will also be assessed as part of one-way analyses of variance.

H4. Sex differences in WFC, FWC, marital process, satisfaction scales and instability, and parental satisfaction scales will be assessed with independent sample t-tests for Army officers; there were too few women (N = 4) among the non-Army officers to permit accurate testing of H4.

H5. Differences in WFC, FWC, marital process, satisfaction scales, and instability and parental satisfaction scales will be assessed for dual-career military family status with independent sample t-tests for Army officers; there were too few dual-career spouses (N = 2) among the non-Army officers to permit accurate testing of H4.

H6. Differences in WFC versus FWC, marital vs. parental satisfaction, and soldier vs. spouse retention intentions will be assessed with pair-sample t-tests for both Army officers (H6A) and non-Army officers (H6B). These comparisons will be performed separately for men and women in the Army sample; there were too few women officers (N = 4) in the non-Army sample to allow for such testing. The same comparison will also be tested for 18 never married officers.

H7A. Demographic variables, including age, years of commissioned service, education, married vs. not married, number of times married, number of children, number of preschool children, number of adolescent children, age of the youngest child, rank, dual-career military family status, and number of quarter years deployed between 2001 and 2006, will be associated with WFC and FWC using Pearson zero-order correlations with two-tailed testing.

H7B. Independent variable predictors, including marital social desirability, intrinsic religiosity, religious practice, satisfaction with one’s own spiritual life, and our marital process scale, of WFC and FWC will be associated with WFC and FWC using Pearson zero-order correlations with two-tailed testing.

H7C. Outcome variables associated with WFC and FWC, including officer retention intentions, spouse retention intentions, satisfaction with the amount of time with family since 2001, satisfaction with the Army as a way of life, satisfaction with your own life overall, satisfaction with your family’s fit with Army life, satisfaction with the Army’s support for the families of deployed soldiers, perceived difficulty of maintaining both the current frequency of deployments and quality of family life, current marital satisfaction, current parental satisfaction, and our marital instability index, will be tested using Pearson zero-order correlations with two-tailed testing.

H7D. Interaction effects between sex and each of the variables used in H7A, H7B, and H7C predicting both WFC and FWC will be tested using univariate analysis of variance, using sex as a fixed variable and the other variables as random. Our expectation was that FWC might matter more for female officers than male officers.

All analyses will be calculated using SPSS version 29, as used by other scholars for computation of statistical analyses (Bos et al. 2025).

4. Results

H1.

Table 2. Rotated factor loadings from three types of factor analysis for eight items related to work-to-family and family-to-work conflicts (N = 257).

Subscale Items	Rotated Factor Loadings a,b
Work-Family Conflict (WFC) Items:	1	2
1. Army demands interfere with my home and family life.	0.132/0.137	0.689/0.682
2. I am often preoccupied with work issues while at home.	0.099/0.092	0.607/0.608
3. Army-related strain often interferes with family responsibilities.	0.286/.284	0.725/0.726
4. Things I want to do with family don’t get done because of Army demands.	0.110/0.111	0.829/0.831
Family-Work Conflict (FWC) Items:
1. I am often preoccupied with personal/family life while at work.	0.724/0.712	0.210/0.219
2. Family-related demands interfere with my work related activities.	0.819/0.829	0.083/0.075
3. Family related strain often interferes with my military responsibilities.	0.801/0.804	0.210/0.217
4. My Army duties sometimes are not done well due to family demands.	0.580/0.576	0.111/0.105
Number of Cases	257
Eigenvalues	3.55	1.75
Determinant	0.040
KMO Sampling Adequacy	0.818
Bartlett Test of Sphericity, chi-square (df = 28)	810.57, p < 0.001
Maximum Likelihood chi-square test (df = 13)	17.37, p = 0.183
Percentage of Residuals > 0.05 in reproduced correlation matrix	3.0

^a Extraction Methods: Maximum likelihood and principal axis (common) factoring; varimax rotation, listwise deletion of missing data; ^b using mean substitution (n = 5) and maximum likelihood extraction, the determinant = 0.041, KMO = 0.816, Bartlett test (df = 28) = 811.69, p < 0.001, goodness of fit, chi-square (df = 13) = 17.98, p = 0.158, residuals (3.0%), eigenvalues 3.52 and 1.76. Results for principal components analysis were similar except residuals were greater (42.0%). Note: For WFC, items 1, 3, and 4 represent modifications of Netemeyer et al.’s (1996) items 1, 2, and 3; for FWC, items 2, 3, and 4 represent modifications of Netemeyer et al.’s (1996) items 1, 5, and 3. Items 2 (WFC) and 1 (FWC) are keyed to the word “preoccupied”, similar to items 3 and 12 from Geurts et al. (2005) or an FWC item from Huffman et al. (2008, p. 523), as well as two items in Carlson et al.’s (2000, p. 273) scale. Item 2 (WFC) is similar to item 20 (Kopelman et al. 1983, p. 204, Figure 2).

presents the results of two approaches to factor analysis of the eight WFC/FWC items for the 257 Army officers in our sample. All of the factor loadings exceeded 0.60 except for item #4, FWC, which loaded at 0.58 and 0.576; for the secondary loadings, none exceeded 0.286. Two factors were extracted, with eigenvalues exceeding 1.0. KMO sampling adequacy (Lorenzo-Seva and Ferrando 2021) was >0.80. The maximum likelihood test was not significant, indicating good fit of the factor solution to the data. The percentage of residuals in the reproduced correlation matrix was only 3.0%. Using mean substitution for five cases (N = 262) yielded similar results.

