Gender-specific patterns and determinants of spillover between work and family: The role of partner support in dual-earner couples

Objective: The study investigates how partner support affects different types of work-tofamily and family-to-work conflicts in dual-earner couples divided by gender and parenthood. Background: In Germany, as in other Western Countries, interrole conflicts between work and family increase, especially within dual-earner couples. Only few studies focused on the effects of partner support on different types of these conflicts. Method: We use longitudinal data deriving from waves 6 to 10 of the German Family Panel (pairfam) to uncover the extent to which the perception of having a supportive partner reduces timeand strain-based work-to-family and family-to-work conflicts. We conduct longitudinal structural equation models based on information of 1,252 persons, which are full-time employed and live in a dual-earner relationship. Results: Whereas for men partner support helps reduce stress-based work-to-family conflicts, for women perceived partner support is not beneficial. Within a subsample of parents, the experience of work-to-family conflicts is likely irrespective of partner support. Overall, women’s family-to-work conflicts appear to be reduced by their partners’ support whereas for men this detrimental effect only applies in the case of stress-based family-towork conflicts. Conclusion: To sum up the findings, the differences for men and women in the effect of partner support on different types of interrole conflicts indicate a still existing impact of traditional gender norms that connect femininity to house work and masculinity to employed work.


Introduction
In many Western countries, the once predominated male bread-winner model has given way to norms of dual-earner households and the desire for steady female employment and careers has spread across societies (Berghammer & Verwiebe 2015;Blossfeld & Drobnič 2001). Due to these changes, both men and women often face problems due to conflicting role expectations (Grunow 2016;Martínez et al. 2011) and difficulties concerning the feasibility of work-family compatibility (Steiner & Krings 2016). Dual-earner couples face particular challenges in the coordination of everyday life (Moen 2003), including the task of maintaining their relationship as a couple (Voorpostel, van der Lippe & Gershuny 2010), especially when work hours of both partners exceed one regular full-time job significantly. Additionally, theories of individualization emphasize the value of selfdevelopment in one's personal life and career, for which the couple relationship must leave additional space (Beck & Beck-Gernsheim 2002). These complex challenges become even more demanding for households with children (Flood & Genadek 2016).
On the one hand, growing attention has been paid to the spillover between both work and family life in sociological and socio-psychological research. Drawing on the initial work of Greenhaus and Beutell (1985) and its advancements (Ashforth, Kreiner & Fugate 2000;Demerouti 2012), work-to-family-conflicts (WFC) and family-to-work-conflicts (FWC) have dominated the research agenda (Bianchi & Milkie 2010;Steiner & Krings 2016).
On the other hand, some authors have addressed the way couple relationships and especially social support within couples can affect different dimensions of life (Kuhn et al. 2018). Feeney and Collins (2015), for instance, discuss the role of partner support in coping with adversity and seizing opportunities in life. Although for many people the partner can be seen as the most important significant other (Lee & Szinovacz 2016;Teo, Choi & Valenstein 2013), there has been little systematic theoretical and empirical analysis of the effects of partner support on spillover in the family-work context. In dual-earner couples, time resources are peculiarly scarce and strain-or time-based work-to-family and family-to-work conflicts can affect both partners (Yucel & Latshaw 2020). Children in the household, gender norms, and the couple's relative resource distribution are additional factors which on the one hand render partner support important and on the other hand add a certain complexity to the underlying mechanisms of supportive structures.
This study contributes to the existing body of research by investigating the genderspecific role played by perceived partners' support in reducing patterns of negative spillover between work and family demands. To this end, we firstly discuss the theoretical and empirical state of research in order to develop hypotheses on the role partner support plays within dual-earner couples in reducing different types of spillover between family and work. These hypotheses are based on arguments for different impact levels of men's and women's perception of partner support and gendered work and family norms. In the second step, we test our assumptions using structural equation models for longitudinal data based on waves 6 to 10 of the Panel Analysis of Intimate Relationships and Family Dynamics (pairfam, release 9.1).
Differences between the partner support estimates using the whole sample, divided by gender and divided by gender for the subgroup of parents are compared. Using both, our theoretical and empirical results, we draw a conclusion on the effects of partner support on men's and women's interrole conflicts, further discuss the role of gendered norms on work, family and parenting, and provide an outlook on future perspectives in this field of family research.

