Work-family conflict and partners’ agreement on fertility preferences among dual-earner couples: Does women’s employment status matter?

Objective: This study tests the effects of work-family conflict, in both directions, on partners’ agreement on fertility preferences among dual-earner couples, as well as whether this relationship varies by women’s employment status. Background: Few studies have examined the relationship between work-family conflict and fertility preferences. Given the high percentages of women working part-time in Germany, it is important to investigate the role working women’s employment status plays to further understand this relationship . Method: Using data from 716 dual-earner couples in Wave 10 of the German Family Panel (pairfam), we use dyadic data analysis to test whether work-family conflict impacts one’s own (“actor effects”) and/or one’s partner’s (“partner effects”) reports of agreement on fertility preferences. We also run multi -group analyses to compare whether these effects vary in “full -time dual-earner” versus “modernized male breadwinner” couples . Results: There are significant actor effects for family-to-work conflict in both types of couples, and for work-to-family conflict in modernized male breadwinner couples only. Partner effects for family-to-work conflict exist only among modernized male breadwinner couples. While there are no gender differences in actor or partner effects, results suggest differences in the partner effect (for family-to-work conflict only) between these two couple types. Conclusion: These findings indicate that work-family conflict is associated with greater partner disagreement on fertility preferences and highlight the differential impact incompatible work and family responsibilities have on fertility decisions when women work full-time versus part-time.


Introduction
In recent decades, research has documented a rise in female labor force participation, a reduction in traditional male breadwinner households, and a decline in fertility rates across Western nations (Bianchi & Milkie, 2010;Blossfeld & Drobnic, 2001).As these major demographic changes unfolded, scholars began to examine the myriad ways shifting roles in the workplace impact men's and women's experiences in their homes (Bianchi et al., 2006;Galinsky et al., 2011).In particular, work-family conflict has been a growing issue of concern, as dual-earner couples attempt to simultaneously balance their responsibilities in the competing work and family domains (Adams & Golsch, 2020).Despite intense scholarly interest in this topic, surprisingly little is known about the consequences of work-family conflict on couples' fertility-related decision making.Specifically, it is currently unknown whether work-family conflict is associated with greater fertility-related conflict among partners.Given the declining fertility rate and the high number of women working part-time in Germany, it is also important to investigate employment variations among dual-earner couples to better understand this relationship within the German context.As prior work suggests, despite the introduction of more progressive family policies in recent years, Germany is still marked by traditional gender ideologies (Florean & Engelhardt, 2020;Geist & Brauner-Otto, 2017;Konig & Cesinger, 2015) and a legacy of policies that promote the married, male breadwinner ideal (Fleckenstein & Seeleib-Kaiser, 2011;Klammer & Letablier, 2007).
The concept of work-family conflict refers to a type of inter-role conflict resulting from the competing (and often incompatible) demands of the work and family domains (Greenhaus & Beutell, 1985).Work-family conflict operates in two directions: work-tofamily conflict ("WTFC") and family-to-work conflict ("FTWC").For example, working late can cause tension with one's spouse (i.e., WTFC), and likewise, fighting with a spouse can lead one to feel distracted at work (i.e., FTWC).Prior work suggests that both WTFC and FTWC are related to a wide variety of work, family, and health-related outcomes (see, e.g., Allen et al., 2000;Amstad et al., 2011;Bakker et al., 2008;Frone et al., 1996;Michel et al., 2009;Yucel & Latshaw, 2020).However, few studies have directly examined the relationship between work-family conflict and fertility-related outcomes (Begall & Mills, 2011;Burch, 2020;Liu & Hynes, 2012;Shreffler et al., 2010).Likewise, of the existing studies, most use individual-level data to examine the effects of work-family conflict on fertility-related outcomes, with only one employing couple-level data to evaluate the effects of work-family conflict on birth/adoption outcomes (see Burch, 2020).Moreover, to our knowledge, no existing study has specifically examined the relationship between workfamily conflict and partners' agreement on fertility preferences as we do here.In the following manuscript, we fill this gap in the literature by using couple-level data and dyadic data analysis to assess the relationship between work-family conflict (in both directions) and partners' agreement on fertility preferences among dual-earner couples.We further contribute by assessing whether there are gender differences in either the spillover and/or crossover effects of work-family conflict, as well as whether these effects vary by women's employment status.To accomplish this, we compare (1) "full-time dualearner" couples (Allen & Finkelstein, 2014) where both partners work full-time and (2) "modernized male breadwinner" couples (Buber-Ennser, 2015;Mauerer & Schmidt, 2019) where the male partner works full-time and the female partner works part-time.Finally, we extend the literature by situating this analysis within the larger institutional context of Germany, a country marked by its continued adherence to traditional gender ideologies and values, despite an increase in dual-earner families (Florean & Engelhardt, 2020;Geist & Brauner-Otto, 2017;Konig & Cesinger, 2015).
In doing so, we answer the following five research questions: (1) Does one's own work-family conflict (in two distinct directions --WTFC and FTWC) impact their own reported agreement with their partner (i.e., actor effects) on fertility preferences?, (2) Are there gender differences in these actor effects?, (3) Does one's own work-family conflict (in two distinct directions --WTFC and FTWC) impact their partner's reported agreement with them (i.e., partner effects) on fertility preferences?, (4) Are there gender differences in these partner effects?, and (5) Do actor and/or partner effects vary among full-time dual-earner versus modernized male breadwinner couples?