Table 3. Rotated factor loadings from three types of factor analysis for eight items related to work-to-family and family-to-work conflicts (N = 45).

Subscale Items	Rotated Factor Loadings a,b
Work-Family Conflict (WFC) Items:	1	2
1. Army demands interfere with my home and family life.	0.190	0.638
2. I am often preoccupied with work issues while at home.	0.004	0.682
3. Army-related strain often interferes with family responsibilities.	0.152	0.880
4. Things I want to do with family don’t get done because of Army demands.	−0.044	0.816
Family-Work Conflict (FWC) Items:
1. I am often preoccupied with personal/family life while at work.	0.613	0.016
2. Family-related demands interfere with my work related activities.	0.880	0.106
3. Family related strain often interferes with my military responsibilities.	0.898	0.156
4. My Army duties sometimes are not done well due to family demands.	0.690	0.036
Number of Cases	45
Eigenvalues	3.28	2.28
Determinant	0.016
KMO Sampling Adequacy	0.752
Bartlett Test of Sphericity, chi-square (df = 28)	163.16, p < 0.001
Maximum Likelihood chi-square test (df = 13)	11.37, p = 0.580
Percentage of Residuals > 0.05 in reproduced correlation matrix	28.0

^a Extraction Methods: Maximum likelihood; varimax rotation, mean substitution used for missing data; ^b ideally factor analysis requires either 200 or more cases or at least a 5:1 ratio of cases to variables used (10:1 is ideal; Thompson et al. 1999, p. 399). Here that ratio was 5.63.

presents the same factor analysis applied to the 45 non-Army officer cases, using mean substitution. The results are similar except that the percentage of residuals in the reproduced correlation matrix is much higher (28.0%). Together, both sets of factor analysis support the expected dimensionality of the eight items into two factors representing WFC and FWC, supporting hypothesis 1.

H2.

Table 4. Characteristics of measurement scales and single item for social desirability for Army officers (N = 257).

Scale or Item Name	N	Mean (SD)	Median	Range	Cronbach’s Alpha
Work-Family Conflict	260	14.74 (3.15)	15.00	5–20	0.814
Family-Work Conflict	259	9.67 (2.97)	8.00	4–17	0.828
Kansas Marital Satisfaction	217	13.03 (3.18)	15.00	3–15	0.949
Kansas Marital Process Scale	231	13.01 (2.91)	14.00	3–15	0.919
Kansas Parental Satisfaction	216	12.57 (2.21)	13.00	6–15	0.813
Kansas Parental Satisfaction (five items)	214	20.99 (3.67)	21.00	8–25	0.898
Intrinsic Religiosity (3 items)	260	12.60 (2.71)	13.00	3–15	0.905
Practical Religiosity (10 items)	224	28.07 (7.81)	29.00	10–40	0.921
Marital Social Desirability	260	3.22 (1.22)	4.00	1–5	NA

presents the Cronbach (1951) alpha estimates of internal consistency reliability for the WFC and FWC scales as well as several other scales used in this study. All of the scales yielded alphas above 0.80, indicating at least “good” reliability. In

Table 5. Characteristics of measurement scales and single item for social desirability for non-Army officers (N ≤ 45).

Scale or Item Name	N	Mean (SD)	Median	Range	Cronbach’s Alpha
Work-Family Conflict	44	15.07 (2.94)	16.00	8–20	0.834
Family-Work Conflict	44	10.66 (3.18)	10.00	4–17	0.845
Kansas Marital Satisfaction	35	12.71 (3.67)	15.00	3–15	0.986
Kansas Parental Satisfaction	38	12.81 (2.31)	13.50	5–15	0.893
Kansas Parental Satisfaction (five items)	38	21.32 (3.83)	22.00	9–25	0.927
Intrinsic Religiosity	44	12.61 (3.38)	14.00	3–14	0.967
Practical Religiosity	28	28.21 (8.71)	29.00	10–39	0.949
Marital Social Desirability	44	3.50 (1.27)	4.00	1–5	NA

, similar results were obtained for the same scales, using the smaller sample of non-Army officers; all of the scales yielded alphas >0.80. Hypothesis 2 was supported, not only for the WFC and FWC scales but for all of the multi-item scales considered in this study in terms of satisfactory levels of internal consistency reliability.

