Spillover and Crossover Effects of Work-Family Conflict Among

Home , Crossover effects

Society and Mental Health 2020, Vol. 10(1) 35–60 Spillover and Crossover Ó American Sociological Association 2018 DOI: 10.1177/2156869318813006 Effects of Work-Family journals.sagepub.com/home/smh Conflict among Married and Cohabiting Couples

Deniz Yucel1 and Beth A. Latshaw2

Abstract The present study uses Wave 8 of the German Family Panel to test the spillover and crossover effects of work-family conflict on job satisfaction, relationship satisfaction, and mental health for individuals (actor effects) as well as their spouses/partners (partner effects) in dual-earning couples. We further contribute by assessing whether the results vary by gender and union type. Results suggest that among married couples, for job satisfaction, there are no gender differences in actor effects (but gender differences in partner effects), and actor and partner effects remain distinct. For relationship satisfaction, there are no gender differences in actor or partner effects, but both effects remain distinct. For mental health, however, there are gender differences in actor effects (but not in partner effects), and both effects remain distinct. Among cohabitors, there are no differences in actor effects by gender, and adding in partner effects does not significantly improve the models predicting all three outcomes. Some results also suggest differences in relationship dynamics between married and cohabiting couples.

Keywords work-family conflict, spillover, crossover, job satisfaction, relationship satisfaction, mental health

INTRODUCTION a child’s school play (WFC), whereas getting little sleep due to caring for a newborn baby can affect Work-family conflict is a topic of growing con- one’s ability to perform at a work meeting (FWC). cern in the twenty-first century as greater numbers Existing literature suggests that both WFC and of dual-earning couples seek to balance their FWC are correlated with a wide range of outcomes responsibilities in these two domains. Greenhaus (see e.g., Allen et al. 2000; Amstad et al. 2011; and Beutell (1985:77) define work-family conflict Bakker, Demerouti, and Dollard 2008; Frone, Rus- as “a form of inter-role conflict in which the role sell, and Barnes 1996; Gareis et al. 2009; Michel pressures from the work and family domains are et al. 2009; Ngo and Lui 1999). For example, mutually incompatible. . . . That is, participation research finds negative associations between in the work (family) role is made more difficult by virtue of participation in the family (work) role.” This type of interrole conflict operates in 1William Paterson University, Wayne, NJ, USA two distinct directions: One’s work role can hinder 2Widener University, Chester, PA, USA one’s role in the family domain (work-to-family Corresponding Author: conflict; WFC), or one’s family role can jeopar- Deniz Yucel, Department of Sociology, William Paterson dize one’s role in the work domain (family-to- University, 300 Pompton Road, 465 Raubinger Hall, work conflict; FWC). For example, having a dead- Wayne, NJ 07470, USA. line at work can affect one’s ability to attend Email: [email protected] 36 Society and Mental Health 10(1)

WFC and FWC and job satisfaction (Allen et al. emphasis being placed on other types of unions, 2000; Bellavia and Frone 2005; Ford, Heinen, such as cohabitors (Symoens and Bracke 2015). and Langkamer 2007; Kossek and Ozeki 1998). Young et al. (2014) conclude their work by Likewise, there is evidence that both forms of encouraging future researchers to more closely work-family conflict negatively correlate with examine variations in crossover effects among marital satisfaction (Chen and Li 2012; Matthews, married and cohabiting couples, particularly to Conger, and Wickrama 1996; Voydanoff 2005) assess whether there are differences in mental and mental health indicators, such as depression health outcomes due to differences across union and psychological strain (Kelloway, Gottlieb, statuses. We extend their work by closely explor- and Barham 1999; Vinokur, Pierce, and Buck ing these differences here. We also shift our theo- 1999). However, most studies to date have tested retical focus toward examining whether crossover the “spillover effects” of work-family conflict effects of work-family conflict on mental health using individual-level data (Amstad et al. 2011). might vary for married and cohabiting couples There has been much less in-depth examination due to differences in the “cost of caring” (Kessler of “crossover effects” (Young, Schieman, and and McLeod 1984) and apply life course perspec- Milkie 2014), particularly in terms of clarifying tive as a new theoretical lens to interpret differen- the complex processes through which stress and ces in work-family conflict by union type. Further, strain are transmitted across partners, resulting in we examine the relationship between work-family negative outcomes (e.g., declines in mental conflict and two other domain-specific outcomes: health). In a prior work, Young et al. (2014) relationship satisfaction (family) and job satisfac- accomplish this goal, becoming one of few studies tion (work). In doing so, we assess the usefulness to fill this gap in the literature and thoroughly of applying Amstad et al.’s (2011) theoretical examine the relationship between spouses’ WFC, framework to better understand the consequences family stressors, FWC, and respondents’ mental of work-family conflict across multiple domains health. Further, they contribute by teasing out of social life. the theoretical mechanisms by which the stress Accordingly, in the following, we examine the process model (Pearlin 1999) explains the effects actor effects (AEs) and partner effects (PEs) of of crossover and whether these dynamics might WFC and FWC on outcomes across three domains differ for men and women. (job satisfaction, relationship satisfaction, and We build on and extend Young et al.’s (2014) mental health) for dual-earning German couples. work by using couple-level data and dyadic anal- We contribute to the literature in three ways. First, ysis to study the crossover effects of work-family we empirically explore the relationship between conflict, something few studies have done (Ham- WFC, FWC, and these three outcomes using mer, Allen, and Grigsby 1997; Lu et al. 2016; Mat- couple-level data and dyadic data analysis. Sec- thews et al. 2006). These few existing studies have ond, our study tests whether there are gender dif- small, nonrepresentative samples that typically ferences in AEs and PEs. Third, we assess whether focus on specific occupations or outcomes, limit- AEs and PEs vary by union type by comparing ing the robustness and generalizability of their married and cohabiting couples. findings. Couple-level data are also important because as Young et al. (2014:5) point out, “spouses do not always have sufficient informa- THEORETICAL FRAMEWORKS tion about the other.” This leads to a type of bias Kenny and Acitelli (2001) call “assumed- The experience of work-family conflict is associ- similarity bias,” where one assumes another’s ated with negative outcomes in the well-being of experience is similar to one’s own experience. workers across multiple areas of social life. Fol- This type of bias might be especially present lowing a theoretical model first proposed by Bel- among cohabitors, who are typically in less com- lavia and Frone (2005) and more recently adopted mitted relationships of shorter durations (Brown by Amstad et al. (2011), these outcomes can be 2000, 2003), perhaps making their reports of part- grouped into three major domains: work-related ners’ behaviors less accurate or reliable because (e.g., burnout, satisfaction, and job performance), they lack sufficient information about one another. family-related (e.g., satisfaction and strain in the In fact, most existing studies on couples have relationship), or domain-unspecific (e.g., mental focused exclusively on married partners, with little health, physical health, and life satisfaction). For Yucel and Latshaw 37 the purpose of this paper, we will focus on one and Singla 2011) in predicting work-family con- outcome from each domain: (1) job satisfaction flict itself as well as various work-, family-, and (work), (2) relationship satisfaction (family), and domain-unspecific outcomes related to it. As (3) mental health (domain-unspecific). Amstad et al. (2011), Ford et al. (2007), and Amstad et al. (2011) adopt two alternate Nohe et al. (2015) point out in their meta-analyses, hypotheses for understanding the relationship the cross-domain perspective has been more between work-family conflict and the domain- “popular” in the existing literature; however, the specific outcomes: the cross-domain and the matching hypothesis has received more empirical matching hypotheses. The first hypothesis (Frone, support. For example, Shockley and Singla Russell, and Cooper 1992a) proposes that there is (2011) found that there is a stronger relationship a “cross-domain” relationship between work- between WFC and job satisfaction as well as family conflict and outcomes such that WFC between FWC and relationship satisfaction. Nev- should be more influential on family-related ertheless, some scholars (see e.g., Li et al. 2013) domain outcomes (e.g., relationship satisfaction) have questioned the generalizability of applying and FWC should more strongly affect work- the matching hypothesis in non-Western nations related domain outcomes (e.g., job satisfaction). like China. Given the likelihood that work-family In other words, conflict that emerges in one conflict might vary across cultures and contexts, domain has a deleterious effect on well-being in we also seek to test the usefulness of the cross- a separate domain of social life. Frone et al. domain and matching perspectives in explaining (1992a) theorized that when one role a person outcomes of WFC and FWC in Germany. occupies (e.g., teacher) interferes with another of their roles (e.g., mother), this individual will strug- gle to fulfill the responsibilities associated with Spillover/Actor Effects the latter role and become dissatisfied with this receiving role as a result (e.g., lower family satis- Existing research outlines two theoretical path- faction resulting from WFC). ways through which work-family conflict affects The “matching hypothesis” suggests the oppo- individuals and their partners. The first, spillover, site possibility—WFC should more negatively transpires when stressors in one domain generate impact work-related domain outcomes, and FWC strain that is then transferred to another domain, should more harshly impact family-related domain thereby affecting an individual’s well-being. It is outcomes. Here, because the root of the conflict said to occur because of interrole conflict (Green- lies in one domain, individuals are likely to expe- haus and Beutell 1985) that arises when individu- rience declines in well-being within that same als occupy multiple social roles (e.g., employee domain. According to Amstad et al. (2011:153), and mother), and it is linked to negative outcomes this matching hypothesis is rooted in the theoreti- in job satisfaction, relationship satisfaction, and cal notion that a person’s “negative affective reac- mental health (see e.g., Bellavia and Frone 2005; tions” and/or “strain reactions” to work-family Grzywacz, Almeida, and McDonald 2002; Grzy- conflict (e.g., lowered satisfaction at work because wacz and Bass 2003; Grzywacz and Marks 2000; they are spending less time with their family) will Michel et al. 2011). Negative spillover is fre- correspond to the domain where the conflict and quently explained using theoretical frameworks problems originate (e.g., increasing demands at such as the role-scarcity hypothesis (Edwards work). Thus, the theoretical underpinnings of the and Rothbard 2000) and conservation of resources matching hypothesis are rooted in what Shockley theory (Hobfoll 1989). The role-scarcity hypothe- and Singla (2011:864) refer to as “source attribu- sis suggests that individuals have a limited number tion,” which also links back to appraisal theories of resources, and thus, strain can develop when (Lazarus 1991) and the suggestion that “when multiple roles pull from the same resources. Sim- self-relevant roles are threatened, individuals ilarly, conservation of resources theory proposes tend to appraise the source of the threat that individuals typically strive to achieve and negatively.” maintain their resources, and thus, strain can Numerous empirical studies have tested the develop when a disruption in one’s workplace or efficacy of the cross-domain and matching family leads to an inability to balance roles and hypotheses (see e.g., Li, Lu, and Zhang 2013; protect resources in either domain, leading to Luk and Shaffer 2005; Nohe et al. 2015; Shockley a decline in well-being. In addition, the stress 38 Society and Mental Health 10(1) process model (Pearlin 1999) suggests that when We draw on these recent trends as well as sev- someone experiences primary stressors at work eral theoretical perspectives on gender to generate or at home, they can lead to secondary stressors predictions about how the AEs of work-family that heighten and/or proliferate conflict, affecting conflict might vary for men and women. Accord- mental health specifically. ing to Ridgeway (2011), despite movement toward These theories have been tested, and the prior egalitarianism in the time use among romantic research highlighted here demonstrates that there partnerships, gender inequality continues to persist is a negative association between work-family due to powerful traditional beliefs and expecta- conflict and job satisfaction, relationship satisfac- tions about gender that can quickly be used to tion, and mental health. In the following, we frame and organize social interactions. For exam- extend this literature and contribute by using ple, due to expectations surrounding the norms of couple-level data to evaluate whether there are roles like wife and mother, women will likely feel gender and/or union status differences in the spill- more responsibility over and invest more time in over and crossover effects of work-family conflict, the family domain, while men will feel more in both directions, on three outcomes that Amstad responsibility over and invest more time in the et al. (2011) would classify as being “domain spe- work domain (see e.g., Bakker et al. 2008; Brum- cific” (job satisfaction and relationship satisfac- melhuis et al. 2008; Voydanoff 2002). Accord- tion) and “domain unspecific” (mental health). In ingly, one might expect that men are more likely doing so, we also more thoroughly assess the use- to experience the negative AEs of WFC (due to fulness of applying Amstad et al.’s (2011) theoret- cultural expectations surrounding their career ical framework to examine work-family conflict roles, making work a greater source of conflict), across social contexts. while women are more likely to experience the negative AEs of FWC (due to cultural expectations surrounding their family roles, making fam- Gender Differences in Spillover/Actor ily a greater source of conflict). For example, Effects women are more likely than men to see their employee role as negatively competing with their The experiences of work-family conflict for men family responsibilities (Simon 1995), and frequent and women have changed over time. In the United work contact is correlated with greater levels of States, men’s reports of work-family conflict have guilt and distress for women but not men (Glavin, increased consistently across the past three deca- Schieman, and Reid 2011). des, particularly for fathers in dual-earner mar- Prior literature testing gender differences is not riages (Aumann et al. 2011; Galinsky, Aumann, consistent (Cinamon and Rich 2002). Some stud- and Bond 2011; Nomaguchi and Johnson 2009). ies have found stronger negative AEs of WFC This could be due in part to recent changes in for men than women (Frone et al. 1996), while both gender ideologies and gendered divisions of others have found that men are more affected by paid and unpaid labor (Winslow 2005). Women’s FWC spillover and women are more affected by labor force participation has increased, and fathers WFC spillover (Amstad and Semmer 2011; are also devoting more time to child care and Symoens and Bracke 2015). Despite the inconsis- housework (Bianchi, Robinson, and Milkie 2006; tent literature, because men and women still dem- Galinsky et al. 2011). Despite this movement onstrate unequal divisions of paid and unpaid toward greater equality, women are still more labor (Bianchi et al. 2006; Nomaguchi and likely to manage the care of children, and tradi- Johnson 2009; Sayer 2005) and traditional cultural tional male breadwinner expectations persist expectations about gender persist throughout (Bianchi et al. 2006; Nomaguchi and Johnson social interactions today (Ridgeway 2011), we 2009; Schieman and Glavin 2008; Schieman, Gla- expect that the AEs of WFC will be stronger for vin, and Milkie 2009). Moreover, scholars have men than women, while the AEs of FWC will be critiqued the workplace for being resistant to shift- stronger for women than men (Hypothesis 1). ing away from the male breadwinner model, thereby disadvantaging mothers (Ridgeway and Correll 2004; Williams 2001) and increasing lev- Crossover/Partner Effects els of work-family conflict for fathers (Aumann, A second theoretical pathway called crossover Galinsky, and Matos 2011). occurs at the couple level when the stress an Yucel and Latshaw 39 individual experiences from WFC or FWC is (Baruch, Biener, and Barnett 1987; Niles and transmitted to one’s partner, thereby affecting his Goodnough 1996; Rosenthal 1985). or her well-being. As Bakker et al. (2008:901) Theoretical perspectives on gender socializa- explain, “Whereas [spillover] is an intraindividual tion and empathy can be employed to generate transmission of stress or strain, crossover is additional hypotheses about why the outcomes of a dyadic, interindividual transmission of stress or partners’ work-family conflict might also vary strain.” Thus, conflict between work and family for men and women, particularly in terms of men- roles not only affects individual actors but their tal health consequences. For example, Symoens romantic partners. Westman (2001, 2006) elabo- and Bracke (2015) found that husbands’ WFC rates on this phenomenon by proposing three plau- had an impact on their wives’ mental health, sible mechanisms of transmission across couples. whereas wives’ WFC did not have a negative First, partners may feel one another’s stress and effect on their husbands’ mental health. According strain because they tend to empathize with the per- to Kessler and McLeod (1984) and Rosenfield and son they share a relationship with. Second, Smith (2010), the relationships between empathy, because partners’ lives are so closely connected, sensitivity, and femininity are transmitted to the strains experienced by one are often felt by women via gender socialization. As a result, the other, simply by virtue of them sharing their when individuals in a woman’s social network experiences, homes, and (often) children together. experience conflict, women might have a height- Finally, strain can affect an individual’s partner ened awareness of the conflict (Rosenfield and due to a decline in effective couple-level commu- Smith 2010) and experience greater levels of dis- nication or social interaction that results from the tress as a result (Kessler and McLeod 1984). work-family conflict the partner experiences (see Thus, women are not necessarily more vulnerable e.g., Amstad and Semmer 2011; Bakker and to stress related to work-family conflict but rather, Demerouti 2009; Bakker et al. 2008; Matthews are socialized to internalize and empathize when et al. 2006; Young et al. 2014). Moreover, the people they care about experience conflict (Simon stress process model (Pearlin 1999) also suggests 1998). Because of these connections between gen- that experiences of primary and secondary stres- der socialization and empathy, we expect that the sors can exacerbate and/or proliferate conflicts, PEs of FWC will also be stronger for women than thereby negatively affecting the mental health of men (Hypothesis 3). both individuals and their partners. Thus, the exist- Finally, we expect that AEs and PEs will differ. ing theoretical models and literature lead us to Unlike AEs, PEs are more likely to be indirect expect significant PEs for WFC and FWC. Specif- because they require transmission across individu- ically, due to crossover, those whose spouses have als and are more dependent on a variety of factors, higher WFC and FWC will report significantly including how close partners are, how much time lower job satisfaction, relationship satisfaction, they spend together, and their level of communica- and mental health (Hypothesis 2). tion (see e.g., Kenny and Cook 1999; Westman 2001, 2006). Thus, we expect that AEs will generally be stronger than PEs (Hypothesis 4). Gender Differences in Crossover/ Partner Effects COMPARING MARRIED AND As stated, we believe that the AEs of WFC will be COHABITING COUPLES stronger for men and the AEs of FWC will be stronger for women. Accordingly, if the effects As new family structures become more common, of WFC and FWC cross over to the partners of particularly in the United States and Europe, it is these individuals, we anticipate that these out- also increasingly important to compare the effects comes will also vary by gender. Because we of work-family conflict across unions. Further, hypothesized that men are more likely to experi- because our study tests whether AEs and PEs ence WFC due to persistent masculine breadwin- vary between married and cohabiting couples, it ner ideologies, female partners will likely have is critical to review general ways relationships dif- to compensate by investing even more time in fer across these two formations. First, research the family domain. Thus, we expect that PEs of suggests that cohabiting unions, compared to mar- WFC will be stronger for women than men riages, are less institutionalized (Cherlin 2004; 40 Society and Mental Health 10(1)