Work-family conflict and fertility
As Greenhaus and Beutell (1985) explained over three decades ago, work-family conflict is a type of inter-role conflict that occurs when "participation in the work (family) role is made more difficult by virtue of participation in the family (work) role" (p.77).As mentioned, work-family conflict also operates in two directions -WTFC and FTWC.In other words, work conditions and/or demands can lead to conflict in one's role at home, or family conditions and/or demands can lead to conflict in one's role at work.Workfamily conflict, in both directions, has been associated with a number of negative outcomes for individuals and partners across multiple domains (Allen et al., 2000;Amstad et al., 2011;Bakker et al., 2008;Frone et al., 1996;Michel et al., 2009).However, only a handful of existing studies have examined the relationship between work-family conflict and fertility outcomes (Begall & Mills, 2011;Burch, 2020;Liu & Hynes, 2012;Shreffler et al., 2010).Moreover, among these few existing studies, none simultaneously examine work-family conflict in both directions and use dyadic data analysis.For example, while Burch (2020) examines couple-level data on work-family conflict, he only includes measures of WTFC (not FTWC).Likewise, while Liu and Hynes (2012) and Shreffler et al. (2010) account for work-family conflict in both directions, they do not use dyadic data.Begall and Mills (2011) assess other aspects of employment in addition to work-family conflict, but only include individual-level data from women (not men).To our knowledge, no existing study uses measures of both WTFC and FTWC and dyadic data analysis to examine the relationship between work-family conflict and fertility (in general) or partners' agreement on fertility preferences (specifically) as we do here.
Among the few studies that have been conducted on this topic, the results are inconsistent.Using individual-level, cross-sectional data on dual-earners in the United States ("U.S."), Shreffler et al. (2010) found that men's perceptions of their wives' workfamily conflict negatively influenced men's fertility intentions (but men's own workfamily conflict did not).Further, they found no significant relationship between women's own work-family conflict (nor their perceptions of their husbands' work-family conflict) and women's fertility intentions.In contrast, using European Social Survey (ESS) data from 23 counties, Begall and Mills (2011) documented an unexpected result.They found a strong, positive association between women's work-family conflict and fertility intentions among European mothers with one child (but did not examine fathers).They theorize that this is likely due to a selection effect: mothers who continue paid work after having one child but still highly value their family (and desire to have more children) probably already experience greater levels of work-family conflict.Differing once again, Liu and Hynes' (2012) study of longitudinal U.S. data found very little support for the notion that employed American women with poor work-life balance delay having or choose not to have children.In contrast, in a more recent longitudinal study of Australian dual-earners, Burch (2020) found that couples' WTFC lowered their likelihood of experiencing a subsequent birth/adoption indirectly (by working through less satisfaction with their family relationship).Thus, it is clear that more work is needed to better understand the complex links between work-family conflict and fertility, particularly in terms of how work-family conflict impacts partners' agreement on fertility preferences.
Prior works employing data from both developing (see, e.g., Bankole &Singh, 1998) anddeveloped countries (see, e.g., Hohmann-Marriot, 2009;Shreffler et al., 2019;Stein et al., 2014) suggest that partner disagreement over reproductive goals is quite common.However, studies rarely utilize reports from both partners and/or dyadic data analysis (as we do here).This topic is especially important within the German context, since their birth rates remain very low and greater partner agreement on fertility intentions has been linked to higher birth outcomes (Shreffler et al., 2019).In contrast, greater partner disagreement over fertility intentions and fertility timing has been linked to pregnancy avoidance and higher contraceptive use among partners who already have at least one child (Testa, 2012;Testa et al., 2014;Testa & Bolano, 2021).Thus, high levels of such disagreement could further lower Germany's birth rate.Beyond these implications for the birth rate, it is also vital to study the link between work-family conflict and partners' reproductive disagreement because, while literature on this topic is scarce, there are a plethora of plausible outcomes linked to fertility-related conflict.For example, experiencing discord over fertility preferences might threaten the health and the longevity of a relationship, resulting in a decline in relationship satisfaction and/or an increase in relationship dissolution.Ultimately, this could translate into higher levels of separation and divorce for couples who report more disagreement.Likewise, partners' disagreement over fertility preferences could have long-lasting and pervasive consequences for women (in particular) if it results in forms of reproductive coercion (Grace & Andersson, 2018), such as unintended pregnancies, that might threaten their educational and/or career goals (Cavalli, 2012;Stein et al., 2014).One existing study has even linked the absence of shared fertility intentions to higher risks of poor prenatal care and preterm births (Hohmann-Marriot, 2009).Accordingly, by examining the relationship between work-family conflict and partners' agreement over fertility plans, we take a vital step by isolating a potential source of reproductive disagreement that could be addressed in the future.