H3. WFC, FWC, marital satisfaction, parental satisfaction (three items), parental satisfaction (five items), and our single item for marital instability (five levels) were predicted from marital social desirability, using one-way analysis of variance. Linear trends were significant for all of the analyses. However, weighted quadratic trends were only significant for WFC (p = 0.025), marital satisfaction (p < 0.001), and marital instability (p < 0.001). In terms of effect sizes (eta-squared fixed effects/omega-squared; random effects), there was considerable variation: WFC (0.155/040), FWC (0.068/0.014), marital satisfaction (0.595/0.263), parental satisfaction (three items) (0.107/0.025), parental satisfaction (five items) (0.083/0.018), and marital instability (0.475/0.179). The largest effect size for marital social desirability was for marital satisfaction, followed by marital instability, suggesting that measures of social desirability need to fit the variables most related to them (Schumm 2015, p. 40). As marital social desirability answers increased from strongly disagree to agree, the means/SDs for marital satisfaction increased as an inverse quadratic (5.74/3.53, N = 19; 10.76/3.35, N = 45; 13.30/1.92, N = 43; 14.44/1.19, N = 107; and 14.79/1.14, N = 38). SPSS provides four different tests for heterogeneity of variance, based on the mean, median, median with adjusted degrees of freedom, and trimmed mean. Because of an increased chance of one of those four tests being significant due to random error, we adjusted the required level of significance to 0.01. Significant heterogeneity of variance was observed only for marital satisfaction [F(4, 247) = 32.21, p < 0.001) and marital instability [F(4, 260, p < 0.001) and for those two variables, for all four tests of heterogeneity, as was expected for marital satisfaction (and in hindsight for its “opposite” marital instability). Thus, an expected heterogeneity of variance was found for marital satisfaction as predicted. Our results indicate that our single-item measure of marital social desirability has unique effects on our scales and differentially predicts marital satisfaction in terms of effect size, heterogeneity, and nonlinear trends.

Furthermore, we investigated different ways of controlling for marital social desirability. One way would be to partial MSD from the correlations among WFC, FWC, marital satisfaction, and marital process; the KMSS/WFC and KMSS/FWC correlations were reduced from −0.293 (p < 0.001) and −0.294 (p < 0.001) to −0.002 and −0.128 (p < 0.10) while the MP/WFC and MP/FWC were reduced from −0.274 (p < 0.001) and −0.341 (p < 0.001) to −0.128 (p < 0.10) and −0.226 (p < 0.001). It appears that controlling for MSD may overcontrol, especially for WFC. Another way to control for MSD would be to delete all cases in which MSD = 5, removing the cases most subject to the highest levels of MSD. The previous correlations, respectively, were now −0.253, −0.271, −0.215, and −0.317 (all p < 0.001), with FWC correlations decreased the least. Being more strict and removing all cases in which MSD > 3 led to correlations of −0.179 (p < 0.10), −0.245 (p < 0.05), −0.135, and −0.315 (p < 0.01). While the WFC correlations became non-significant, the FWC correlations remained substantial and significant. The relationships between MSD and marital process and marital satisfaction were muted, compared to very large homogeneities of variance and quadratic trends. Most of the homogeneity of variance tests were barely p < 0.05, some were not significant, and the quadratic trends barely reached p < 0.05, indicating that much of the peculiar relationships between MSD and marital variables suggestive of the role of MSD were minimized. In summary, our analyses confirmed the value of measuring and controlling for marital social desirability and presumably for other forms as well (e.g., individual social desirability, parental social desirability, etc.), as discussed elsewhere (Schumm 2015, p. 40).

H4.

Table 6. Measurement scales as a function of sex differences.

Scale	Sex	Mean (SD)	N	t	df	p a	d b
Work-Family Conflict	Male	14.83 (3.16)	237	1.44	257	0.151	0.32
	Female	13.82 (2.99)	22
Family-Work Conflict	Male	9.75 (2.97)	237	1.68	256	0.095	0.381
	Female	8.62 (2.92)	21
Kansas Marital Satisfaction	Male	13.00 (3.25)	203	−0.68	214	0.498	−0.194
	Female	13.62 (2.10)	13
Kansas Parental Satisfaction (3 items)	Male	12.47 (2.23)	201	−2.41	213	0.017 c	−0.663
	Female	13.93 (1.27)	14
Kansas Parental Satisfaction (5 items)	Male	20.81 (3.71)	199	−2.53	211	0.012	−0.697
Current instability levels	Male	1.55 (0.97)	211	3.04	30.51	0.005	0.943
	Female	1.19 (0.40)
Marital Social Desirability	Male	3.23 (1.22)	238	0.318	257	0.751	0.072
	Female	3.14 (1.24)	21

Note: Heterogeneity of variances was observed for instability levels (p = 0.003); hence, equal variances were not assumed for the t-test. Results for intrinsic religiosity, practical religiosity, officer career plans, and spouse career plans were not significant. ^a Two-tailed tests; ^b Hedges’s correction for effect size; ^c for the Kansas Parental Satisfaction Scale, the assumption of homogeneity of variance was violated, Levene’s test, F(200, 13) = 4.08 (p = 0.045), with t(19.16) = −3.89, p < 0.001.

presents results for sex differences on selected scales and items. Male officers reported higher levels of WFC (g = 0.32) and FWC (g = 0.38) compared to female officers, but those differences were not significant. Mothers reported significantly higher parental satisfaction than fathers (0.66 < g < 0.70), but though wives reported higher marital satisfaction than husbands (g = 0.19), the difference was not significant. Male officers reported significantly higher current instability levels than female officers (g = 0.94). Other variables tested yielded non-significant results (Table 6). There were too few women (N = 4) among the non-Army officers to permit cross-validation for sex differences.