Yabiku and Gager 2009), leading to lower levels these differences among union types. Symoens of well-being for cohabitors (Soons, Kalmijn, and Bracke (2015) predicted that FWC would and Teachman 2009). In other words, the norms cause greater spillover strain for cohabiting indi- governing the behavior of individuals who cohabit viduals than married individuals because married are not as clearly delineated as they are among men and women have more committed, secure spouses (Cherlin 2004). As a result, cohabiting relationships and thus will not feel their well-being partnerships tend to be associated with lower lev- is threatened if they must devote more time to els of relationship satisfaction and commitment as work. Cohabitors, who are more prone to relation- well as a higher risk of dissolution (see e.g., ship uncertainty, are therefore at greater risk of Brown 2003, 2004; Brown, Manning, and Payne well-being declines resulting from FWC (see 2015). e.g., Knobloch and Knobloch-Fedders 2010; Kno- Second, cohabitation can have different mean- bloch, Miller, and Carpenter 2007). In contrast, the ings to those who enter into the arrangement. For same study hypothesized that WFC would have some, it is more akin to being single than married more consequential PEs for married couples because (Rindfuss and VandenHeuvel 1990), while for greater devotion and investment in one’s family others, it is mostly a short-term precursor to mar- among married partners equates to more negative riage or even a substitution for marriage (Heuve- outcomes in well-being when conflict arises. line and Timberlake 2004). In the United States While Symoens and Bracke (2015) ultimately today, cohabitation continues to be primarily found no differences in AEs and PEs for cohabit- short-lived and is very rarely viewed as a substitu- ing and married couples, we draw on their study to tion for marriage (Smock and Manning 2004). make our own predictions here. Due to higher lev- Cohabitors continue to receive far fewer rights els of commitment, relationship closeness, and and legal protections than married couples. In intimacy among married people as well as the addition, despite the fact that the majority of cou- increased emphasis on freedom and self-fulfill- ples will cohabit prior to first marriages, few ment among cohabitors, we infer that married cohabiting unions are long term in duration, and spouses will be more affected by both the WFC cohabitation is more common among individuals and FWC of their partners. In opposition, because with a lower socioeconomic status (Brown et al. cohabiting unions are not as committed or as likely 2015). to last over the long term, we infer that cohabitors While cohabitation has increased in all Euro- will be less affected by the WFC and FWC of their pean countries in recent decades, the highest rates partners. In other words, as Westman (2001, 2006) are found in Germany, Austria, and Norway explains, negative PEs are transmitted because of (Perelli-Harris et al. 2014). In German-speaking the empathy, shared connection, and support cou- countries, cohabitation is primarily viewed as ples experience by virtue of being in longer-term, a younger stage in the life course that tends to occur more committed relationships. Because cohabitors before marriage and children. In addition, cohabita- likely feel less commitment, connection, and sta- tion is associated with having greater levels of free- bility, the negative AEs of WFC and FWC might dom and an ability to explore self-fulfillment, seem more pronounced in the absence of a strong sometimes with multiple partners. Similar to the support system from their partners. United States, German cohabitors have very few Finally, we can also draw on differential vul- legal rights or responsibilities (Gassen and Perelli- nerability theories to generate predictions about Harris 2015; Le Goff 2002). Still, differences exist how the consequences of work-family conflict within Germany. In the Western region, cohabitors vary by marital status. According to this perspec- generally desire to marry and associate marriage tive, unmarried individuals are likely more vulner- with financial stability and plans to have children. able to experiencing distress related to the conse- In contrast, East German cohabitors express a lower quences of strain and conflict between work and desire to marry, see fewer differences between family roles (see e.g., Pearlin and Johnson 1977; cohabitation and marriage, and interpret having Simon 1998). In other words, marriage can serve children with someone as being more serious than as a protective barrier to the higher risks unmar- marriage (Perelli-Harris et al. 2014). ried individuals face (e.g., isolation and financial Based on this literature, we expect that AEs hardship). Thus, we expect that AEs will be stron- and PEs will vary between married and cohabiting ger among cohabiting couples and PEs will be couples in Germany. Very few studies have tested stronger among married couples (Hypothesis 5). Yucel and Latshaw 41

MATERIALS AND METHODS following list you see a number of statements that people can use to describe themselves. Please This study uses Wave 8 of the German Family read each statement and indicate from among the Panel (pairfam), which focuses on partnership four answers the one that corresponds to the way and family dynamics in Germany and is funded you feel in general. (1) My mood is melancholy. by the German Research Foundation (Bru¨derl (2) I am happy. (3) I am depressed. (4) I am sad. et al. 2017; Huinink et al. 2011). The survey, first (5) I am in desperation. (6) My mood is gloomy.”6 launched in 2008–2009, collects data from The answer categories range from 1 (almost never) a national, random sample of 12,402 anchor per- to 4 (almost always). The factor loadings of these sons as well as their spouses and partners annually. six items range from .60 to .75 and .58 to .71 for Questions about work-life balance were only married and cohabiting couples, respectively, asked in Waves 6 and 8.1 This study focuses on which all exceed the threshold level of .50 (Hair the more recent wave and is limited to married et al. 2006). Cronbach’s alpha scores for men and cohabiting couples where both spouses/part- and women are .78 and .83, respectively (for mar- ners are employed (1,043 married and 323 cohab- ried couples), and .78 and .81, respectively (for iting couples). cohabiting couples), indicating high internal reli- On average, among the couples in the sample, ability. After reverse coding the second item, sum- wives are approximately 38, and husbands are ming these six items, and taking the average of the approximately 41 years old, while female and summed score, higher numbers indicate lower lev- male cohabitors are around 29 and 31, respec- els of mental health. tively. Approximately 49 percent of the married men and 54 percent of the married women com- 2 pleted upper education, whereas 57 percent of Independent Variables the cohabiting men and 64 percent of the cohabiting women completed upper education. On aver- This study uses two main independent variables: age, married couples have been married for 11 WFC and FWC. To measure WFC, respondents years, and cohabiting couples have been together are asked: for 4 years. Among cohabiting couples, around 27 percent have plans to get married. Table 1 (1) Because of my workload in my job, presents detailed descriptive statistics for the cou- vocational training, or university education, ples in our sample and the full list of variables. my personal life suffers. (2) Even when I am doing something with my friends, partner, or family, I must often think about Dependent Variables work. (3) After the stress of work, I find it difficult to relax at home and/or to enjoy In this study, the three dependent variables are job my free time with others. (4) My work pre- satisfaction, relationship satisfaction, and mental vents me from doing things with my friends, health. Job satisfaction is measured by one indica- partner, and family more than I’d like. tor: “How satisfied are you with school, education and career?”3 The answer categories range from Each item ranges from 1 (not at all) to 5 (abso- 0 (very dissatisfied) to 10 (very satisfied). This lutely).7 The factor loadings of these four items item was highly skewed, and therefore, a log trans- range from .65 to .86 and .62 to .82 for married formation of the variable was used. Higher num- and cohabiting couples, respectively. Cronbach’s bers indicate greater levels of job satisfaction. alpha scores for men and women are .75 and .82, Relationship satisfaction is also measured by one respectively (for married couples), and .75 and item: “All in all, how satisfied are you with your .81, respectively (for cohabiting couples), indicat- relationship?”4 The answer categories range from ing high internal reliability. After summing these 0 (very dissatisfied) to 10 (very satisfied). This five items and taking the average to create a scale, item was also highly skewed; therefore, a log higher scores indicate greater levels of WFC. transformation of the variable was used.5 Higher To measure FWC, respondents are asked: numbers indicate greater levels of relationship satisfaction. Finally, mental health is measured using (1) Because I am often under stress in my six indicators. Respondents are asked: “In the private life, I have problems concentrating 42 Table 1. Descriptive Statistics for all Variables.