Concepts
Interactions between family and work domain have been the subject of research for nearly half a century. Greenhouse and Beutell (1985) were the first to develop a widely applied approach on the interaction of both domains, structuring the life of most people in society. The description of interrole conflicts arising from the specific needs and demands of both roles forms the theoretical foundation to dozens of studies, especially in the case of double-income couples (Byron 2005). These conflicts can be divided into the categories 'time, strain, and behaviour', displaying the variety of possible spillover patterns (Carlson, Kacmar & Williams 2000).
A vital starting point is that everybody has a limited supply of two important resources -time and energy. Managing these resources is the most challenging for dual-earner couples and families who have to juggle two life roles: work and family (Schooreel & Verbruggen 2016). Partners' resources are not only needed in each life domain; partners' resources in these two life domains also compete with each other, since men and women are faced with "the challenge of three jobs -two at work and one at home" (Moen & Yu 2000: 291). A stronger involvement in one life domain (e.g. in the family domain) is therefore likely to tie up resources which are needed in the other life domain (e.g. in the work domain), resulting in conflicts. One such conflict that may arise when one domain begins to interfere with the other is time-based conflict. If time constraints in one life domain are not met, this complicates participation in another role. In line, many work hours were found to be one major antecedent of WFC (Michel et al. 2011) and interrole conflicts arise mainly in context of full-time work and extra hours. Strain-based conflicts arise when strain in one life domain interferes with participation in another. To provide two examples: productivity in the workplace is likely to decrease due to conflicts with one's partner or children. Also, strain at work may impede proper family role task performance.
However, spillover can also have positive effects, which are discussed using keywords such as 'work-family-enrichment' (Greenhaus & Powell 2006). Positive spillover effects include, broadly speaking, learning new skills and expanding one's knowledge, developing positive emotional attitudes and psychosocial resources, and improving time management skills (Carlson et al. 2006).
In summary, spillover, positive or negative, is bidirectional and can occur from family to work and from work to family, showing the strong interrelatedness of these two life domains. Spillover effects are not equally likely, however, because the work and family domain and their borders are not equally permeable (Ashforth, Kreiner & Fugate 2000). More specifically, family boundaries seem to be weaker than work boundaries, and thus aspects of work are more likely to interfere with the family domain than vice versa (Frone, Russell & Cooper 1992). However, as becomes tangible in the notion of "ideal" workers and "ideal" parents (Lott & Klenner 2016), a couple's boundary management becomes more demanding and boundaries are becoming increasingly blurred. Moreover, how couples manage their multiple roles and how they create or maintain boundaries between work and family life changes across the life course. In all of these considerations, gender differences in demands and resources should not be overlooked.

Gender-specific effects
The compatibility of two careers, love and work and family and work, is often seen as difficult to harmonise, particularly when children are involved (Solga & Wimbauer 2005). One may assume that time-based WFCs and especially FWCs are more often experienced by women and some studies already provide support for this (Voydanoff 2004). Such conflicts and arrangements with conflictual settings are often explained via theories of exchange and bargaining theory, adopted from economics (Lundberg & Pollak 1996). In short, negotiation within couples is led by marital power, which is determined by socioeconomic resources (Bittman et al. 2003). The partner with the higher income or better career prospects has the better negotiation position to advance his/her interests to a certain extent. Those interests are typically interpreted as the chance to focus more on one's career, while the 'outgunned' partner has to split him-or herself between his or her own employment schedules and household and care tasks. In Germany as in many other countries, men have an income and job prestige advantage over women resulting in a higher negotiation power of male partners. In line, the male breadwinner/female part-time career model is a wide-spread way of (re)distributing responsibilities for paid and unpaid work within couples. Many women arrange, at least temporarily, a reduction in paid working time as an adaption to a conflictual spillover of work and family (Damaske & Frech 2016), while men, even if they have a lower socioeconomic status than their partner, much less frequently reduce work hours (Berghammer 2014). Moreover, in nearly every couple's socioeconomic combination women spend more time on housework than their male partners (Berghammer 2014;Dechant, Rost & Schulz 2014;Baxter, Hewitt & Haynes 2008). Despite the spread of gender egalitarian views (Nitsche & Grunow 2016;Pampel 2011), the amount of unpaid extra work does not change, with women doing the lion's share of unpaid work at home, even in full-time employed dual-earner couples (Lyonette & Crompton 2015). From this perspective, decisions on the division of labour between partners are to a lesser degree attributable to negotiation processes. Couples are 'doing gender' -leaving women with a more demanding time schedule.