H5.

Table 7. Measurement scales as a function of spouse as civilian or military.

Scale	Spouse	Mean (SD)	n	t	df	p a	d b
Work-Family Conflict	Military	13.24 (3.37)	21	2.29	225	0.023	0.524
	Civilian	14.89 (3.13)	206
Family-Work Conflict	Military	9.48 (2.93)	21	0.13	224	0.899	0.029
	Civilian	9.56 (2.91)	207
KS Marital Satisfaction	Military	13.95 (1.84)	19	1.90	29.3 c	0.068	0.297
	Civilian	13.05 (3.13)	196
Marital Process	Military	14.00 (1.70)	21	2.44	33.37 d	0.020	0.297
	Civilian	12.97 (2.91)
KS Parental Satisfaction (3 items)	Military	14.24 (0.90)	17	6.39	35.7 e	<0.001	0.822
	Civilian	12.52 (2.15)	190
KS Parental Satisfaction (5 items)	Military	23.59 (1.37)	17	6.27	40.9 e	<0.001	0.760
	Civilian	20.93 (3.60)	188
Current instability levels	Military	1.24 (0.44)	21	2.72	44.2 f	0.009	0.337
	Civilian	1.56 (0.97)	207
Number of Children	Military	1.38 (0.92)	21	3.17	226	0.002	0.723
	Civilian	2.22 (1.17)	207
Age of Youngest Child	Military	2.50 (1.95)	18	7.60	43.87 g	<0.001	0.912
	Civilian	6.96 (5.06)	191
Number of Preschool Children	Military	1.17 (0.62)	18	3.13	22.75 g	0.005	0.622
	Civilian	0.67 (0.80)	191
Number of Adolescent Children	Military	0.05 (0.22)	21	4.68	70.2 g	<0.001	0.485
	Civilian	0.35 (0.67)	206
Age of Officer	Military	35.67 (2.56)	21	1.73	224	0.086	0.36
	Civilian	37.02 (3.51)

^a Two-tailed; ^b Hedges’s correction for effect size; ^c for the Kansas Marital Satisfaction Scale, the assumption of homogeneity of variance was violated, Levene’s test, F(195, 18) = 4.76 (p = 0.030). ^d For the Kansas Marital Process Scale, the assumption of homogeneity of variance was violated, Levene’s test, F(206, 20) = 7.63 (p = 0.006). ^e For the Kansas Parental Satisfaction Scale, the assumption of homogeneity of variance was violated, Levene’s test, F(189, 16) = 7.98 (p = 0.005). For the five-item version, Levene’s test, F(187, 16) = 9.44 (p = 002). ^f For current instability levels, the assumption of homogeneity of variance was violated, Levene’s test, F(206, 20) = 9.06 (p = 0.003). ^g The Levene’s tests for homogeneity of variance for age of youngest child, number of preschool children, and number of adolescent children, respectively, were F(190, 17) = 13.73 (p < 0.001), F(190, 17) = 5.50 (p = 0.020), and F(205, 20) = 22.39 (p < 0.001). Note: The results for marital social desirability were not significant (p = 0.968; g = 0.01).

presents the differences in variables as a function of dual-military spouses. Officers with military spouses reported lower WFC (p = 0.023; g = 0.524), nearly higher marital satisfaction (p = 0.068; g = 0.297), higher marital process (p = 0.020; g = 0.364), higher parental satisfaction (p < 0.001; 0.760 < g < 0.822), and lower instability levels (p = 0.009; g = 0.337). Although differences in age were not significant (p = 0.086; g = 0.394), officers with military spouses reported fewer children in total (p = 0.002; g = 0.723), fewer adolescent children (p < 0.001; g = 0.485), more preschool children (p = 0.005; g = 0.622), and a lower age of the youngest child in their family (p < 0.001; g = 0.912). There were too few dual-career spouses (N = 2) among the non-Army officers to permit accurate testing of H4. In general, Army officers with military wives reported more favorable family processes than those officers with civilian or unemployed spouses.

H6.

Table 8. Paired scale differences.