Married Couples Cohabiting Couples (N = 1,043) (N = 323) Variable Name Range Mean/Proportion SD Mean/Proportion SD

Dependent variables Female job satisfaction 1–3 1.83 .83 1.91 .84 Male job satisfaction 1–3 1.86 .81 1.81 .83 Female relationship satisfaction 1–3 2.18 .85 2.25 .83 Male relationship satisfaction 1–3 2.22* .84 2.11* .82 Female mental health 1–3.83 1.53*** .43 1.71*** .51 Male mental health 1–3.33 1.49** .42 1.57** .40 Independent variables Female: Work-to-family conflict 1–5 2.20*** .99 2.70*** 1.00 Female: Family-to-work conflict 1–5 1.66* .78 1.76* .70 Male: Work-to-family conflict 1–5 2.52 .89 2.51 .89 Male: Family-to-work conflict 1–5 1.65 .64 1.70 .66 Control variables Relationship duration (log) –2.48 to 5.75 4.54*** .99 .82*** 1.30 Presence of preschool children in the household 0–1 .33*** — .16*** — Female work hours 1–70 28.33*** 12.01 37.51*** 9.84 Male work hours 1–85 43.62*** 9.09 41.49*** 10.11 Female lower education 0–1 .08 — .06 — Female intermediate education 0–1 .38** — .30** — Female upper education 0–1 .54*** — .64*** — Male lower education 0–1 .16*** — .08*** — Male intermediate education 0–1 .35 — .34 — Male upper education 0–1 .49** — .57** — Female nationality 0–1 .95** — .98** — Male nationality 0–1 .96 — .97 — Female income (log) 4.32–9.10 7.01** .70 7.16** .60 Male income (log) 5.70–9.39 7.76*** .48 7.41*** .53 Couple lives in East Germany 0–1 .33 — .35 — Plans to get married (for cohabiting couples)a 0–1 — — .27 —

Note. The indicators of the dependent variable are coded so that higher scores indicate greater job satisfaction and relationship satisfaction and worse mental health. aPlans to get married was asked to cohabiting couples in the survey. Therefore, there is no information for married couples. *p \ .05. **p \ .01. ***p \ .001 (two-tailed tests). Yucel and Latshaw 43

on my work. (2) Because of my personal unions) and constrained models (i.e., coefficients schedule, I often lack time to do my work. are constrained to be equal for both genders and (3) The time I need for my partner, family, for both types of unions), the unconstrained model and friends keeps me from being more is preferred. However, if there are no differences, involved in my job, vocational training, or the constrained models with pooled estimates are university education. (4) Conflicts in my preferred (Kenny and Cook 1999). Finally, to personal life reduce my work performance. test whether AEs and PEs vary between married and cohabiting couples, multigroup analyses are Each item ranges from 1 (not at all) to 5 (abso- used (see Results section for details). lutely). The factor loadings of these four items range from .61 to .87 and .62 to .83 for married and cohabiting couples, respectively. Cronbach’s RESULTS alpha scores for men and women are .74 and .78, respectively (for married couples), and .74 and Descriptive Findings .76, respectively (for cohabiting couples), indicat- Descriptive findings for all variables are displayed ing high internal reliability. After summing these in Table 1. There are no differences between mar- five items and taking the average to create a scale, ried men (women) and cohabiting men (women) in higher scores indicate greater levels of FWC. terms of their reports of job and relationship satisfaction; however, cohabiting men and women are significantly more likely than married men and Control Variables women (respectively) to have worse mental health. There are no differences in reports of This study also uses several other variables that WFC and FWC between cohabiting and married have been shown to be associated with our inde- men; however, cohabiting women are more likely pendent and/or dependent variables (Bruck, Allen, to report WFC and FWC than married women. and Spector 2002; De Simone et al. 2014; Min- Cohabiting couples also have significantly shorter notte, Minnotte, and Bonstrom 2015; Roehling, relationships than married couples, while spouses Jarvis, and Swope 2005; Schieman et al. 2009; have significantly more preschool-aged children Voydanoff 2005, 2007; Yucel 2017a, 2017b): edu- than cohabitors. cation, nationality, work hours,8 income, marriage (relationship) duration,9 the presence of preschool- aged children in the household, plans to marry in the future (for cohabiting couples), and whether Multivariate Analyses the individual lives in the East or West Germany.10 Consistent with the approach in Yucel and Gassa- nov (2010) and Yucel (2017a), multivariate analyses proceed in five steps. Results for married cou- Analytical Strategy ples are displayed in Tables 2 through 4, while Table 5 presents the final models for cohabiting This study follows the same analytical approach of couples across all three dependent variables. prior research (Yucel 2017a; Yucel and Gassanov Tables showing the full models for cohabiting 2010). The actor-partner interdependence model couples can be provided on request. (APIM) framework is used, and dyadic data analysis is carried out using structural equation modeling (SEM) in AMOS 22.0. As part of dyadic data analysis, first, the AEs and PEs for our two main Married Couples independent variables, WFC and FWC,11 are First, we test if there are any AEs between tested. Next, using chi-square difference tests, respondents’ WFC, FWC, and their reports of this study tests the gender differences in AEs job satisfaction, relationship satisfaction, and men- and PEs between men and women and whether tal health. As shown in the first model (Model 1) AEs and PEs are equal to each other. The relative of Tables 2 through 4, significant AEs emerge. fits of these nested models in SEM are compared. For both husbands and wives, those with higher If the difference in chi-square between uncon- WFC (b = –.094, p \ .001; b = –.065, p \ .001 strained models (i.e., coefficients are estimated for husbands and wives, respectively) report sig- separately for both genders and for both types of nificantly lower job satisfaction. Higher FWC is 44

Table 2. Predictive Model of Job Satisfaction among Married Couples (N = 1,043 Couples).

Model 2 Model 4 Model 5 Model 1 Testing for Gender Model 3 Testing for Testing for Testing for Differences in Testing for Gender Differences Differences in Actor Actor Effects Actor Effects Partner Effects in Partner Effects and Partner Effects b (SE) b (SE) b (SE) b (SE) b (SE)

Individual-level actor effects Work-to-family conflict W: –.118*** (.032) –.160*** (.023) –.158*** (.023) –.158*** (.023) –.102*** (.016) H: –.200*** (.031) Family-to-work conflict W: –.121*** (.038) –.110*** (.028) –.108*** (.028) –.109*** (.028) –.042* (.020) H: –.109** (.042) Individual-level partner effects Work-to-family conflict W: .000 (.031) –.046* (.023) –.102*** (.016) H: –.094** (.033) Family-to-work conflict W: –.001 (.037) .023 (.028) –.042* (.020) H: .043 (.044) Chi-square 10.671* 14.450* 5.996* 10.456* 47.932*** df 462 46 Comparative Fit Index .97 .97 .97 .97 .97 Root mean square error of approximation .04 .04 .04 .04 .05 R2 (wives) .10 .11 .12 .12 .09 R2 (husbands) .11 .10 .09 .20 .09

Model 2 Model 4 Model 5 Model 1 Testing for Gender Model 3 Testing for Gender Testing for Testing for Differences in Testing for Differences in Differences in Actor Actor Effects Actor Effects Partner Effects Partner Effects and Partner Effects b (SE) b (SE) b (SE) b (SE) b (SE)

Individual-level actor effects Work-to-family conflict W: –.053 (.032) –.063** (.023) –.068** (.023) –.068** (.023) –.056** (.018) H: –.075* (.032) Family-to-work conflict W: –.168*** (.037) –.185*** (.028) –.194*** (.029) –.194*** (.029) –.132*** (.022) H: –.213*** (.043) Individual-level partner effects Work-to-family conflict W: –.001 (.032) –.045* (.023) –.056** (.018) H: –.089** (.032) Family-to-work conflict W: –.104** (.037) –.069* (.029) –.132*** (.022) H: –.032 (.043) Chi-square 23.114*** 24.749*** 2.942 7.064 24.477*** df 462 4 6 Comparative Fit Index .98 .98 .98 .98 .98 Root mean square error of approximation .05 .05 .02 .03 .05 R2 (wives) .06 .07 .09 .09 .08 R2 (husbands) .06 .05 .07 .07 .07

Table 4. Predictive Model of Mental Health among Married Couples (N = 1,043 Couples).

Model 2 Model 3 Model 4 Model 5 Model 1 Testing for Gender Testing for Testing for Testing for Testing for Differences in Partner Gender Differences Differences in Actor Actor Effects Actor Effects Effects in Partner Effects and Partner Effects b (SE) b (SE) b (SE) b (SE) b (SE)

Individual-level actor effects Work-to-family conflict W: .089*** (.016) .125*** (.011) .090*** (.017) .088*** (.016) .072*** (.008) H: .156***(.015) .155*** (.016) .154*** (.015) Family-to-work conflict W: .120*** (.019) .117*** (.014) .119*** (.019) .120*** (.019) .067*** (.010) H: .123*** (.021) .121*** (.021) .122*** (.021) Individual-level partner effects Work-to-family conflict W: .030* (.013) .021* (.010) .072*** (.008) H: .027 (.015) Family-to-work conflict W: .034* (.016) .019 (.014) .067*** (.010) H: .027 (.021) Chi-square 10.154* 21.790*** 1.969 4.570 115.341*** df 462 4 6 Comparative Fit Index .98 .98 .98 .98 .98 Root mean square error of approximation .04 .05 .03 .03 .05 R2 (wives) .16 .19 .16 .17 .15 R2 (husbands) .21 .17 .22 .22 .15