Gendered parenthood effects
Children play a significant role in processes of re-traditionalization throughout the life course. Immediately after the birth of a child, women increase their share of unpaid house and care work (Baxter, Hewitt & Haynes 2008;Nitsche & Grunow 2016) and reduce their hours in paid work in the medium-/ long-term (Berghammer 2014). This correlation is not found for men. Some findings, instead, indicate that male partners increase their work hours after the transition to fatherhood or at least do not decrease them (Bünning & Pollmann-Schult 2016). For both, women and men, children in the household increase the workload at home (Jacobs & Gerson 2004;Künzler et al. 2001). Indeed, the sheer increase in family-related work could itself be a catalyst to work-family spillover; caring for children is associated with strong normative obligations. In addition to possible timebased conflict, norms of motherhood or 'intensive mothering' (Guendouzi 2006;Hays 1996) and 'new fathers' (Duyvendak & Stavenuiter 2004;Hook & Wolfe 2012) remain in conflict with the 'ideal worker norm' (Williams 2001). While for women the social pressure to act in accordance with the former gender norms is higher than for men, the ideal worker norm is still quite clearly connoted as masculine (Collinson & Hearn 2005). The feeling of not having enough time for the child makes mothers in particular feel guilty (Borelli et al. 2017;Guendouzi 2006), but fathers with non-traditional values experience the same feelings (Martínez et al. 2011). The converse, feeling guilty because of a lack of time or energy for the job is also conceivable.
Due to these gender differences, it can be assumed that mothers tend to perceive higher degrees of WFC, while FWC is in fathers (Hill 2005). Several researchers have argued this way (Wayne, Musisca & Fleeson 2004) emphasizing the importance of gender differences in life circumstances (Shockley et al. 2017). A deviation from the ideal worker norm often goes hand in hand with disadvantages for women (Benard & Correll 2010) and men (Coltrane et al. 2013). For mothers, working more hours is in direct contradiction to expectations of intensive mothering. Extensive mothering practices are fraught with many hurdles and conflicts for employed mothers, too. Fathers can have problems in fulfilling new demands of active fathering, which to a large extent might be dependent on their personal interpretation of the role of fathers.