Variable	Mean (SD)	N	r	t	df	p	g
Work–Family Conflict vs. Family–Work Conflict
Work–Family Conflict	14.75 (3.13)	257	0.343	23.23	256	<0.001	1.62
Family–Work Conflict	9.68 (2.97)
Men	14.86 (3.14)	235	0.349	22.38	234	<0.001	1.46
	9.77 (2.96)
Women	13.67 (2.97)	21	0.180	6.13	20	<0.001	1.29
	1.62 (2.92)
Marital vs. Personal Satisfaction
Marital Satisfaction	13.17 (3.03)	198	0.317	2.24	197	<0.03	0.158
Personal Satisfaction	12.68 (2.15)
Men	13.11 (3.10)	186	0.309	2.21	185	<0.03	0.161
	12.60 (2.16)
Women	14.27 (1.19)	11	0.297	0.58	10	0.574	0.162
	14.00 (1.41)
Retention Intention Difference
Office Retention	3.92 (0.30)	232	0.190	7.11	231	<0.001	1.000
Spouse Retention	3.46 (1.01)
Men	3.93 (0.30)	214	0.199	6.86	213	<0.001	0.467
	3.47 (0.99)
Women	3.94 (0.25)	16	0.000	1.00	15	0.333	0.237
	4.00 (0.00)

Note: Two-tailed significance tests; Hedges’s correction used for effect size estimates. The correlations between WFC and FWC fit with spillover theory, which predicts overlap between those concepts.

presents paired-sample mean differences in WFC versus FWC, marital vs. parental satisfaction, and soldier vs. spouse retention intentions for Army officers while

Table 9. Paired scale differences.

Variable	Mean (SD)	N	r	t	df	p	d
Work–Family Conflict vs. Family–Work Conflict
Work–Family Conflict	15.07 (2.94)	44	0.224	7.66	43	<0.001	1.14
Family–Work Conflict	10.66 (3.18)
Marital vs. Parental Satisfaction
Marital Satisfaction	12.94 (3.33)	33	0.098	0.31	32	0.758	0.053
Parental Satisfaction	12.73 (2.43)
Retention Intentions	3.92 (0.28)	36	0.271	2.50	35	0.017	0.955
	3.53 (0.97)

Note: Two-tailed significance tests; Hedges’s correction used for effect size estimates.

presents the same results with non-Army officers. For the Army officers, WFC was significantly greater than FWC (p < 0.001; g = 1.62); marital satisfaction was higher than parental satisfaction (p < 0.03; g = 0.158), and officers’ views of their retention intentions was greater than that which they attributed to their spouses (p < 0.001; g = 1.00). For the 18 never married Army officers, WFC (14.22/2.76) was significantly greater than for FWC (10.50/2.60); r = 0.493 (p < 0.05) and t(17) = 5.86 (p < 0.001; g = 1.32). For the non-Army officers (Table 9), the patterns were similar except that the results were not significant for marital vs. parental satisfaction, with a much smaller effect size (g = 0.053 vs. 0.158). In terms of sex differences, the larger magnitude of work–family conflict compared to family–work conflict remained significant (p < 0.001) for both male and female officers; differences between marital and parental satisfaction remained small, with slightly higher scores for marital satisfaction, and there were significant differences only for male officers. Differences for retention intentions remained significant for male officers but not for female officers. There were too few women officers (N = 4) in the non-Army sample to allow for such testing. The moderate correlations between WFC and FWC, shown in Table 8, would be expected from spillover theory, which predicts empirical overlap between those two concepts. In general, WFC presented more problems than did FWC; officers viewed their retention more favorably than they thought their spouses did; and marital satisfaction tended to be greater than parental satisfaction.

To control for marital social desirability (MSD), we re-ran the same analyses, using only cases in the three lower levels of MSD, which should reduce MSD in this sample but which also reduced our sample size and thus statistical power. For marital vs. parental satisfaction, the result flipped, with parental satisfaction exceeding marital satisfaction (p = 0.101, g = 0.181, and N = 82), possibly a result of eliminating many cases with high reported levels of marital satisfaction. For career plans, officer retention continued to exceed spouse retention (p < 0.001, g = 0.551, and N = 111) but with about half the effect size as before. Notably, WFC continued to exceed FWC (p < 0.001, g = 1.49, and N = 124) with nearly the same effect size. Thus, two of our comparisons seemed resistant to controls for marital social desirability.

H7A. Demographic variables predicted WFC and FWC using Pearson zero-order correlations with two-tailed testing. The results for demographic variables yielded few significant results (

Table 10. Pearson zero-order correlations between work–family and family–work conflict with variables related to concurrent or discriminant validity.