Note. This study also controls for the following variables: marital duration, work hours, education, nationality, income, living in East or West Germany, and having preschool children living in the household. Higher scores indicate worse mental health. Best-fitting model is highlighted in bold. *p \ .05. **p \ .01. ***p \ .001 (two-tailed tests). Yucel and Latshaw 47 associated with lower relationship satisfaction for wives of husbands with higher WFC report lower both wives (b = –.084, p \ .001) and husbands job satisfaction (b = –.033, p \ .01). Thus, Model (b = –.093, p \ .001). In addition, both spouses 3 is preferred over Model 4. For relationship satis- with higher WFC (b = .156, p \ .001; b = .089, faction and mental health, PEs work similarly p \ .001 for husbands and wives, respectively) between husbands and wives (Dx2 = .35, df =2, and FWC (b = .123, p \ .001; b = .120, p \ p . .10; Dx2 = 2.60, df =2,p . .10, respectively). .001 for husbands and wives, respectively) report Thus, Model 4, the restricted model, is preferred worse mental health. over Model 3. Specifically, individuals who are Model 2, Tables 2 through 4, shows the results married to spouses with higher FWC also report for gender differences in AEs. According to the lower relationship satisfaction (b = –.025, p \ chi-square difference test, comparing Models 1 .05). Finally, individuals who are married to and 2, the effects are not significantly different spouses with higher WFC conflict report lower for job satisfaction and relationship satisfaction mental health (b = .021, p \ .05). These findings (Dx2 = 4.51, df =2,p . .10; Dx2 = 3.30, df =2, are independent of the significant AEs in the p . .10, respectively). This indicates that there model. are no gender differences in AEs. Specifically, Model 5 in Tables 2 through 4 tests whether the negative effects of WFC on job satisfaction AEs and PEs are statistically equal. For example, (b = –.080, p \ .001) and FWC on relationship the analysis examines whether a person’s job sat- satisfaction (b = –.086, p \ .001) are similar isfaction is equally associated with one’s own between spouses. On the other hand, for models reports and one’s spouse’s reports of WFC and that predict mental health, the models are signifi- FWC. The chi-square difference tests comparing cantly different between Models 1 and 2 (Dx2 = Models 4 and 5 are significant, so Model 4 is pre- 11.64, df =2,p \ .01). This indicates that there ferred for job satisfaction, relationship satisfac- are indeed gender differences in AEs for mental tion, and mental health (Dx2 = 43.52, df =4, health. Specifically, the AE of WFC on mental p \ .001; Dx2 = 25.49, df =2,p \ .001; Dx2 = health is significantly stronger for husbands (b = 110.77, df =2,p \ .001, respectively). AEs and .156, p \ .001) than wives (b = .089, p \ .001). PEs are significantly different from each other. There are no gender differences, however, in the Specifically, the AE of WFC on job satisfaction AEs of FWC on mental health. (b = –.080, p \ .001) is significantly stronger Model 3 in Tables 2 through 4 tests whether than the husbands’ PE on their wives’ job satisfac- PEs significantly improve the path models of job tion (b = –.033, p \ .01), whereas wives’ PE on satisfaction, relationship satisfaction, and mental their husbands’ job satisfaction is not significant health. According to the chi-square difference (b = .008, p . .10). Moreover, the results suggest test, comparing Models 2 and 3 across Tables 2 that AE of FWC on relationship satisfaction (b = through 4, adding PEs does improve our under- –.089, p \ .001) is significantly stronger than standing of the correlates of job satisfaction, rela- the PE of FWC (b = –.025, p \ .05). Moreover, tionship satisfaction, and mental health among the AE of WFC on mental health (b = .088, p \ married couples (Dx2 = 9.04, df =4,p \ .10; .001; b = .154, p \ .001 for wives and husbands, Dx2 = 8.84, df =4,p \ .10; Dx2 = 19.82, df =4, respectively) is significantly stronger than the PE p \ .001, respectively). Model 3 is preferred (b = .021, p \ .05). Finally, the AE of FWC on over Model 2. Specifically, wives of husbands mental health for both wives and husbands is sig- who report higher WFC have lower job satisfac- nificant (b = .120, p \ .001; b = .122, p \ .001, tion (b = –.033, p \ .01), and husbands of wives respectively), whereas the PE of FWC is not. who report higher FWC have lower relationship Overall, the best-fitting model for job satisfac- satisfaction (b = –.029, p \ .05). Finally, husbands tion is Model 3 (see the bold text in Table 2), of wives who report higher WFC and FWC have where there are no gender differences in AEs but worse mental health (b = .030, p \ .05; b = gender differences in PEs and these effects remain .034, p \ .05, respectively). distinct. The best-fitting model for relationship Model 4 in Tables 2 through 4 tests whether satisfaction is Model 4 (see the bold text in Table there are gender differences in PEs. Comparing 3), where there are no gender differences in AEs Model 3 and Model 4, the results indicate that and PEs but these effects remain distinct. The there are gender differences in PEs for job satis- best-fitting model for mental health is Model 4 faction (Dx2 = 6.84, df =2,p \ .05). Specifically, (see the bold text in Table 4), where there are 48 Society and Mental Health 10(1)

Table 5. Final Model for Job Satisfaction, Relationship Satisfaction, and Mental Health among Cohabiting Couples (N = 323 couples)

Job satisfaction Relationship Satisfaction Mental Health b (SE) b (SE) b (SE)

Individual-level actor effects Work-to-family conflict –.239*** (.038) –.012 (.036) .151*** (.019) Family-to-work conflict –.082 (.052) –.143** (.048) .116*** (.026) Individual-level partner effects Work-to-family conflict Family-to-work conflict Chi-square 4.949 13.670* 7.246 df 666 Comparative Fit Index .98 .98 .98 Root mean square error of approximation .02 .05 .03 R2 (female partners) .18 .08 .21 R2 (male partners) .11 .11 .22

Note. This study also controls for the following variables: relationship duration, work hours, education, nationality, income, living in East or West Germany, plans to marry their current partner in the future, and having preschool children living in the household. Higher scores indicate greater job and relationship satisfaction but worse mental health. *p \ .05. **p \ .01. ***p \ .001 (two-tailed tests). gender differences in AEs but no gender differen- significantly add to the model. This is the best ces in PEs and these effects remain distinct. For model for predicting job satisfaction, relationship the job satisfaction model (see Model 3 in Table satisfaction, and mental health. This model also 2), the significant coefficients are the AEs (for shows that for cohabitors, individuals with higher both genders) and husbands’ PE of WFC (b = WFC report lower job satisfaction (b = –.114, p \ –.080, p \ .001; b = –.033, p \ .01, respectively). .001). In addition, those with higher FWC report For the relationship satisfaction model (see Model lower relationship satisfaction (b = –.049, p \ 4 in Table 3), the significant coefficients are the .01). Lastly, those with higher WFC and FWC AEs and PEs (for both genders) of FWC (b = report worse mental health (b = .151, p \ .001; –.089, p \ .001; b = –.025, p \ .05, respectively). b = .116, p \ .001, respectively). On the other For the mental health model (see Model 4 in Table hand, among cohabitors, partners’ reports do not 4), the AE of WFC for wives and husbands is b = significantly add to the models predicting job sat- .088 (p \ .001) and b = .154 (p \ .001), respec- isfaction, relationship satisfaction, and mental tively. The AE of FWC for wives and husbands health. Only the best-fitting model for all three on mental health is b = .120 (p \ .001) and b = dependent variables is displayed in Table 5. The .122 (p \ .001), respectively. On the other hand, full tables showing all five models for all three the PE of WFC on mental health is b = .021 dependent variables can be provided on request. (p \ .05). The best-fitting models for all three outcomes for married couples are also shown in Figure 1. Multigroup Analyses As the last step, this paper tests whether the AEs of WFC and FWC on job satisfaction, relationship Cohabiting Couples satisfaction, and mental health vary between mar- Table 5 shows the same five steps as described ried and cohabiting couples.12 Following the same previously for cohabiting couples. The results are analytical approach from prior research (Yucel different from married couples, indicating that 2017a; Yucel and Gassanov 2010), multigroup the best-fitting model is Model 2, where there analyses are used to examine group differences are no gender differences in AEs and PEs do not by running a series of nested models where all Yucel and Latshaw 49

Figure 1. Unstandardized coefficients for the best-fitting models in structural equation model (for job satisfaction, relationship satisfaction, and mental health) for married couples. Note. WFC = work-to-family conflict; FWC = family-to-work conflict; JS = job satisfaction; RS = relationship satisfaction; MH = mental health; e1–e6 = error terms associated with each indicated variable. *p \ .05. **p \ .01. ***p \ .001.

AEs of WFC and FWC are constrained to be equal suggest that adding equality constraints at each across groups in consecutive steps (Byrne 2013).13 step improves the fit of the baseline model (Model Then, using chi-square difference tests, the model 1). Thus, these findings suggest that most AEs do fit between each of these constrained models is not differ between married and cohabiting couples. compared to the model fit of the baseline uncon- On the other hand, two models indicate that strained model (Model 1).14 Most chi-square tests constraining some parameters to be equal across 50 Society and Mental Health 10(1) groups worsens the baseline model’s fit signifi- Our findings confirm the application of several cantly. Specifically, Step 4 in Table 6 constrains key theoretical frameworks in assessing the the AE of females’ WFC on mental health to dynamics of WFC and FWC on job satisfaction, equality across married and cohabiting couples. relationship satisfaction, and mental health. First, Based on the difference in the chi-square test, when considering the usefulness of applying the this result suggests that the AE of females’ WFC cross-domain (Frone et al. 1992a) versus matching on mental health differs significantly for married hypotheses (Amstad et al. 2011) to interpret the and cohabiting couples (Dx2 = 5.891, df =1, findings, our research lends strong support to the p \ .05) such that it is stronger among cohabiting latter. As predicted by the matching hypothesis, couples. In addition, Steps 2 and 4 in Table 6 con- and in support of the meta-analyses by Amstad strain the AEs of men and women’s WFC on job et al. (2011) and Nohe et al. (2015), we find that satisfaction to equality across married and cohab- for domain-specific outcomes, only the domain iting couples, respectively. Based on the differ- where the root of the conflict lies is significantly ence in the chi-square test, the AE of males’ and correlated with declines in well-being in that females’ WFC on job satisfaction differs signifi- same domain. This is evident in several of our cantly for married couples versus cohabiting cou- key findings. Among both married and cohabiting ples (Dx2 = 3.31, df =1,p\.10) such that the neg- couples, only WFC negatively impacts work- ative AE of females’ WFC on mental health and related domain outcomes (e.g., job satisfaction), the AEs of females’ and males’ WFC on job satis- and only FWC negatively impacts family-related faction are stronger among cohabiting couples. domain outcomes (e.g., relationship satisfaction). The highlighted numbers in Table 6 also show Thus, our findings align with two key theoretical a significant worsening of the model fit for job sat- underpinnings of the matching hypothesis— isfaction and mental health. A table displaying appraisal theories (Lazarus 1991) and source attri- nonsignificant chi-square test results when com- bution (Shockley and Singla 2011). For example, paring models for relationship satisfaction can be as predicted by the matching hypothesis, when provided on request. a salient role (spouse) is strained due to the increasing demands of another role (worker), an individual will begin to negatively appraise (via DISCUSSION lower job satisfaction) the original source of the strain on their relationship (working too many Prior research has studied the effects of WFC and hours). Despite this finding, the matching and FWC on a variety of outcomes (Allen et al. 2000; cross-domain perspectives are not without limita- Amstad et al. 2011; Bakker et al. 2008; Cooklin tions. Because of the inherent overlap and interac- et al. 2014; Kinnunen, Geurts, and Mauno 2004; tion that takes place between roles in the Minnotte 2016; Pedersen and Minnotte 2012), but workplace and home, to examine outcomes of most studies that have examined spillover and cross- work-family conflict rooted in the work versus over effects using couple-level data have had small the family domains, as though they are separate, sample sizes (Abeysekera and Gahan forthcoming; mutually exclusive entities, might oversimplify Amstad and Semmer 2011; Hammer et al. 1997; a more complex, fluid reality (Clark 2000). Matthews et al. 2006), with few exceptions (Shi- Further, in support of prior literature, results mazu et al. 2013). Even fewer studies have com- suggest that married individuals who experience pared married couples to other union types higher levels of work-family conflict have lower (Symoens and Bracke 2015). Additionally, whereas job satisfaction, lower relationship satisfaction, most prior research on work-family conflict focuses and worse mental health. These findings illustrate on workers from specific occupations or regions, this the continued usefulness of several theoretical study uses a large, representative sample of dual- concepts. As the role scarcity hypothesis (Edwards earning married and cohabiting couples in Germany, and Rothbard 2000) proposes, individuals have allowing for stronger generalizability. With this in limited resources and can feel strain when their mind, our study contributes by using couple-level roles as employees and spouses compete for their data to test the spillover and crossover effects of time. As conservation of resources theory explains work-family conflict (in both directions) on three (Hobfoll 1989), this strain can also be exacerbated outcomes across multiple domains and assessing when stressful events (e.g., an unexpected dead- whether they vary by gender or union type. line at work) cause an individual to be unable to Yucel and Latshaw 51

Table 6. Fit Indices for Nested Models in Multigroup Analyses for Mental Health and Job Satisfaction.