Theoretical frameworks
To generate predictions about the relationship between work-family conflict (in two directions) and partners' agreement on fertility preferences, we adopt a theoretical model first proposed by Bellavia and Frone (2005) and more recently elaborated upon by Amstad et al. (2011).As the latter work explains, the impact of work-family conflict is wide reaching, spanning a diverse collection of individual and couple-level outcomes across multiple social contexts.Accordingly, the plethora of outcomes associated with WTFC and FTWC can be categorized into three major domains: (1) work-related (e.g., job turnover, organizational commitment), (2) family-related (e.g., parenting, relationship conflict), and (3) domain-unspecific (e.g., substance use, depression).Here, we examine conflict among partners over fertility preferences, which falls under the family-related domain.According to this framework, there are two hypotheses that might explain the connections between work-family conflict and outcomes in each domain: (a) the "matching hypothesis" and (b) the "cross-domain hypothesis."The "matching hypothesis" (Amstad et al., 2011) predicts that domain-specific outcomes are more likely to be influenced by the form of work-family conflict rooted in the same domain.In other words, because the source of FTWC is the family domain, the matching hypothesis predicts that it should be more consequential on fertility preferences (a family-domain outcome) than WTFC will be (since its source is a separate domain -work).In contrast, the "cross-domain hypothesis" (Frone et al., 1992) proposes that WTFC and FTWC should have stronger associations with outcomes that exist in different domains.Thus, because the source of WTFC is the work domain, the cross-domain hypothesis predicts that it should be more consequential on fertility preferences (a family-domain outcome) than FTWC will be (since its source is the same domain -family).According to meta-analyses conducted on work-family conflict, there is greater empirical support for the matching hypothesis (Amstad et al., 2011;Nohe et al., 2015).Thus, we take these findings into account as we formulate our own predictions about how work-family conflict might impact partners' agreement on fertility preferences.