The role of perceived partner support
Social support is a multifaceted concept and has no universal definition (Diewald 1991). This support is typically divided into two subtypes: emotional and instrumental support (Adams, King & King 1996;Perrewé & Carlson 2002). While emotional support becomes clear in the degree to which one partner shows understanding for the other partner's issues, instrumental support involves the provision of specific and concrete assistance. Instrumental support is often measured by the amount of housework done by the partner, while measures of emotional support typically capture a partner's empathy or emotional responsiveness to one's personal affairs (Busch, Bröckel & Golsch 2014). Beside these direct effects of emotional and instrumental support, both types can also have indirect effects, which go beyond the concrete situation of providing support and can cause sustainable positive outcomes for recipients of partner support (Shumaker & Brownell 1984).
The perception of social support provision may differ from the actual availability of social support as well as the actual received support (Haber et al. 2007). What is important to note in the present context is that perceived partner support is assumed to have a greater impact on an individual's experience in dealing with WFC and FWC than actual provided support and that individuals reporting higher levels of social support are also more likely to believe that adequate support resources are available when needed (Lin & Wu 2014), thus reducing future stress and strain experiences in the work-family context. This holds not only for perceived provision of help and assistance, but also for the belief that someone is available who is willing to listen with empathy. A second important aspect is the likely increase in self-confidence as well as practical skills in handling a WFC or FWC. Put differently, a lack of support or unhelpful support may result in increased levels of future stress and strain (Overall, Fletcher & Simpson 2010). Positive effects of supportive structures, therefore can be seen as a holistic concept (Adams, Golsch & Maiwald 2020).
The potential for reverse causality has to be kept in mind, however. Inability to effectively manage both types of conflict (WFC and FWC) may not only be rooted in one's own personality, knowledge, skills and abilities. A person's failure to deal with such conflicts may also be due to lower perceptions of partner support in the long termirrespective of whether there is a decrease in the quality and frequency of actual provided support. Either way, unsolved conflicts in the work-family context may challenge a couple relationship and its inherent sources of partner support (Amstad et al. 2011).
Partner support plays a vital role in couples, having positive effects not only individually for the two partners but also for the couple itself, since providing this support establishes a reliable and reciprocal relationship (Shumaker & Brownell 1984). Difficulties in dual-earner couples may not only arise in the establishment of a functioning day-to-day schedule, but also in the harmonization of the individual 'needs' and 'wants' of both partners. Lundberg and Pollak (1996) emphasize the benefit of cooperative negotiations over non-cooperative behaviour in a couple relationship. Failure to establish a modus of cooperation can be seen as the failure of a couple relationship (Adams, Golsch & Maiwald 2020;Maiwald 2013). A general modus of cooperation can be considered an indirect effect of reliable support provision in couple relationships. Because of the holistic effect of a modus of cooperation in a couple relationship, it can be assumed that emotional and instrumental support are strongly connected and perceived support may help to reduce time-based and strain-based conflicts.
Despite increased research on the work-family interface, the role of perceived partner support is not yet sufficiently understood, particularly in the context of Germany. Prior research has examined how social support helps couples cope with WFC and FWC and to what extent this differs regarding the type of support, the circumstances in which couples find themselves, and the type of conflict. One of the major shortcomings of previous studies is that a distinction is made between support at work and family support in general (Michel et al. 2010) while relatively few studies focus on partner support in particular. Taken together, the results of these studies show some inconsistencies, which may be due to varying definitions and operationalizations or the particular country context, but most findings reveal partner support has a reducing effect of partner support on FWC and, to a lesser extent, on WFC.
Instrumental and emotional support showed a reducing effect on FWC in different cultural settings (Aycan & Eskin 2005;Burke & Greenglass 1999;Minnotte & Minnotte 2018;Selvarajan, Cloninger & Singh 2013) and in a meta-analysis (Michel et al. 2011). Recent findings, instead, only reveal reducing effects of emotional partner support on FWC, but no effect of instrumental support (Uysal Irak, Kalkışım & Yıldırım 2020). Inconsistent with these findings, in Hong Kong support in dual-earner couples only showed an indirect effect on FWC by moderating negative outcomes of work overload (Aryee et al. 1999). Inconsistencies can also be found for WFC. While some results indicate a reducing effect of emotional support (Uysal Irak, Kalkışım & Yıldırım 2020), others show a reducing effect of instrumental support on WFC (Selvarajan, Cloninger & Singh 2013). Aryee and colleagues (1999) found an effect of partner support solely on WFC and not on FWC, while Michel and colleagues (2011) report the converse.
Additionally, some studies do not focus on work-family conflicts in particular, but on strongly related predictors of WFC and FWC. Low levels of partner support were detected as a cause of work-related stress for women and men, while high levels of support were associated with a reduction in work-related stress and emotional exhaustion (Pluut et al. 2018). Additionally, partner support was found to buffer the effect of parental overload as an antecedent to FWC (Aryee et al. 1999). These effects also point to partner support as an expression of a modus of cooperation -a positive environment in the relationship entangling what Cobb and Jones (1984) term indirect effects of social support. In summary, it can be assumed that: H1: Perceived partner support reduces FWC and WFC.
The short literature review, on the other hand, reveals some inconsistent findings in the effects of partner support (and other coping activities) regarding different types of interrole conflicts. Many studies do not focus on one particular coping mechanism, but the different types of WFC and FWC may make it worth to look at coping mechanisms separately. Although we highlight the importance of partner support, it has to be stated that the area of influence is not infinitely large. In particular, WFC is mainly caused by the employment environment (Byron 2005). Perceived co-worker and supervisor support emanate from the life area in which this stress arises. This social support in the workplace can be an effective resource for reducing stress of time and is thus more likely to decrease WFC than partner support. An exception to this, although not the focus of the present paper, is dual-earner couples in which both partners work in related professions and are thus in a good position to provide mutual emotional and instrumental support. Returning to a more general consideration of couples, partner support may still be relevant for WFC, particularly when workplace support is lacking. While partner support can help people cope with strain-based work-to-family conflicts, its influence on time-based conflicts might be negligible, however. If there is not enough time to meet one's demand for time with one's partner, children, or friends due to work hours or unfriendly work schedules, the effect of a supportive partner is limited. In contrast, stress emanating from the family environment is under the direct influence of the partner and his or her support, be it emotional or instrumental, is expected to be stress-buffering. In summary, it can be assumed that: H2a: The effect of perceived partner support is stronger for FWC than for WFC.
H2b: Perceived partner support has no effect on time-based WFC.
However, this may not hold true for men and women to the same extent. The social support literature already provides insights into gender differences in partnerships which are often discussed in the context of the 'support gap hypothesis'. Here, women are assumed to receive less (helpful) support from their partner than men and this is often traced back to gender differences in personality or to gender-role expectations. Given the argument that perceptions of partner support are most relevant in the context of this study, it is important to stress that gender differences in either observed or perceived partner support only occur on particular aspects of support provision (Neff & Karney 2005;Verhofstadt, Buysse & Ickes 2007).
Although, empirical studies in the field are rather scarce, these differences underneath the surface of partner support might be relevant in the context of managing interrole conflicts. Results of Neff & Karney (2005), for example, show that female support is more relevant in situations of greater stress. More general, classical approaches indicate that women attach greater importance to meeting a partner's needs and strengthening their partner's wellbeing than men do (Goffman 1977;Kohen 1981). Women are also more likely to actively seek support from their partner, whereas men are more likely to try to get along without support. The background to these phenomena is formed by gender-specific socialization processes (Jensen, Rauer & Volling 2013). Additionally, women may be more likely to face conflicts than men and thus they are also more likely to be in greater need of partner support. Many studies reveal gender differences in the experience of work-family conflicts as well as the received social support and its effects. While empirical evidence shows ambiguous results (Pluut et al. 2018), some studies assume that partner support has greater impact for women in the context of work and career (Busch, Bröckel & Golsch 2014). Additionally, many women perform their roles in accordance with traditional gender norms and spend significantly more time on household duties and parenting (Dechant, Rost & Schulz 2014;Lyonette & Crompton 2015). Therefore, partner support may be more relevant for women than for men with regard to reducing time-and strain-based stress. In summary, it can be assumed that: H3: Perceived partner support has a stronger reducing effect on FWC and WFC for women than for men.
Finally, the role of parenthood must also be taken into consideration but to date there are relatively few empirical studies in this area. In dual-earner couples with preschool children, familial obligations increase. These additional obligations are time and energy demanding (Jacobs & Gerson 2004;Künzler et al. 2001) -resources that are already in short supply for dual-earner couples. As in childless couples, partner support can play a buffering or preventive role in work-family conflicts. What is special here is the role of the second parent as a support provider. The double role of intimate other and co-parent makes partner support even more necessary to avoid or buffer work-family conflicts. Due to stronger norm conflicts, especially working mothers seem to be dependent on partner support: H4: Perceived partner support has a greater effect on FWC among mothers than fathers.