Mental Health Job Satisfaction Models df Chi-square CFI RMSEA df Chi-square CFI RMSEA

Step 1: The baseline model 10 17.407+ .98 .03 12 19.400+ .98 .02 Step 2: Actor effect of husbands’ 11 17.447+ .98 .03 13 22.522* .99 .02 work-to-family conflict is equal across groups Step 3: Actor effect of husbands’ 11 17.456+ .98 .03 13 19.626 .98 .02 family-to-work conflict is equal across groups Step 4: Actor effect of wives’ 11 23.298* .98 .03 13 22.522* .99 .02 work-to-family conflict is equal across groups Step 5: Actor effect of wives’ 11 17.427+ .98 .03 13 19.626 .98 .02 family-to-work conflict is equal across groups

Note. Multigroup analyses are only performed for variables that are the same for both married and cohabiting couples. The control variable “plans to marry in the future” cohabiting couples were therefore not used for multigroup analysis. No partner effects emerged for cohabiting couples, therefore only actor effects were compared between married and cohabiting couples. Higher scores indicate worse mental health and greater job satisfaction. CFI = Comparative Fit Index; RMSEA = root mean square error of approximation. +p \ .10. *p \ .05 (two-tailed tests).

balance both employee and spousal roles (because associated with work-family conflict are transmit- of their inability to conserve the amount of time ted across spouses. This finding that when hus- they need to both meet the deadline and be bands and wives experience WFC and FWC, the home to help their family). As a result of these job satisfaction, relationship satisfaction, and men- conflicts, our findings suggest that individuals tal health of their spouses are negatively affected can suffer a decline in job satisfaction, relationship can be interpreted through the theoretical lenses satisfaction, and mental health. Likewise, our find- of crossover stress and empathy among significant ings match the predictions of Pearlin’s (1999) others (Bakker and Demerouti 2009; Westman stress process model, which proposes that primary 2001, 2006). Essentially, when partners share stressors (e.g., unexpected work deadlines) might closely connected lives and frequently communi- lead to secondary stressors (e.g., arguments cate their work and family stressors, the “cost of between partners over how long a spouse should caring” for one another (Kessler and McLeod stay at work that evening). Thus, work-family con- 1984) likely translates into a decline in well-being flict can potentially heighten existing conflicts for both individuals and their spouses across mul- and/or proliferate additional conflicts, negatively tiple domains. These findings related to the trans- impacting individuals’ mental health. mission of stress due to empathy among partners While these findings on spillover are consistent are underscored by existing literature (see e.g., with well-established existing literature, our main Bakker et al. 2008; Matthews et al. 2006), partic- contribution lies in our use of couple-level data to ularly in terms of negative mental health outcomes compare the crossover effects of work-family con- (Young et al. 2014). Still, as expected, while add- flict by gender and union type. Among married ing in PEs improves our models, AEs are generally couples, our findings of significant negative PEs stronger and more significant than PEs because of WFC and FWC on job satisfaction and relation- they have a more direct and consequential impact ship satisfaction, respectively, support Amstad et on individuals’ behavioral outcomes. al.’s (2011) matching hypothesis, while the nega- When testing gender differences in AEs and tive PE of FWC on mental health supports West- PEs of WFC and FWC for married couples, how- man’s (2001, 2006) assertion that stress and strain ever, we find mixed support for the usefulness of 52 Society and Mental Health 10(1) applying theoretical frameworks on the persis- findings lend strong support to Amstad et al.’s tence of traditional gender expectations (Ridge- (2011) matching hypothesis. This suggests that way 2011). In contrast to our hypotheses, the neg- Amstad et al.’s theoretical framework can be use- ative AEs of both WFC and FWC on job fully applied to better understand the spillover and satisfaction and relationship satisfaction, respec- crossover dynamics of work-family conflict tively, are similar between husbands and wives, among individuals in both married and cohabiting and there are no gender differences in the PEs of partnerships. Our findings on cohabitors also align either FWC and WFC on relationship satisfaction with differential vulnerability theories and their and mental health, respectively. This lack of gen- prediction that cohabitors are often more vulnera- der differences in some of our analyses could be ble (emotionally and economically) and thus due to growing egalitarianism and equality in the might feel less commitment, connection, and sta- work and family experiences of men and women bility in their relationship, making the negative (Frone et al. 1992b; Young et al. 2014). Alterna- spillover effects of WFC seem more pronounced tively, some gender differences do emerge. As in the absence of a strong support system from predicted, the PE of WFC on job satisfaction is a partner (Pearlin and Johnson 1977; Simon 1998). stronger for women than men. In addition, the The lack of significant PEs for cohabiting AE of WFC on mental health is significantly couples also confirms our prediction that higher stronger for husbands than wives. In other words, levels of commitment, relationship closeness, in accordance with existing literature (see e.g., and intimacy among married people coupled Frone et al. 1996), it does appear that husbands’ with the increased emphasis on freedom and self- WFC crosses over to their wives, perhaps due to fulfillment among cohabitors (Brown 2003, 2004; the continued emphasis on empathy throughout Brown et al. 2015) might result in PEs being stron- women’s gender socialization. Likewise, men’s ger for married couples. This also matches the the- continued adherence to the work domain, even oretical mechanisms of crossover that Westman as societal expectations of involved fatherhood (2001, 2006) outlines, suggesting that negative become more dominant (Winslow 2005), is asso- crossover effects are typically transmitted because ciated with stronger negative effects of WFC on of the empathy and shared connection couples husbands’ mental health. Thus, our findings point share when they are in long-term, committed rela- to the need for more effective government and tionships (something cohabitors often lack). Thus, workplace policies designed to help married cou- it is possible that the cost of caring (Kessler and ples balance work and family. In light of recent lit- McLeod 1984; Wethington 2000) and the emo- erature documenting a rise in reports of work- tional burden of crossover stress, particularly its family conflict among married, dual-earning negative relationship to mental health (Young fathers in particular (Aumann et al. 2011; Galink- et al. 2014), is not as pronounced among cohabi- sky et al. 2011; Nomaguchi and Johnson 2009), tors as it is among spouses. These variations by this finding on mental health is critical. union status lend support to the use of couple-level While some of our results on the relationship data to study crossover stress among couples as it between work-family conflict and mental health reduces the possibility of assumed-similarity bias replicate those found in existing literature (Young and the possibility that cohabitors in particular et al. 2014), we extend prior work by using might lack sufficient information about one couple-level data to examine whether outcomes another. across multiple domains of social life vary by Life course perspective can also be applied to union status, assessing whether crossover stress better make sense of the absence of significant (in particular) operates differently among spouses PEs among cohabiting couples. This theoretical versus cohabiting partners. Among cohabitors, perspective “emphasizes the importance of time, findings suggest there are some significant AEs context, process, and meaning on human develop- between WFC, FWC, and our outcomes but no ment and family life” (Bengston and Allen significant PEs. Further, we find that the negative 1993:471). When using life course perspective to AEs of men’s and women’s WFC on job satisfac- examine partnering and relationship processes, tion and the negative AE of women’s WFC on one might infer that younger cohabitors interpret mental health are stronger among cohabiting cou- their unions as being shorter-term, less committed, ples than married couples. On one hand, similar to or more about convenience than longevity and the results on married couples, some of these thus are less likely to transition to marriage. In Yucel and Latshaw 53 other words, due to their being in a different causality is occurring or assume causal, directional life stage with varying relationship dynamics positioning between our key variables (for a com- (see e.g., Sassler 2010), cohabitors in our study prehensive review of antecedents of WFC and could be in qualitatively different than married FWC, see Michel et al. 2011). While we can respondents. only be certain of correlation between the varia- It is also possible that these variations we find bles in our study, the wealth of prior research the- among married and cohabiting partners are driven oretically and empirically demonstrating that by other factors that are not easily captured in sur- WFC and FWC likely precede changes in job sat- vey data. For example, prior research suggests that isfaction, relationship satisfaction, and mental factors such as differing norms (Brown and Booth health lends confidence to our findings (see e.g., 1996; Cherlin 2004), gender ideologies (Davis and Allen et al. 2000; Amstad et al. 2011; Bellavia Greenstein 2009; Sassler and Miller 2011), atti- and Frone 2005; Nohe et al. 2015). This is also tudes (Brown 2000; Stanley, Whitton, and Mark- supported by longitudinal research (Yucel and man 2004), and/or levels of social acceptance Fan 2018), where changes in work-family conflict and support from families (Brown 2003; Brown over time are associated with changes in well- and Booth 1996) might either select people into being outcomes among married couples. Second, cohabitation or emerge as a result of cohabiting. although this research makes a contribution by While we cannot measure or control for these fac- exploring the spillover and crossover effects of tors in the present study, our findings underscore WFC and FWC on various outcomes, one of our the growing importance of examining family con- key variables (relationship satisfaction) is mea- text (e.g., union status) and how it shapes work- sured using only one item. This could create reli- family conflict itself as well as various outcomes ability issues and/or introduce measurement error. associated with it (see e.g., Kossek and Lee Despite this, some prior research has shown strong 2017; Young 2015). As the role cohabition and reliability and validity for single-item measures of marriage play in family formations and life course relationship satisfaction (Huston, McHale, and experiences continues to shift (Brown et al. 2015), Crouter 1986; Huston and Vangelisti 1991). Third, the import of social context should not be over- the single item we have used to measure job satis- looked. We suggest scholars continue to test faction fails to measure it in the conventional way. the mediating and/or moderating effects of union Prior research shows mixed findings on the rela- status on outcomes (particularly mental health) tionship between the level of education and job using a larger, more diverse sample of cohabiting satisfaction (for a review, see Fabra and Camiso´n couples. In addition, future research should con- 2009) and that job and career satisfaction are sider more nuanced and integrated theoretical related but separate constructs (Kuchinke, Kang, approaches for explaining variations in the effects and Oh 2008; Lounsbury et al. 2007). Given this of work-family conflict among cohabiting and evidence, we infer that one might be satisfied married couples (Kossek and Lee 2017). with their job but not necessarily be satisfied with their education or that one’s job satisfaction does not necessarily indicate a satisfaction with Limitations and Suggestions for Future the career they may (or may not) have. Fourth, although this study mainly focused on WFC and Research FWC, various working conditions (i.e., job There are several limitations of this research. First, demands and job resources) could likely influence the cross-sectional design of the data limits our the outcomes in this study, as shown in some prior ability to make causal arguments. In addition, meta-analyses (Schaufeli 2017; Schaufeli and some of the relationships between our key varia- Taris 2014). In addition to their main effects, job bles are likely bidirectional. For example, those demands and job resources can act as moderators. with poor mental health and problems at work For example, job demands worsen the negative and/or in their relationships might experience effects of WFC and FWC on well-being and health higher WFC and FWC. Likewise, being dissatis- outcomes (Van Yperen and Hagedoorn 2003), fied at work might lead to discord at home whereas job resources are expected to alleviate (WFC), and experiencing marital strain might the negative effects of WFC and FWC (Hughes cause problems at work (FWC). Without the use and Parkes 2007). Thus, future research could of longitudinal data, we cannot conclude that expand on this study by testing the AEs and PEs 54 Society and Mental Health 10(1) of job demands and job resources on similar out- health over time. This study aims to contribute to comes. Moreover, though not our focus, the spill- prior research by acknowledging both the actor over and crossover effects of WFC and FWC effects (AEs) and partner effects (PEs) of WFC and might be more negatively related to our outcomes FWC on outcomes using couple-level data and dyadic data analysis. Second, there is high attrition when couples have a more unequal division of among cohabiting couples, and thus, using both household labor (Stevens, Kiger, and Riley waves would reduce the sample significantly, and 2001) and possess a more egalitarian gender ideol- that would prevent us from doing any statistical tests ogy (Minnotte et al. 2010), when families have for cohabiting couples and making meaningful com- greater child care responsibilities or parenting parisons between married and cohabiting couples. duties (Lin and Burgard 2018), and/or when there Lastly, the time change between Waves 6 (2013– is a lack of available child care (Frye and Breaugh 2014) and 8 (2015–2016) is only two years. Consis- 2004). Future research might explore the AEs/PEs tent with some prior research that also examines of such family-related variables and attitudes on changes in work-family conflict over a relatively similar outcomes. short period of time (e.g., one year in research by Kinnunen, Geurts, and Mauno 2004), in this study, the two years between Waves 6 and 8 might not be CONCLUSIONS a long enough time to experience any meaningful change in WFC and FWC. In conclusion, this study makes a significant con- 2. Upper education is equivalent to a postsecondary tribution by empirically evaluating the relation- education that either leads or does not lead to a uni- ship between WFC, FWC, and three outcomes versity degree or equivalent according to U.S. educa- (relationship satisfaction, job satisfaction, and tional standards (i.e., an associate’s, bachelor’s, mas- mental health) across multiple domains of social ter’s or PhD). life, testing both spillover and crossover effects. 3. Using a single item to measure job satisfaction (i.e., Further, we use couple-level data and dyadic “How satisfied are you with your job overall?”) has data analysis to test gender differences in AEs been common in prior research (Artz and Kaya 2014; Cotti, Haley, and Miller 2014; Hill 2005). and PEs and assess whether these effects differ However, the single-item measure we use in this by union type. Moreover, we employ a large rep- study (the only one offered in our data source) does resentative sample of dual-earning married and not measure job satisfaction in the conventional cohabiting couples in Germany, allowing for way. Please see the Discussion section for further stronger generalizability than existing studies explanation. focused only on married couples, workers from 4. While using multiple items to measure relationship specific occupations or regions, or outcomes in satisfaction is preferred, using a single item has one domain. Our findings point to a need for con- also been common in prior research (Hill 2005; Min- tinued research on how WFC and FWC affect not notte, Minnotte, and Bonstrom 2015; Schoen, Rog- just individuals but couples, particularly through ers, and Amato 2006; Yucel 2017b). Prior research has shown strong correlations between a single-item a closer examination of crossover effects transmit- measure for relationship satisfaction and other related across spouses. Future research might also tionship satisfaction scales, establishing reliability build on our findings by continuing to explore dif- (Huston, McHale, and Crouter 1986; Huston and ferences in the correlates of WFC and FWC Vangelist 1991). among couples in different types of unions. As 5. Skewness and kurtosis tests as well as Shapiro Wilk cohabitation becomes an increasingly common tests all revealed deviations from normality for both family form across the United States and Europe, the job satisfaction and relationship satisfaction more couple level-data and research are needed, scales. As recommended by Field (2009), the skewed particularly using larger data sets with greater scales were transformed using a log transformation, numbers of cohabitors. which has also been used in prior research for skewed satisfaction measures (see e.g., Waismel-Manor, Levanon, and Tolbert 2016). NOTES 6. The Center for Epidemiologic Studies Depression Scale (CES-D) scale has been widely used in prior 1. We did not use both waves in this study for several research to measure mental health (Baumann et al. reasons. First, the main scope of the paper was not 2008; Gutierrez and Michaud 2017). As cited in exploring the changes in work-to-family conflict Tho¨nnissen et al. (2016), the items used in this study (WFC) and family-to-work conflict (FWC) on job to measure mental health were adapted from the satisfaction, relationship satisfaction, and mental State-Trait Depression Scale, which was also used Yucel and Latshaw 55