Actor effects
There are also two theoretical avenues through which work-family conflict can travel across domains and impact partners' agreement on fertility preferences: (1) spillover (actor) effects and (2) crossover (partner) effects.The former, spillover, is due to the presence of inter-role conflict (Greenhaus & Beutell, 1985), such as conflict between one's worker role and one's parental role.As the "role-scarcity hypothesis" (Edwards & Rothbard, 2000) predicts, when a person's multiple social roles attempt to pull from a finite set of resources, strain often transpires and negatively affects that individual as a result.Moreover, as Pearlin's stress process model (1999) explains, the primary stressors individuals experience can result in secondary stressors that intensify and/or exacerbate the effects of spillover.Thus, we expect that when there is conflict between one's work and family roles (i.e., the primary stressor), one will also experience more conflict with one's partner (i.e., the secondary stressor), which can lead to or amplify a couples' uncertainty over other important topics, such as fertility intentions.In other words, when work interferes with family, and/or when family interferes with work, it likely creates greater levels of tension and stress for individuals and their partners.Accordingly, greater workfamily conflict is likely associated with couples experiencing more arguments, increased conflict, and greater levels of disagreement over a plethora of issues, one of which being whether they desire to have a(nother) child together.Disagreement over fertility intentions seems especially likely to be influenced by work-family conflict, since arguments and conflict with one's partner over whether they have the time and resources needed to care for one or more children are likely intimately linked to each partner's existing perception of their current work-life balance.More specifically, while we expect that greater WTFC and FTWC will both be associated with lower levels of partner agreement on fertility preferences, due to greater empirical support for the matching hypothesis (Amstad et al., 2011;Nohe et al., 2015), we anticipate that FTWC will have a more significant impact (Hypothesis 1).

Actor effects by gender
Starting in the latter half of the 20 th century, the gendered division of labor began to shift and be renegotiated in households across the U.S. and Europe (Galinsky et al., 2011;Winslow, 2005).For example, married fathers are spending more time completing housework and childcare, while married mothers are completing more hours of paid work (Bianchi et al., 2006).Just as the number of dual-earner couples increased, scholars documented a concurrent rise in reports of work-family conflict for men and women.For example, several studies have suggested that work-family conflict is increasing among married fathers in dual-earner couples (Aumann et al., 2011;Galinsky et al., 2011;Nomaguchi, 2009).While there is evidence of movement toward gender egalitarianism, gender inequalities still persist in many households and social institutions.For instance, women still complete nearly two times as much housework and childcare as men do and many workplaces have not been responsive to changing gender dynamics in the home (Bianchi et al., 2006;Gerson, 2010).
A number of theories have been proposed to predict differences in work-family conflict by gender (Shockley et al., 2017), resulting in opposing hypotheses about the impact this form of inter-role conflict has for men and women.We compare two of these frameworks here.First, the "rational view" (Gutek et al., 1991) suggests that a person's work-family conflict is directly related to the amount of time they invest in the work and family domains.As mentioned, women tend to spend more time completing family responsibilities (housework and childcare) than men, while men tend to spend more time completing workplace responsibilities (paid employment) than women.Accordingly, the rational view predicts that women will experience and be impacted by FTWC more often than men, while men will experience and be impacted by WTFC more often than women.In contrast, the "sensitization perspective" (Duxbury & Higgins, 1991;Gutek et al., 1991) takes gender role socialization into account, linking the work and family domains to masculine and feminine ideals.Because traditional gender roles teach men to place greater value on their roles as workers, while women are taught to place greater importance on their roles as caregivers, men and women likely perceive work-family conflict in different ways.In essence, sensitization theory predicts that men will report and be more negatively affected by FTWC because it disrupts their more salient worker identity, while women will find WTFC more intrusive because it threatens their more salient family identity.While both the "rational view" and "sensitization perspective" have been employed and tested in the literature, Shockley et al.'s (2017) meta-analysis suggests that the "rational view" receives the most empirical support.Thus, we adopt this approach here and predict that, due to gendered patterns in time use, the actor effects of WTFC will be stronger for men than for women, while the actor effects of FTWC will be stronger for women than for men (Hypothesis 2).