Variables	WFC	FWC
Demographic Factors
Age	−0.035	−0.034
Years as an officer	0.078	0.095
Education	−0.111 +	0.054
Married vs. not married	0.001	−0.101
Number many times married	0.092	−0.027
Number of children	0.055	−0.139 *
Preschool children (ages 1–5)	0.064	0.041
Adolescent children (ages 13–17)	−0.057	−0.038
Sex (1 = male; 2 = female)	−0.089	−0.104 +
Adopted as a child (1 = yes; 2 = no)	−0.050	−0.105 +
Age of youngest child	0.002	−0.043
Rank	−0.020	0.114 +
Spouse in the military	−0.151 *	0.008
Quarter years deployed	−0.146 *	−0.001
Independent Predictors
Marital social desirability	−0.387 ***	−0.274 ***
Intrinsic religiosity	−0.067	−0.007
Religious practice scale	−0.015	0.028
Satisfaction with your own personal spiritual life	−0.300 ***	−0.021
Marital process scale	−0.274 ***	−0.341 ***
Outcome Factors
Officer’s retention intentions	−0.133 *	−0.025
Spouse’s retention intentions (officer’s report)	−0.249 ***	−0.148 *
Satisfaction with amount of time you have had with your family since 2001	−0.451 ***	−0.078
Satisfaction with the Army as a way of life	−0.373 ***	−0.103 +
Satisfaction with your own life, overall	−0.337 ***	−0.149 *
Satisfaction with your family’s fit with Army life	−0.379 ***	−0.142 *
In general, your satisfaction with the Army’s support for the families of deployed soldiers	−0.255 ***	0.037
How difficult is it for officers such as yourself to maintain both the current frequency of deployments and your quality of family life	0.429 ***	0.117 +
Current marital satisfaction	−0.293 ***	−0.294 ***
Current parental satisfaction	−0.301 ***	−0.095
Current parental satisfaction (long scale)	−0.304 ***	−0.102
Marital instability index	0.336 ***	0.234 ***

Significance tests are two-tailed. + p < 0.10; * p < 0.05; and *** p < 0.001.

). Number of children was negatively related to FWC (−0.139; p < 0.05); WFC was negatively related to being in a dual-military career family (−0.151; p < 0.05) and to total number of quarter years deployed since 2001 (−0.146; p < 0.05). Given 28 total correlations, one would expect one or two significant by chance (p < 0.05), so our results should be considered exploratory.

H7B. Independent variable predictors of WFC and FWC were tested against WFC and FWC using Pearson zero-order correlations with two-tailed testing (Table 10). The marital process scale was significantly correlated with both WFC (−0.274; p < 0.001) and FWC (−0.341; p < 0.001). While intrinsic religiosity and religious practice scales were not significantly related to either WFC or FWC, satisfaction with own personal spiritual life was correlated significantly with WFC (−0.300; p < 0.001). Marital social desirability was correlated negatively with both WFC (−0.387; p < 0.001) and FWC (−0.274; p < 0.001).

H7C. Outcome variables potentially associated with WFC and FWC were assessed using Pearson zero-order correlations with two-tailed testing. Associations with WFC and the outcome variables were usually strong (r > 0.250; p < 0.001). Associations with FWC were less frequently significant and, even if significant, were smaller (r < 0.250).

H7D. We tested for interactions between the variables in Table 10 and interaction effects with sex, predicting both WFC and FWC. Since we had 62 tests, three significant results would have been expected by chance, which is what occurred, only for FWC. Hence, our results must be interpreted with caution, but may have some usefulness for future research. Regarding total quarters deployed, for men the regression equation was 9.69 + 0.011x with beta = 0.012 (not significant); for women, it was 10.34 − 0.523x with beta = −0.494 (p < 0.05). That result would seem to suggest that female officers find longer deployment experience to have a substantially greater impact on FWC. A similar pattern was found for officers’ own career plans. For men, the regression equation was 9.375 + 0.114x with beta = 0.011 (not significant), while for women, it was 38.50–7.50 with beta = −0.589 (p < 0.01). In other words, there was a much stronger relationship between their own career plans and FWC for women compared to men. For the variable satisfaction with your own life, overall, we found opposite patterns as a function of sex. For men, the regression equation was 8.35 + 0.662x with beta = 0.179 (p < 0.01); for women, an opposite pattern was found with an equation of 11.54 − 1.50x with beta = −0.342 (p < 0.13). For women, high FWC appeared to be associated with lower life satisfaction while the opposite occurred for men. In general, more variables predicted WFC than FWC, but FWC seemed more important to the career plans of women officers.

5. Limitations

While we considered symbolic interaction, systems theory, and conservation of resources theory here, one limitation was our lack of use of work–family enrichment or facilitation models (Greenhaus and Powell 2006; Silveira-Rodrigues et al. 2021; Wayne et al. 2004; Wong et al. 2022) or of life/developmental course models (Allen et al. 2019; De Angelis et al. 2018; Huffman et al. 2013; Segal et al. 2015; Wilmoth and London 2013), spillover theories (Greenhaus and Powell 2006), or the Contextual Model of Family Stress (Boss et al. 2016; Farnsworth and O’Neal 2025), which in part were developed after our data had been collected. Some research continues to focus only on negative spillover effects (Borowski et al. 2021), probably reflecting a tendency towards pessimism with respect to work–family conflicts in the military. However, these limitations also represent opportunities for future theory and research.