in prior research (Spaderna, Schmukle, and Krohne 11. Between 97 and 99 percent of partners in the sample 2002). In another research study, it was shown that agreed on their responses for the following control both the CES-D and State-Trait Depression scales variables: the presence of preschool-aged children are suitable to assess depressive symptoms (Lehr in the household, plans to get married in the future et al. 2008). Further research established high reli- (for cohabiting couples), marital (relationship) dura- ability as well as construct and external validity tion, and whether respondents lived in East or West of this scale to measure mental health (Krohne et Germany. Thus, consistent with prior research al. 2002). Since then, this scale has been used in (Yucel 2017a; Yucel and Gassanov 2010), only some additional studies (see e.g., Weigl et al. 2016). the female partner’s report was used. 7. The scales we use for WFC and FWC in pairfam 12. The question of whether the PEs of WFC and FWC data are adapted from other national surveys such on job satisfaction, relationship satisfaction, and as the National Study of the Changing Workforce mental health vary between married and cohabiting (NSCW; Huinink et al. 2011), where WFC and couples cannot be tested using multigroup analyses FWC are measured by items that contain words since PEs only emerge for married couples and do such as “family or other important people in your not exist among cohabiting couples. Therefore, life” and “family/personal life.” Despite the fact only AEs of WFC and FWC are compared between that these measures of WFC and FWC are common married and cohabiting couples. in prior research, the items containing such phrases 13. Multigroup analyses are only performed for varia- do indicate a domain that is broader than family bles that are the same for both married and cohabit- alone. Prior research has thus used work-family bal- ing couples. Therefore, the control variable “plans ance interchangeably with work-life balance (see to marry the current partner in the future” for cohab- reviews by Chang, McDonald, and Burton 2010; itors was not used. Rogelberg 2007). 14. The baseline model (Model 1) is where all AEs are 8. To test the nonlinear effect of work hours, as sup- allowed to vary between married and cohabiting ported in prior research (Pencavel 2015), we mea- couples. sured this variable in three categories: part-time, full-time, and overtime work hours (reference cate- gory). The results did not show any unfavorable REFERENCES effects for those who work overtime for relationship satisfaction and mental health, but it had a negative Abeysekera, Lakmal, and Peter Gahan. Forthcoming. AE on job satisfaction for men and women (only “Work-Family Conflict among Australian Dual- among cohabiting couples). Additionally, both earner Couples: Testing The Effects of Role Salience effects were only moderate (p \ .10), and they Crossover and Gender.” The International Journal of did not change the main effects of WFC and Human Resource Management. doi:10.1080/09585 FWC on our outcome measures. 192.2017.1296015 9. We tested for interactions between WFC, FWC, and Allen, Tammy, David Herst, Carly Bruck, and Martha the duration of cohabitation. We used continuous Sutton. 2000. “Consequences Associated with measures for duration of cohabitation as well as cat- Work-to-family Conflict: A Review and Agenda egorical measures (been together for less than 5 for Future Research.” Journal of Occupational years [\5], N = 233; and for at least 5 years [5 Health Psychology 5(2):278–308. years], N = 90). We could not divide the duration Amstad, Fabienne, Laurenz L. Meier, Ursula Fasel, of cohabitation into more than two categories due Achim Elfering, and Norbert K. Semmer. 2011. “A to the small sample size. Nonetheless, these interac- Meta-analysis of Work-Family Conflict and Various tions were not significant. Outcomes with a Special Emphasis on Cross-domain 10. Consistent with prior research on differences among Versus Matching-domain Relations.” Journal of cohabitors within Germany (see e.g., Perelli-Harris Occupational Health Psychology 16(2):151–69. et al. 2014), this study controlled for whether an Amstad, Fabienne, and Norbert K. Semmer. 2011. individual lives in East or West Germany. We also “Spillover and Crossover of Work- and Family- tested for interactions between this dummy variable related Negative Emotions in Couples.” Psychology and WFC and FWC, but none of them were signif- of Everyday Activity 4:43–55. icant. This indicated that there was no need to divide Artz, Benjamin, and Ilker Kaya. 2014. “The Impact of the sample into East or West Germany. In addition, Job Security on Job Satisfaction in Economic Con- the number of previous marriages (for married cou- tractions Versus Expansions.” Applied Economics ples) and partners with whom the anchor cohabited 46(24):2873–90. (for cohabiting couples) were only constructed for Aumann, Kerstin, Ellen Galinsky, and Kenneth Matos. the anchor but not the partner. Therefore, these var- 2011. The New Male Mystique. New York: Families iables were not included in the analysis. and Work institute. 56 Society and Mental Health 10(1)

Bakker, Arnold B., and Evangelia Demerouti. 2009. Castiglioni, Stefan Fiedrich, Christine Finn, Madison “The Crossover of Work Engagement between Garrett, Kristin Hajek, Michel Herzig, Bernadette Working Couples: A Closer Look at the Role of Huyer-May, Ru¨diger Lenke, Bettina Mu¨ller, Timo Empathy.” Journal of Managerial Psychology 24: Peter, Claudia Schmiedeberg, Philipp Schu¨tze, Nina 220–36. Schumann, Carolin Tho¨nnissen, Martin Wetzel, and Bakker, Arnold B., Evangelia Demerouti, and Maureen Barbara Wilhelm. 2017. The German Family Panel F. Dollard. 2008. “How Job Demands Affect Part- (pairfam). GESIS Data Archive, Cologne. ZA5678 ners’ Experience of Exhaustion: Integrating Work- Data file Version 8.0.0, doi:10.4232/pairfam. 5678. Family Conflict and Crossover Theory.” Journal of 8.0.0 Applied Psychology 93(4):901–11. Brummelhuis, Lieke, Tanja van der Lippe, Esther Baumann, Miche`le, Raymond Meyers, Etienne Le Kluwer, and Henk Flap. 2008. “Positive and Nega- Bihan, and Claude Houssemand. 2008. “Mental tive Effects of Family involvement on Work-related Health (GHQ12; CES-D) and Attitudes towards the Burnout.” Journal of Vocational Behavior 73(3): Value of Work among Inmates of a Semi-open 387–96. Prison and the Long-term Unemployed in Luxem- Byrne, Barbara. 2013. Structural Equation Modeling bourg.” BMC Public Health 8:214. with AMOS: Basic Concepts, Applications, and Pro- Baruch, Grace K., Lois Biener, and Rosalind C. Barnett. gramming. New York: Routledge. 1987. “Women and Gender in Research on Work and Chang, Artemis, Paula McDonald, and Pauline Burton. Family Stress.” American Psychologist 42(2): 2010. “Methodological Choices in Work Life Bal- 130–36. ance Research 1987 To 2006: A Critical Review.” Bellavia, Gina, and Michael Frone. 2005. “Work-Family International Journal of Human Resource Manage- Conflict.” Pp. 113–48 in Handbook of Work Stress, ment 21(13):2381–413. edited by J. Barling, E. K. Kelloway, and M. Frone. Chen, Lung Hung, and Tsui-Shan Li. 2012. “Role Bal- Thousand Oaks, CA: Sage. ance and Marital Satisfaction in Taiwanese Couples: Bengston, Vern, and Katherine Allen. 1993. “The Life An Actor-Partner interdependence Model Approach.” Course Perspective Applied to Families over Time.” Social indicators Research 107(1):187–99. Pp. 469–504 in Sourcebook of Family Theories and Cherlin, Andrew. 2004. “The Deinstitutionalization of Methods: A Contextual Approach,editedbyP.G. American Marriage.” Journal of Marriage and Fam- Boss,W.J.Doherty,R.LaRossa,W.R.Schumm, ily 66(4):848–61. and S. K. Steinmetz. New York, NY: Plenum Press. Cinamon, Rachel, and Yisrael Rich. 2002. “Gender Dif- Bianchi, Suzanne, John Robinson, and Melissa Milkie. ferences in the Importance of Work and Family 2006. Changing Rhythms of American Family Life. Roles: Implications for Work-Family Conflict.” Sex New York: Russell Sage. Roles 47:531–41. Brown, Susan. 2000. “Union Transitions among Cohab- Clark, Sue. 2000. “Work/Family Border Theory: A New itors: The Significance of Relationship Assessments Theory of Work/Family Balance.” Human Relations and Expectations.” Journal of Marriage and Family 53:747–70. 62(3):833–46. Cooklin, Amanda, Elizabeth Westrupp, Lyndall Straz- Brown, Susan. 2003. “Relationship Quality Dynamics of dins, Rebecca Giallo, Anne Martin, and Jan Nichol- Cohabiting Unions.” Journal of Family Issues 24: son. 2014. “Mothers’ Work-Family Conflict and 583–601. Enrichment: Associations with Parenting Quality Brown, Susan. 2004 Moving from Cohabitation to Mar- and Couple Relationship.” Child: Care, Health and riage: Effects on Relationship Quality.” Social Sci- Development 41:266–77. ence Research 33:1–19. Cotti, Chad D., M. Ryan Haley, and Laurie A. Miller. Brown, Susan, and Alan Booth. 1996. “Cohabitation 2014. “Workplace Flexibilities, Job Satisfaction Versus Marriage: A Comparison of Relationship and Union Membership in the US Workforce.” Brit- Quality.” Journal of Marriage and Family 58(3): ish Journal of Industrial Relations 52(3):403–25. 668–78. Davis, Shannon N., and Theodore N. Greenstein. 2009. Brown, Susan, Wendy Manning, and Krista Payne. 2015. “Gender Ideology: Components, Predictors, and “Relationship Quality among Cohabiting Versus Consequences.” Annual Review of Sociology 35: Married Couples.” Journal of Family Issues 38(12): 87–105. 1730–53. De Simone, Silvia, Jessica Lampis, Diego Lasio, Fran- Bruck, Carly S., Tammy D. Allen, and Paul E. Spector. cesco Serri, Gianfranco Cicotto, and Daniela Putzu. 2002. “The Relation between Work-Family Conflict 2014. “influences of Work-Family Interface on Job and Job Satisfaction: A Finer-grained Analysis.” and Life Satisfaction.” Applied Research in Quality Journal of Vocational Behavior 60(3):336–53. of Life 9(4):831–61. Bru¨derl, Josef, Karsten Hank, Johannes Huinink, Bern- Edwards, Jeffrey R., and Nancy P. Rothbard. 2000. hard Nauck, Franz J. Neyer, Sabine Walper, Philipp “Mechanisms Linking Work and Family: Clarifying Alt, Elisabeth Borschel, Petra Buhr, Laura the Relationship between Work and Family Yucel and Latshaw 57