Data
We use data from the Panel Analysis of Intimate Relationships and Family Dynamics (pairfam, release 9.1). The pairfam is a German multi-actor panel study following a cohort approach (birth cohorts 1971-73, 1981-83, and 1991-93) and focusing on partnerships and family processes, with nine waves available to date (Brüderl et al. 2018). Within these waves, the panel provides information on family, educational, and labour market histories as well as a broad range of measurements regarding the participant's relationship life, work life and the interweaving of both dimensions. The anchor's current partners were interviewed separately using a slightly shorter questionnaire (Huinink et al. 2011).

Sample
Items that measure WFC and FWC are surveyed in every second wave from wave 6 to 10. Information on partner support is only provided in the waves 1, 3, 5, 7, and 9. For our analyses we pooled wave 6 to 10. Respondents of the estimation sample were surveyed three times (either waves 6 to 8 or waves 8 to 10), in order to facilitate a before and after comparison of WFC and FWC. Information for respondents participating in all of the waves was only taken from the later three. Therefore, most information is obtained from waves 8 to 10 (n = 934) rather than from waves 6 to 8 (n = 318). Since we are studying people in a dual-earner relationship, we limited our sample to respondents living in a heterosexual relationship with both partners employed in three consecutive waves. To contribute to the argument of limited time resources, we limit the sample to couples in which anchor persons work for at least 30 hours per week and their partners for at least 20 hours (n = 1,252). These work hours represent 75 and 50 percent of a regular full-time employment in Germany, meaning that only couples are included who have significantly more work hours than one regular full-time job (40 hours of regular work). Socioeconomic characteristics of the sample are described in Table 1.