Our data were cross-sectional and do not permit conclusions about causal relationships among our variables. We did not measure WFC or FWC at two times, preventing us from determining test–retest reliability (Matthews et al. 2010, found correlations of 0.75 for WFC and 0.61 for FWC over one month). The response rates were low, although in line with some other previous military research (Cavaleiro et al. 2019, 30.2%; Military Family Advisory Network 2024, 20.4%; Pearce et al. 2025, 30.9%; Skomorovsky 2014; Skomorovsky et al. 2015, 21.3%; Skomorovsky et al. 2019, 17.2%; Skomorovsky et al. 2025a, 18.0%, 30.9%, and 30.3%; and Sullivan et al. 2020 and Park et al. 2023, 22.5%) and civilian research (Day and Chamberlain 2006, 18.7%; Matthews et al. 2010, 21%; and Thompson et al. 1999, 32.0%). The sample we used for validation featured an N of only 45, lower than the ideal of a sample size similar to that of the primary sample. The spouses of the soldiers surveyed were not part of the research; their opinions and beliefs were not sought or elicited, which eliminated any chance to assess certain dyadic variables, unlike other recent research (Huffman et al. 2017; Park et al. 2023). Furthermore, some research has found that female civilian spouses suffer the most from work–life imbalance (Vuga Bersnak et al. 2023), a group that was not part of our sample of Army respondents. Our study did not assess the impact of overseas combat deployments on the mental health of the children of our respondents, despite the importance of such research (Card et al. 2011; Chandra et al. 2010; Department of Defense 2010; Esposito-Smythers et al. 2011; Farnsworth and O’Neal 2025; Gorman et al. 2010; Hawkins et al. 2012; Huebner et al. 2007; Lester et al. 2010; McFarlane 2009; Mmari et al. 2009) nor on that of military spouses (Baptist et al. 2011; Woodall et al. 2020), especially while their soldiers were deployed overseas (Dimiceli et al. 2010; Renshaw et al. 2008; Skomorovsky 2014; Warner et al. 2009). It also did not focus on implications for military team members (Pickering 2018, p. 13). We did not assess personality traits (locus of control, affectivity, openness, agreeableness, conscientiousness, neuroticism, extraversion, and type A/B) (Allen et al. 2012; Bruck and Allen 2003; Wayne et al. 2004), genetic factors (Allen et al. 2023), sexual minority family structures (Allen and French 2023), or adverse childhood experiences (ACEs) (Woodall et al. 2020; Sullivan et al. 2020). We did not measure resilience (Smith et al. 2008) as did Wong et al. (2022). Our measures of WFC and FWC did not distinguish between spouse and parental components of work–family conflict as did Day and Chamberlain (2006). Our research did not use qualitative methods (Plummer 2023). Although we used a modified version of WFC and FWC scales developed previously, as others have done (Sullivan et al. 2020), we did not include alternative previously studied measures of work and family conflict that would have allowed for assessments of concurrent validity, especially longer measures that would have contained more than two dimensions of work and family issues. Among the 45 non-Army participants, there were only two spouses who were also in the military and there were only four women, rendering analysis of sex and military spouse variables too small for re-analysis as predictors of WFC, FWC, KMSS, and KPSS. While cross-national comparisons would have been useful (Allen et al. 2020), our sample included too few officers from outside the United States for effective analysis and their nationalities were not identified in our survey. Our study was conducted in 2007, many years prior to 2025, such that changes in the validity of our measures and patterns of correlations may have changed in the interim. Our participants were mostly active duty U.S. Army officers, rather than reserve component or enlisted personnel, or servicemembers from other nations or military branches (Navy, Air Force, and Marine Corps), limiting the generalization of our results. Specifically of the 45 non-Army respondents, 16 from the U.S. Army Reserve, 14 from the Army National Guard (states not specified), two from the U.S. Air Force, four from the U.S. Navy, and nine from international nations (nationalities not specified in our survey). However, long delays between data collection and reporting of results seem not uncommon with surveys of military personnel (e.g., Farnsworth and O’Neal 2025, 12 years; Hobfoll et al. 2012, 7 years; Huffman et al. 2008, 7 years; Huffman et al. 2014, 15 years; Park et al. 2023, 11 years; Sullivan et al. 2020, 8 years; Tidwell and Lucier-Greer 2025, 14 years; Vuga and Juvan 2013, 10 years; and Woodall et al. 2020, 7–9 years). One advantage of such delays is increased security for the respondents, whose characteristics have no doubt changed substantially over the decades. The study also lacked a measure of religious affiliation, in part because Army chaplains, regardless of their own affiliation, are directed to serve all soldiers and families, regardless of affiliations and perhaps to protect respondents’ privacy. However, future research should include a measure of a variety of such affiliations.