Constructs.” The Academy of Management Review Grzywacz, Joseph G., and Brenda L. Bass. 2003. “Work, 25(1):178–99. Family, and Mental Health: Testing Different Mod- Fabra, M. Eugenia, and Cesar Camiso´n. 2009. “Direct els of Work-Family Fit.” Journal of Marriage and and Indirect Effects of Education on Job Satisfaction: Family 65(1):248–61. A Structural Equation Model for the Spanish Case.” Grzywacz, Joseph G., and Nadine F. Marks. 2000. Economics of Education Review 28(5):600–10. “Reconceptualizing the Work-Family Interface: An Field, Andy. 2009. Discovering Statistics Using SPSS. Ecological Perspective on the Correlates of Positive 3rd ed. London: Sage Publications. and Negative Spillover between Work and Family.” Ford, Michael T., Beth A. Heinen, and Krista L. Lang- Journal of Occupational Health Psychology 5(1): kamer. 2007. “Work and Family Satisfaction and 111–26. Conflict: A Meta-analysis of Cross-domain Rela- Gutierrez, Italo A., and Pierre-Carl Michaud. 2017. tions.” Journal of Applied Psychology 92(1):57–80. “Whistle While You Work: Job insecurity and Older Frone, Michael R., Marcia Russell, and Grace M. Workers’ Mental Health in the United States.” Work- Barnes. 1996. “Work-Family Conflict, Gender, and ing Paper. Retrieved October 15, 2018 (https://www Health-related Outcomes: A Study of Employed .cirano.qc.ca/files/publications/2017s-21.pdf). Parents in Two Community Samples.” Journal of Hair, Joseph F., William C. Black, Barry J. Babin, Rolph Occupational Health Psychology 1(1):57–69. E. Anderson, and Ronald L. Tatham. 2006. Multivar- Frone, Michael R., Marcia Russell, and M. Lynne iate Data Analysis. 6th ed. Uppersaddle River, NJ: Cooper. 1992a. “Antecedents and Outcomes of Pearson Prentice Hall. Work-Family Conflict: Testing A Model of the Hammer, Leslie B., Elizabeth Allen, and Tenora D. Work-Family Interface.” Journal of Applied Psy- Grigsby. 1997. “Work-Family Conflict in Dual- chology 77(1):65–78. earner Couples: Within-individual and Crossover Frone, Michael R., Marcia Russell, and M. Lynne Effects of Work and Family.” Journal of Vocational Cooper. 1992b. “Prevalence of Work-Family Con- Behavior 50:185–203. flict: Are Work and Family Boundaries Asymmetri- Heuveline, Patrick, and Jeffrey M. Timberlake. 2004. cally Permeable?” Journal of Organizational Behav- “The Role of Cohabitation in Family Formation: ior 13(7):723–29. The United States in a Comparative Perspective.” Frye, N. Kathleen, and James A. Breaugh. 2004. Journal of Marriage and Family 66(5):1214–30. “Family-friendly Policies, Supervisor Support, Hill, E. Jeffrey. 2005. “Work-Family Facilitation and Work-Family Conflict, Family-Work Conflict, and Conflict, Working Fathers and Mothers, Work- Satisfaction: A Test of Conceptual Model.” Journal Family Stressors and Support.” Journal of Family of Business and Psychology 19(2):197–206. Issues 26(6):793–819. Galinsky, Ellen, Kerstin Aumann, and James T. Bond. Hobfoll, Stevan. 1989. “Conservation of Resources. A 2011. Times Are Changing: Gender and Generation New Attempt at Conceptualizing Stress.” American at Work and at Home. New York: Families and Work Psychologist 44(3):513–24. institute. Hughes, Emily, and Katherine Parkes. 2007. “Work Gareis, Karen C., Rosalind Chait Barnett, Karen A. Ertel, Hours and Well-being: The Roles of Worktime Con- and Lisa F. Berkman. 2009. “Work-Family Enrich- trol and Work-Family interference.” Work & Stress ment and Conflict: Additive Effects, Buffering, or 21(3):264–78. Balance.” Journal of Marriage and Family 71: Huinink, Johannes, Josef Bru¨derl, Bernhard Nauck, 696–707. Sabine Walper, Laura Castiglioni, and Michael Feld- Gassen, Nora, and Brienna Perelli-Harris. 2015. “The haus. 2011. “Panel Analysis of intimate Relation- increase in Cohabitation and the Role of Union Sta- ships and Family Dynamics (Pairfam): Conceptual tus in Family Policies: A Comparison of 12 European Framework and Design.” Journal of Family Countries.” Journal of European Social Policy 25(4): Research 23:77–101. 431–49. Huston, Ted L., Susan M. McHale, and Ann C. Crouter. Glavin, Paul, Scott Schieman, and Sarah Reid. 2011. 1986. “When the Honeymoon’s over: Changes in the “Boundary-spanning Work Demands and Their Con- Marriage Relationship over the First Year.” Pp. sequences for Guilt and Psychological Distress.” 109–32 in The Emerging Field of Personal Relation- Journal of Health and Social Behavior 52(1):43–57. ships, edited by R. Gilmour and S. Duck. Hillsdale, Greenhaus, Jeffrey H., and Nicholas J. Beutell. 1985. NJ: Erlbaum. “Sources of Conflict between Work and Family Huston, Ted L., and Anita L. Vangelisti 1991. Roles.” The Academy of Management Review “Socioemotional Behavior and Satisfaction in Mari- 10(1):76–88. tal Relationships: A Longitudinal Study.” Journal Grzywacz, Joseph G., David M. Almeida, and Daniel A. of Personality and Social Psychology 61:721–33. McDonald. 2002. “Work-Family Spillover and Daily Kelloway, E. Kevin, Benjamin H. Gottlieb, and Lisa Reports of Work and Family Stress in the Labor Barham. 1999. “The Source, Nature, and Direction Force.” Family Relations 51(1):28–36. of Work and Family Conflict: A Longitudinal 58 Society and Mental Health 10(1)

Investigation.” Journal of Occupational Health Psy- Scales (STDS-T): A Comparative Analysis of Cut- chology 4(4):337–46. off Scores.” Diagnostica 54(2):61–70. Kenny, David, and Linda Acitelli. 2001. “Accuracy and Li, Chaoping, Jiafang Lu, and Yingying Zhang. 2013. Bias in the Perception of the Partner in a Close Rela- “Cross-domain Effects of Work-Family Conflict on tionship.” Journal of Personality and Social Psychol- Organizational Commitment and Performance.” ogy 80:439–48. Social Behavior and Personality 41(10):1641–54. Kenny, David A., and William Cook. 1999. “Partner Lin, Katherine Y., and Sarah A. Burgard. 2018. Effects in Relationship Research: Conceptual Issues, “Working, Parenting and Work-Home Spillover: Analytic Difficulties, and Illustrations.” Personal Gender Differences in the Work-Home interface Relationships 6:433–48. across the Life Course.” Advances in Life Course Kessler, Ronald C., and Jane D. McLeod. 1984. “Sex Research 35:24–36. Differences in Vulnerability to Undesirable Life Lounsbury, John W., Lauren Moffitt, Lucy W. Gibson, Events.” American Sociological Review 49:620–31. Adam W. Drost, and Mark Stevens. 2007. “An inves- Kinnunen, Ulla, Sabine Geurts, and Saija Mauno. 2004. tigation of Personality Traits in Relation to Job and “Work-to-family Conflict and Its Relationship with Career Satisfaction of Information Technology Pro- Satisfaction and Well-being: A One Year Longitudi- fessionals.” Journal of information Technology nal Study on Gender Differences.” Work and Stress 22(2):174–83. 18:1–22. Lu, Chang-qin, Jing-Jing Lu, Dan-yang Du, and Paula Knobloch, Leanne K., and Lynne M. Knobloch-Fedders. Brough. 2016. “Crossover Effects of Work-Family 2010. “The Role of Relational Uncertainty in Conflict among Chinese Couples.” Journal of Mana- Depressive Symptoms and Relationship Quality: gerial Psychology 31(1):235–50. An Actor-Partner Interdependence Model.” Journal Luk, Dora M., and Margaret A. Shaffer. 2005. “Work of Social and Personal Relationships 27:137–59. and Family Domain Stressors and Support: Within- Knobloch, Leanne K., Laura E. Miller, and Katy E. Car- and Cross-domain influences on Work-Family Con- penter. 2007. “Using the Relational Turbulence flict.” Journal of Occupational and Organizational Model to Understand Negative Emotion Within Psychology 78:489–508. Courtship.” Personal Relationships 14(1):91–112. Matthews, Lisa S., Rand D. Conger, and Kandauda A. S. Kossek, Ellen, and Kyung-Hee Lee. 2017. Wickrama. 1996. “Work-Family Conflict and Mari- “Work-Family Conflict and Work-Life Conflict.” tal Quality: Mediating Processes.” Social Psychology Oxford Research Encyclopedia of Business and Man- Quarterly 59(1):62–79. agement. Retrieved October 15, 2018 (http://busi Matthews, Russell A., Regan E. Del Priore, Linda K. ness.oxfordre.com/view/10.1093/acrefore/97801902 Acitelli, and Janet L. Barnes-Farrell. 2006. 24851.001.0001/acrefore-9780190224851-e-52 ). “Work-to-Relationship Conflict: Crossover Effects Kossek, Ellen, and Cynthia Ozeki. 1998. “Work-Family in Dual-earner Couples.” Journal of Occupational Conflict, Policies, and the Job-Life Satisfaction Health Psychology 11(3):228–40. Relationship: A Review and Directions for Organiza- Michel, Jesse S., Lindsey M. Kotrba, Jacqueline K. tional Behavior–Human Resources Research.” Jour- Mitchelson, Malissa A. Clark, and Boris B. Baltes. nal of Applied Psychology 83(2):139–49. 2011. “Antecedents of Work-Family Conflict: A Krohne, Heinz Walter, Stefan C. Schmukle, Heike Spa- Meta-analytic Review.” Journal of Organizational derna, and Charles D. Spielberger. 2002. “The State- Behavior 32:689–725. Trait Depression Scales: An international Compari- Michel, Jesse S., Jacqueline K. Mitchelson, Lindsey M. son.” Anxiety, Stress, and Coping 15(2):105–22. Kotrba, James M. LeBreton, and Boris B. Baltes. Kuchinke, K. Peter, Hye-Seung Kang, and Seok-Young 2009. “A Comparative Test of Work-Family Conflict Oh. 2008. “The influence of Work Values on Job Models and Critical Examination of Work-Family Link- and Career Satisfaction and Organizational Commit- ages.” Journal of Vocational Behavior 74(2):199–218. ment among Korean Professional Level Employees.” Minnotte, Krista Lynn. 2016. “Extending the Job Asia Pacific Education Review 9:552–64. Demands-Resources Model: Predicting Perceived Lazarus, Richard. 1991. Emotion and Adaptation. New Parental Success among Dual-earners.” Journal of York: Oxford University Press. Family Issues 37(3):416–40. Le Goff, Jean-Marie. 2002. “Cohabiting Unions in Minnotte, Krista Lynn, Michael C. Minnotte, and Jordan France and West Germany: Transitions to First Birth Bonstrom. 2015. “Work–Family Conflicts and Mari- and First Marriage.” Demographic Research 7: tal Satisfaction among U.S. Workers: Does Stress 593–624. Amplification Matter?” Journal of Family and Eco- Lehr, Dirk, Andreas Hillert, Edgar Schmitz, and Nadia nomic Issues 36(1):21–33. Sosnowsky. 2008. “Assessing Depressive Disorders Minnotte, Krista Lynn, Michael C. Minnotte, Daphne E. Using the Center For Epidemiologic Studies-Depres- Pedersen, Susan E. Mannon, and Gary Kiger. 2010. sion Scale (CES-D) and State-Trait Depression “His and Her Perspectives: Gender Ideology, Yucel and Latshaw 59