Work-to-family conflicts and family-to-work conflicts
We use four items, based on the multidimensional measure introduced by Carlson and colleagues (2000), to identify conflicts from work to family and from family to work. FWCs are measured using four statements; two for strain-based conflicts (SFWC): "Have problems to focus on my work" and "Conflicts in private life impair my job performance"; two for time-based conflicts (TFWC) "Time for partner etc. restrains me from engaging more in work" and "Lacking time for my job due to private appointments". The same is true for WFC, where strain-based conflicts (SWFC) are measured by "Difficult to relax at home after strains of the work" and "Often have to think on my work"; while time-based conflicts (TWFC) were measured by the statements "Personal life gets short" and "Job keeps me from doing things with partner etc.". For both FWC and WFC, the statements were answered using the same 5-point scale, ranging from totally disagree (1) to totally agree (5). All single items have substantial loading on the respective latent measure of WFC and FWC (Appendix 1).

Perceived partner support
Emotional and instrumental partner support is captured via the following three single items: "Partner lets me know that he/she understands me", "Partner listens, gives chance to express myself", and "Partner supports me when I have a problem" on a 5-point response scale ranging from never (1) to always (5). All single items show have substantial loading on the respective latent measure of perceptions of partner support (Appendix 1). In the analyses, we do not additionally differ between emotional and instrumental support. As argued above, both types of partner support are strongly connected and contribute to the perception of partner support and its implications.

Further variables
Gender is measured bivariate (0 = female; 1 = male). We controlled for children in the household with the scale "child under the age of 1", "child aged 1 to 3", "child aged 3 to 7", "child aged 7 to 14", and "no child" corresponding to the different levels of care and attention required by the child. In further group analyses, both, gender and the presence of children under 15 (0 = no; 1 = yes) are used as distinction characteristics. We control for actual weekly work hours of the anchor and the partner as metric variables. Following the life and career stage perspective, we control for the birth cohort of employees as a proxy for their career stage (1 = 1991-93; 2 = 1981-83; 3 = 1971-73). We include the educational level in relation to the level of the partner (1 = anchor holds higher educational degree; 2 = same educational level; 3 = partner holds higher educational degree) to control for potential bargaining power within the relationship. Additionally, we control for the place of residence (0 = East; 1 = West) and relationship duration in months.

Figure 1:
General model of analysis

Method
We use latent variable structural equation models to investigate the effect of partner support on different types of work-family conflicts. Structural equation models are able to combine techniques of (longitudinal) regression analyses and confirmatory factor analyses (CFA) and have an advantage when it comes to the inclusion of latent variables (in our case both, partner support and different types of WFC and FWC) by conducting a CFA in the main analysis. Figure 1 shows the general model without control variables. Because of the time-lagged information of both partner support (every uneven wave) and WFC and FWC (even waves from wave 6 on), we conduct the effect of partner support at 1 on the different types of conflicts at 1+1 and add an autoregressive stability path from these conflicts at 1−1 to 1+1 in order to control for the level of conflict before the measure point of partner support (Acock 2013: 113 ff.). The number of missing values among the main variables of the model is low. We excluded 4 persons from the analysis who had missing values either for perceived social support (n=1) or types of work-family conflicts in one wave (n=3). By using full information maximum-likelihood estimation (FIML), we did not exclude persons with missing values on control variables. FIML does not impute missing values, but estimates the most likely covariance and standard errors for variables with missing values. This method, therefore, clearly has an advantage over classical methods like listwise deletion, since the results are less biased and comparable to those obtained using multiple imputation (Johnson & Young 2011). Table 2 shows descriptive statistics and bivariate correlations of the main variables. People have a high average value of perceived support (mean=3.98, SD=.67). Conflicts from family to work are relatively uncommon (SFWC: mean = 1.67, SD = .72; TFWC: mean = 1.5, SD = .64); however, WFC was reported more often (SWFC: mean = 2.24, SD = .94; TWFC: mean = 2.68, SD = 1.11). 1 There are no significant differences in the distribution across gender, neither for perceived support, nor for different types of WFC and FWC conflicts. People living with a child up to age 15 in the household reported slightly lower values of perceived partner support and slightly higher values of TFWC, but no differences were found for the other types of conflicts.