On the other hand, our study had several strengths. First, our sample size was large enough to exceed the Comrey and Lee (2016) requirement that factor analyses be performed with samples of at least 200 cases and to obtain adequate estimates of internal consistency reliability for our four multi-item scales. Second, our sample while limited in scope (field grade officers in the U.S. Army at an Army leadership training school) was also relatively rare due to various barriers to obtaining permission to survey such career personnel. To the best of our knowledge, only Skomorovsky et al. (2025b) has sampled a larger number of senior officers (N = 684, as part of a larger survey of the Canadian military, N = 4349). We also measured, albeit with one item, marital social desirability, unlike many previous studies that have overlooked the possibility of social desirability response bias even though they had depended on self-report measures (Borowski et al. 2021, p. 65). Our use of a modified version of Netemeyer et al.’s (1996) scale was fortuitous as Min et al. (2021) found Netemeyer et al.’s scale to be the overall best among the four scales they compared for reliability and validity. The study occurred after a prolonged war of several years, offering a contrast to research conducted during more peaceful times (Hollis 2008).

6. Discussion

Some research from the era of our study found relatively low effects of ongoing overseas combat deployments on military family life, especially for officer families (Karney and Crown 2007). Our sample represented relatively privileged officers of field grade rank attending an academic course in a garrison installation within the United States. Would such advantages make our sample immune from military stressors, including work–family conflict (WFC) and family–work conflict (FWC)? In terms of measuring WFC and FWC, our eight items did factor into two distinct scales and both scales featured adequate reliability. Our single-item measure of marital social desirability did evidence discriminant validity compared to parental satisfaction; eliminating cases very high on MSD appeared to be the most effective way to control for it. We found relatively few sex differences across our variables. Officers in dual-career military families reported having fewer older children but more younger children and reported lower WFC but higher levels of marital and parental satisfaction. One of the strongest findings was that WFC was consistently higher than FWC. Marital satisfaction tended to be higher than parental satisfaction. Officers rated their own retention intentions as greater than what they thought their spouses were. In general, demographic variables did not predict WFC or FWC. There were results that suggested that FWC was felt as more important by female officers. Virtually all of our variables indicating satisfaction with Army life were lower when WFC was higher; similar but weaker patterns were observed for FWC. In general our results indicate that despite their privileged position, even field grade Army officers in 2007 were experiencing difficulties with military stressors and deployments.

Our results have implications for family life education and other interventions with military families (Drummet et al. 2003). Although it is beyond the scope of this paper, it would be interesting to observe how OPTEMPO by year might change perceptions of WFC and FWC. WFC and FWC might be compared across units or installations to try to head off pending problems. While several demographics of Army personnel appear to have changed little since 2007, recent research from 2023 has pointed to new concerns related to racial and sexual minority groups and those living alone, who are more likely to report poorer health than White, married groups (Military Family Advisory Network 2024). Seventeen percent of active duty Army members reported suicidal thoughts. Loneliness was higher among women (65.8% vs. 53.7% for men) and among LGBTQ+ respondents (77.2%). To quote the MFAN results, “Higher life satisfaction was reported by married respondents, those with [only] two household members, those with no children under 18, caregivers, and those identifying as white. In contrast, lower life satisfaction was more common among separated or divorced individuals, those living alone, multiracial respondents, and those who transitioned over 11 years ago [left the Army]. Additionally, lower satisfaction was noted among LGBTQ+ individuals and those identifying as Black or African American.” (p. 24). By themselves, these results point to areas of needed improvement in terms of Army family services.

WFC and FWC may be modifiable with effective interventions, even if few government programs for veterans have yet addressed those concerns (Borowski et al. 2021). Wadsworth and Southwell (2011), Pickering (2017), and Richmond et al. (2024) have enumerated a number of policies that may help to reduce work-to-family and family-to-work conflicts in all militaries. Support from family-friendly defense policies, organizational leaders, chaplains/religiosity (Rawat 2017), and immediate supervisors can be very helpful in mitigating stress associated with WFC and FWC. Access to parental leave and a high quality, reliable childcare may be helpful. Flexible work arrangements can reduce WFC, when feasible. While the military may offer a host of support services, those services depend on effective frontline leadership and communication about their availability, without stigma, to families. Support must mean more than words but action to support soldiers and their families.

7. Conclusions

As Pickering (2018) herself concluded, “It should be evident that striving to achieve some sort of balance between one’s work and personal life is a challenge faced by many military personnel” (p. 15), which our results indicate includes higher ranking Army military officers. It appears that many researchers have been able to develop brief, reliable and valid measures of work–family and family–work conflict, even though most scales have used slightly different wordings to assess those concepts. Our results indicated that measures of social desirability response bias may need to be specific to the outcome variables of concern. In general we found few sex differences, but female officers seemed to be more influenced by family–work conflict than male officers. One of our most substantial findings was that work–family conflict was more prevalent than family–work conflict within our sample. Also, we found that marital satisfaction tended to be higher than parental satisfaction and that officers usually found their own retention intentions to be higher than their perception of that of their spouses. Satisfaction with the military was consistently and strongly related to lower levels of work–family conflict while similar but weaker trends were found for family–work conflict. Most of our results were found to cross-validate with a subsample of non-Army officers in our sample. In sum, our results confirm the need for military support for families, even among higher ranking officer families.