Work-to-Family Conflict, and Marital Satisfaction.” Rosenfield, Sarah, and Dena Smith. 2010. “Gender and Sex Roles 63:425–38. Mental Health” Pp. 256–67 in A Handbook for the Ngo, Hang-Yue, and Siu-Yun Lui. 1999. “Gender Dif- Study of Mental Health: Social Contexts, Theories, ferences in Outcomes of Work-Family Conflict: and Systems. 2nd ed., edited by T. Scheid and The Case of Hong Kong Managers.” Sociological T. Brown. New York: Cambridge University Press. Focus 32(3):303–16. Rosenthal, Carolyn. 1985. “Kin Keeping in the Familial Niles, Spencer G., and Gary E. Goodnough. 1996. Division of Labor.” Journal of Marriage and the “Life-Role Salience and Values: A Review of Recent Family 47(4):965–74. Research.” The Career Development Quarterly 45: Sassler, Sharon. 2010. “Partnering across the Life 65–86. Course: Sex, Relationships, and Mate Selection.” Nohe, Christoph, Laurenz L. Meier, Karlheinz Sonntag, Journal of Marriage and Family 72(3):557–75. and Alexandra Michel. 2015. “The Chicken or the Sassler, Sharon, and Amanda J. Miller. 2011. “Waiting Egg? A Meta-analysis of Panel Studies of the Rela- to Be Asked: Gender, Power, and Relationship Pro- tionship between Work-Family Conflict and Strain.” gression among Cohabiting Couples.” Journal of Journal of Applied Psychology 100(2):522–36. Family Issues 32(4):482–506. Nomaguchi, Kei, and David Johnson. 2009. “Change in Sayer, Liana. 2005. “Gender, Time and Inequality: Work-Family Conflict among Employed Parents Trends in Women’s and Men’s Paid Work, Unpaid Between 1977 and 1997.” Journal of Marriage and Work and Free Time.” Social Forces 84(1): Family 71(1):15–32. 285–303. Pearlin, Leonard. 1999. “The Stress Process Revisited: Schaufeli, Wilmar B., and Toon W. Taris. 2014. “A Crit- Reflections on Concepts and Their Interrelation- ical Review of the Job Demands-Resources Model: ships.” Pp. 395–415 in The Handbook of the Sociol- Implications for Improving Work and Health.” Pp. ogy of Mental Health, edited by C. Aneshensel and 43–68 in Bridging Occupational, Organizational J. Phelan. New York: Kluwer. and Public Health, edited by G. Bauer and O. Ha¨m- Pearlin, Leonard I., and Joyce S. Johnson. 1977. “Marital mig. Dordrecht: Springer. Status, Life Strains, and Depression.” American Schaufeli, Wilmar. 2017. “Applying the Job Demands- Sociological Review 42:704–15. Resources Model.” Organizational Dynamics 2(46): Pedersen, Daphne E., and Krista Lynn Minnotte. 2012. 120–32. “Self- and Spouse-reported Work-Family Conflict Schieman, Scott, and Paul Glavin. 2008. “Trouble at the and Dual-earners’ Job Satisfaction.” Marriage & Border?: Gender, Flexibility at Work, and the Family Review 48(3):272–92. Work-Home interface.” Social Problems 55(4): Pencavel, John. 2015. “The Productivity of Working 590–611. Hours.” The Economic Journal 125:2052–76. Schieman, Scott, Paul Glavin, and Melissa A. Milkie. Perelli-Harris, Brienna, Monika Mynarska, Ann Berring- 2009. “When Work Interferes with Life: Work- ton, Caroline Berghammer, Ann Evans, Olga Isu- Nonwork Interference and the Influence of pova, Renske Keizer, andreas Kla¨rner, Trude Lappe- Work-related Demands and Resources.” American ga˚rd, and Daniele Vignoli. 2014. “Towards a New Sociological Review 74(6):966–88. Understanding of Cohabitation: Insights from Focus Schoen, Robert, Stacy J. Rogers, and Paul R. Amato. Group Research across Europe and Australia.” 2006. “Wives’ Employment and Spouses’ Marital Demographic Research 31:1043–78. Happiness: Assessing the Direction of Influence Ridgeway, Cecilia. 2011. Framed by Gender: How Gen- Using Longitudinal Couple Data.” Journal of Family der Inequality Persists in the Modern World. New Issues 27(4):506–28. York: Oxford University Press. Shimazu, Akihito, Kazumi Kubota, Arnold Bakker, Eva Ridgeway, Cecilia L., and Shelley J. Correll. 2004. Demerouti, Kyoko Shimada, and Norito Kawakami. “Motherhood as a Status Characteristic.” Journal of 2013. “Work-to-Family Conflict and Family-to- Social Issues 60(4):683–700. Work Conflict among Japanese Dual-earner Couples Rindfuss, Ronald R., and Audrey VandenHeuvel. 1990. With preschool Children: A Spillover-crossover Per- “Cohabitation: A Precursor to Marriage or an Alter- spective.” Journal of Occupational Health 55(4): native to Being Single?” Population and Develop- 234–43. ment Review 16(4):703–26. Shockley, Kristen M., and Neha Singla. 2011. Roehling, Patricia V., Lorna Hernandez Jarvis, and “Reconsidering Work-Family Interactions and Satis- Heather E. Swope. 2005. “Variations in Negative faction: A Meta-analysis.” Journal of Management Work-Family Spillover among White, Black, and 37(3):861–86. Hispanic American Men and Women: Does Ethnic- Simon, Robin. 1995. “Multiple Roles, Role Meaning, ity Matter?” Journal of Family Issues 26:840–66. and Mental Health.” Journal of Health and Social Rogelberg, Steven. 2007. Encyclopedia of Industrial and Behavior 36(2):182–94. Organizational Psychology. Vol. 2. Thousand Oaks, Simon, Robin W. 1998. “Assessing Sex Differences in CA: Sage Publications. Vulnerability among Employed Parents: The 60 Society and Mental Health 10(1)

Importance of Marital Status.” Journal of Health and Waismel-Manor, Ronit, Asaf Levanon, and Pamela S. Social Behavior 39(1):38–54. Tolbert. 2016. “The Impact of Family Economic Smock, Pamela J., and Wendy D. Manning. 2004. Structure on Dual-earners’ Career and Family Satis- “Living Together Unmarried in the United States: faction.” Sex Roles 75(7–8):349–62. Demographic Perspectives and Implications for Weigl, Matthias, Nicole Stab, Isabel Herms, Peter Family Policy.” Law & Policy 26(1):87–117. Angerer, Winfried Hacker, and Ju¨rgen Glaser. Soons, Judith, Matthijs Kalmijn, and Jay Teachman. 2016. “The Associations of Supervisor Support and 2009. “Is Marriage More Than Cohabitation? Well- Work Overload with Burnout and Depression: A being Differences in 30 European Countries.” Jour- Cross-sectional Study in Two Nursing Settings.” nal of Marriage and Family 71(5):1141–57. Journal of Advanced Nursing 72(8):1774–88. Spaderna, Heike, Stefan C. Schmukle, and Heinz Walter Westman, Mina. 2001. “Stress and Strain Crossover.” Krohne. 2002. “Report about the German Adaptation Human Relations 54:717–51. of the State-Trait Depression Scales (STDS).” Diag- Westman, Mina. 2006. “Models of Work-Family interac- nostica 48(2):80–89. tions: Stress and Strain Crossover.” Pp. 498–522 in Stanley, Scott M., Sarah W. Whitton, and Howard J. International Encyclopedia of Organizational Markman. 2004. “Maybe I Do: Interpersonal Com- Behavior, edited by R. K. Suri. New Delhi: Pentagon mitment and Premarital or Nonmarital Cohabita- Press. tion.” Journal of Family Issues 25(4):496–519. Wethington, Elaine. 2000. “Contagion of Stress.” Stevens, Daphne, Gary Kiger, and Pamela J. Riley. 2001. Advances in Group Processes 17:227–51. “Working Hard and Hardly Working: Domestic Williams, Joan. 2001. Unbending Gender: Why Work Labor and Marital Satisfaction among Dual-earner and Family Conflict and What to Do about It. New Couples.” Journal of Marriage and Family 63: York: Oxford University Press. 514–26. Winslow, Sarah. 2005. “Work-Family Conflict, Gender, Symoens, Sara, and Piet Bracke. 2015. “Work-Family and Parenthood, 1977–1997.” Journal of Family Conflict and Mental Health in Newlywed and Issues 26:727–55. Recently Cohabiting Couples: A Couple Perspec- Yabiku, Scott T., and Constance T. Gager. 2009. “Sexual tive.” Health Sociology Review 24(1):48–63. Frequency and the Stability of Marital and Cohabit- Tho¨nnissen, Carolin, Barbara Wilhelm, Philipp Alt, Ste- ing Unions.” Journal of Marriage and Family 71: fan Fiedrich, and Sabine Walper. 2016. “Scales and 983–100. instruments Manual Waves 1 to 8.” Retrieved Octo- Young, Marisa. 2015. “Work-Family Conflict in Con- ber 15, 2018 (http://www.pairfam.de/fileadmin/user_ text: The Impact of Structural and Perceived Neigh- upload/redakteur/publis/Dokumentation/Manuals/Sca borhood Disadvantage on Work-Family Conflict.” les_Manual_pairfam_8.0.pdf). Social Science Research 50:311–27. Van Yperen, Nico W., and Mariet Hagedoorn. 2003. “Do Young, Marisa, Scott Schieman, and Melissa A. Milkie. High Job Demands Increase Intrinsic Motivation or 2014. “Spouse’s Work-to-Family Conflict, Family Fatigue or Both? The Role of Job Control and Job Stressors, and Mental Health among Dual-earner Social Support.” The Academy of Management Jour- Mothers and Fathers.” Society and Mental Health nal 46(3):339–48. 4(1):1–20. Vinokur, Amiram D., Penny F. Pierce, and Catherine L. Yucel, Deniz. 2017a. “The Dyadic Nature of Relation- Buck. 1999. “Work-Family Conflicts of Women in ships: Relationship Satisfaction among Married and the Air Force: Their Influence on Mental Health Cohabiting Couples.” Applied Research Quality and Functioning.” Journal of Organizational Behav- Life 13(1):1–22. ior 20:865–78. Yucel, Deniz. 2017b. “Work-Family Balance and Mari- Voydanoff, Patricia. 2002. “Linkages between the tal Satisfaction: The Mediating Effects of Mental and Work-Family interface and Work, Family, and Indi- Physical Health.” Society and Mental Health 7(3): vidual Outcomes.” Journal of Family Issues 23(1): 175–95. 138–64. Yucel, Deniz, and Wen Fan. 2018. “Work-Family Con- Voydanoff, Patricia. 2005. “Social Integration, Work- flict and Well-being among Married Couples Revis- Family Conflict and Facilitation, and Job and Marital ited: A Longitudinal and Dyadic Approach.” Unpub- Quality.” Journal of Marriage and Family 67: lished manuscript. 666–79. Yucel, Deniz, and Margaret A. Gassanov. 2010. Voydanoff, Patricia. 2007. “Work, Family, and Commu- “Exploring Actor and Partner Correlates of Sexual nity: Exploring Interconnections.” Journal of Mar- Satisfaction among Married Couples.” Social Sci- riage and Family 69:894–96. ence Research 39(5):725–38.