Descriptives and bivariate correlations
Simple bivariate correlations indicate that partner support is negatively associated with strain-based (r = -.16, p < .000) and time-based FWC (r = -.14, p < .000). Lower negative correlations are found for strain-based WFC (r = -.07, p = .021), while no bivariate association can be found between partner support and time-based WFC (r = -.02, p = .534). All different types of work-family conflicts are positively correlated. Not surprisingly, the highest values can be found for the associations between the same types of conflict across both time points (r > .44, p < .000) and for the association between strain-and time-based FWC respectively WFC.

Multivariate models
The multivariate analyses contain four models each for the whole sample, divided by gender and divided by having at least one child in the household. For the groupcomparison models, variances of one observed partner-support variable and one conflict variable at 1−1 were set to zero to compute model fit indices. All models allow correlation 1 Values are taken from time point 1−1 and do not differ between the waves.
of the covariances of the conflict-variables at different time-points as well as the observed partner support scores. Regarding the gender-sensitive models, there are no significant gender differences in the loadings of emotional and instrumental support on the latent variable.  In an earlier version, we also controlled for the birth cohort. The effect was insignificant for all models and had no effect on the results, but caused problems to compute model fits for single models. Therefore, we excluded it for all models to grant higher comparability between the models.

Full models
All four models show a good model fit (Table 3). The estimates reveal a significant reducing effect of partner support on strain-based and time-based FWC (for both β = -.13, p > .000). These results and the high correlation indicate that strain-and time-based FWC can be seen as one concept and that perceived partner support reduces these conflicts, irrespective of the form of expression. In contrast, both estimates from partner support on SWFC and TWFC are not significant (SWFC: β = -.04, p = .196; TWFC: β = .00, p = .993).
Regarding hypotheses H1, H2a and H2b, perceived partner support only revealed a reliable reduction of FWC, while the analyses did not show any effect on TWFC and a non-significant reducing effect on SWFC.

Figure 2:
Standardized effects of partner support on strain and time-related conflicts divided by gender

Models divided by gender
In the next step, we estimated the effect of perceived partner support on the different types of work-family conflicts separately for men (n = 608) and women (n = 644). Again, all models show a satisfying model fit (Appendix 2). In line with hypothesis H3, effect sizes and significance levels of a reduction effect of partner support on SFWC and TFWC were higher for women than for men (SFWC: = -.11, p = .016; = -.16, p < .000; TFWC: = -.02, p = .658; = -.09, p = .045), although wald tests do not suggest that these differences are significant. In contrast to our hypothesis, the results of partner support on SWFC are unexpected and quite surprising. While for women perceived support of the partner did not reveal any effect ( = .01, p < .790), perceived support reduced SWFC significantly for men ( = -.13, p = .010). The wald test shows a significant difference (p = .047). Differentiation by gender did not change the outcome for perceived partner support on TWFC. It can be concluded that the assumption of H3 holds true for FWC, although the differences for men and women are not disproportionately high. Surprisingly, the gendered effect for SWFC was contrary to these expectations.

Figure 3:
Standardized effects of partner support on strain and time-related conflicts of parents divided by gender

Models divided by mothers and fathers
Finally, we investigated the effect of perceived partner support on FWC and WFC for persons living with a child up to age 15 at 1 in the household, divided by fathers (n = 304) and mothers (n = 265). The stress-buffering effect of perceived partner support on SFWC was found to be stronger among mothers ( ℎ = -.26, p < .000) compared to all women and fathers ( ℎ = -.13, p = .040), who do not display differences compared to all men. This underlines the assumptions of hypothesis H4. Also in line with our expectations, TFWC was reduced significantly by partner support for mothers ( ℎ = -.16, p = .028), while only a weak and insignificant effect was found for fathers ( ℎ = -.08, p = .254). In the case of WFC, the surprising outcome of a stronger reducing effect of partner support on strain-based conflicts for men than for women also holds true for fathers ( ℎ = -.08, p = .192) and mothers ( ℎ = .03, p = .614). While the effect for fathers was found to be insignificant, no effect was observed on SWFC for mothers or TWFC for both mothers and fathers.

Conclusion
The purpose of this study was to identify the role of perceived partner support in work-tofamily and family-to-work conflicts of dual-earner couples. These conflicts are additionally divided into different sources of compatibility problems -time-based and stress-basedwith theoretical implications for their relationship to the effectiveness of partner support. The theoretical discussion suggested there are important gender differences in the significance of partner support as well as differences for households with children, due to gendered parenting norms and a general increase in workload. Consequently, the empirical part of the analysis distinguished between men and women in households with and without children. To this end, collected data were used to perform structural equation modelling. Partner support significantly reduces the level of family-to-work conflicts. Interestingly, the study cannot find any robust evidence supporting the hypothesis that partner support generally de-escalates work-to-family conflicts. This finding is particularly noteworthy because dual-earner couples are more vulnerable to work-to-family conflicts than vice versa. The evidence presented in this study suggests that the responsibility of implementing strategies to reduce these conflicts lies in the hands of companies and that it is even more important for organizations to create good occupational conditions and to provide a good workplace culture.
Subgroup analyses revealed important gender differences, however. Partner support is associated with lower levels of stress-based work-to-family conflicts in men but has no buffering effect for women. Conversely, perceived partner support decreases women's time-based and stress-based family-to-work conflicts to a higher degree. Here gender roles come into play and seem to be associated with seeking and receiving emotional and instrumental support. Perception of social support may also be connected to gender role expectations. From a sociological perspective on gender, it can be argued that men's workto-family conflicts are relatively well accepted, whereas women's work-to-family conflicts are still more often a source of incomprehension and a women's full-time employment might still be seen as a luxury, especially for mothers. At the same time, normative expectations of participation in housework and child care, in particular, are not yet socially institutionalized to the same extent for men and women. Thus, men seem to receive a greater benefit from a partner who will take care of things at home and shows more understanding of their job stress. The work-to-family interface is then experienced as less conflictual.
Since women still tend to do a greater share of the housework than their partner their need for partner support is often higher but not fulfilled. Additionally, in many cases women receive less support in the workplace than men. Also, women are more likely to judge their own competence and performance against higher standards than men and are less sure that they will meet the requirements of work and family (Correll 2004). Deviating from traditional role expectations thus makes the role of partner support more crucial for women.
Further support for this interpretation has been offered by analyses of couples with children. Again, there is a significant gender difference in the role of partner support in family-to-work conflicts: partner support has the potential to reduce mother's family-towork conflicts and the difference between this and the reduction of father's family-to-work conflicts is higher than for the whole sample. Furthermore, huge gender differences have not been found in the analysis of work-to-family conflicts.
The present study has some limitations. Selective survey drop-out is a well-known complication in the analysis of longitudinal data from couples. Men and women in good working relationships are not only happier but also more likely to remain in the panel than persons in unhappy or conflictual relationships, who are also at higher risk of a separation. This attrition in panel studies may be related to perceived partner support as well as time-and stress-based conflicts, and thus, this is likely to introduce bias into model estimates. A further limitation is that some effects are more likely to emerge within shorter time spans, while spacing between waves in the pairfam study creates a gap of one year for the effect of support. This may introduce further inaccuracies. Due to limited information in the data, the models cannot control for supportive structures at the work place. Future research could also focus on branch specific aspects, although possible differences do not hit the heart of our arguments on the mechanisms between partner support and interrole conflicts. Finally, due to gender role expectations, the perspective of self-perceived support and conflicts may be an additional conduit for biased data.
Notwithstanding these limitations, this study provides an insight into the role played by partner support in reducing men's and women's work-to-family and family-to-work conflicts and thereby encourages further research on gender-specific effects. This study especially underlines three proposals. Future research should directly address the effects of life events (birth of a child, new job demands etc.) on possible changes in partners' support behaviour in order to uncover the causal interconnections among concepts. Closely related to this, in a dyadic analysis one could test the effect of experiencing workto-family and family-to-work conflicts on the provision and perception of social support, with a specific focus on cross-over effects. Lastly, and most importantly, a within-couple perspective, as described in this study, requires a more sophisticated mixed-method approach that allows exploration of dynamic intra-couple processes that cannot be fully understood within the framework of a standardized survey. Note. Source: pairfam, waves 6 to 10, n = 1,252. Standard errors in brackets (). All loadings significant at p < .000. Values of perceived partner support retrieved from a Confirmatory Factor Analysis, values of interrole conflicts retrieved from Full Models (Figure 1). Note. Source: pairfam, waves 6 to 10. Control variables: Actual Work Hours (Anchor and Partner), Age of youngest child in HH (only Figure 2), Place of Residence, and Relationship Duration.