Partner effects
In addition to spillover (actor effects), work-family conflict can travel across partners via crossover (partner effects).Bakker et al. (2008) note that spillover should be viewed as "intraindividual" process, while crossover is an "interindividual" process, because spillover examines individual-level outcomes associated with one's own work-family conflict, while crossover accounts for couple-level transmission of work-family conflict.In other words, crossover isolates the negative influence a person's own WTFC and/or FTWC has on their partner's outcomes.Westman (2001Westman ( , 2006) ) suggests three possibilities for explaining how work-family conflict related strain moves across members of a couple during crossover.It is due to: (1) the empathy partners feel for each other in times of difficulty, (2) the close physical spaces and interconnected relationships shared by partners, and/or (3) the decreased effectiveness of couples' interpersonal communication during times of stress.In the case of work-family conflict and fertility, perhaps conflict between work and family roles can make partners question their ability to establish workfamily balance in the future if they choose to increase their family size.
Using couple-level data and accounting for both actor and partner effects is also important from a theoretical and methodological standpoint.Scholars have long stressed the importance of using couples (not individuals) as the unit of analysis when examining both of our key variables: work-family conflict (see, e.g., Young et al., 2014;Yucel & Latshaw, 2020) and fertility-related outcomes (see, e.g., Thomson et al., 1990;Shreffler et al., 2019;Stein et al., 2014;Stykes, 2015).While many existing studies on both topics use reports from only one member of a couple (often women), this practice can lead to "assumed similarity bias" (Kenny & Acitelli, 2001), which occurs when someone assumes another person has the same experience they do (which might not be accurate).Likewise, prior work suggests that couples do not always agree on their fertility intentions and preferences (see, e.g., Hohmann-Marriot, 2009;Shreffler et al., 2019), further illustrating the importance of including information from both partners here.Moreover, couples have a degree of interdependence as a unit that cannot be captured using individual-level data, particularly when it comes to fertility decision-making (Stein et al., 2014).Accordingly, we expect that adding in partner effects and accounting for the dyadic nature of our data is not only theoretically sound, but will also significantly improve our empirical model.Given the importance of partner effects, we thus expect that one's greater WTFC and FTWC will be associated with their partner's lower levels of reported agreement on fertility preferences (Hypothesis 3).

Partner effects by gender
Just as we predict that the actor effects of work-family conflict will vary for men and women, we anticipate that any partner effects resulting from the crossover of work-family conflict to partners will also differ by gender.We previously adopted the "rational view" (Gutek et al., 1991), which proposes that men (women) likely experience more WTFC (FTWC) due to a traditionally gendered division of labor (where men typically spend more time completing paid work and women spend more time completing unpaid housework and childcare).If we apply this same theoretical perspective here, we might anticipate that the partner effects of WTFC will be greater for women, while the partner effects of FTWC will be greater for men.However, if we adopt a feminist lens to examine the links between gender role socialization and partners' agreement on fertility preferences, we generate slightly different hypotheses about the gendered nature of crossover.For example, previous work on work-family conflict and mental health has documented greater negative partner effects for women than for men (Symoens & Bracke, 2015).This could be due to women's tendency to internalize stress experienced by themselves, their partners, and/or their close contacts (Kessler & McLeod, 1984;Simon, 1998).According to Simon (1998), this is rooted in gender role socialization, where women are taught that femininity is associated with being nurturing, empathetic, caring, etc.As a result of this socialization process, women have been shown to be more aware of conflict occurring to themselves and others (Rosenfield & Smith, 2010).Thus, when male partners experience work-family conflict, it is perhaps more likely to negatively impact female partners than vice versa, since women are socialized to be more aware of, acknowledge, and help manage the stress others experience (Simon, 1998;Symoens & Bracke, 2015).In other words, if a husband is experiencing work-family conflict in either direction, his wife is more likely to perceive and report disagreement or conflict resulting from it than vice versa.Thus, we predict that women are more likely than men to experience stronger, negative partner effects from greater levels of both WTFC and FTWC (Hypothesis 4).

The role of working women's employment status
It is also important to consider whether the relationship between work-family conflict and partners' agreement on fertility preferences varies by the employment composition of dual-earner couples.Here, we compare two aforementioned couple types: (1) "full-time dual-earner" couples (where both partners work full-time) and ( 2) "modernized male breadwinner" couples (where the male partner works full-time and the female partner works part-time).This comparison is particularly germane given the larger institutional context of our data.Across Western European countries, Germans report some of the highest levels of work-family conflict, particularly for men (Gallie & Russell, 2009).This has been attributed, in part, to a continued adherence to male breadwinner gender ideologies (Florean & Engelhardt, 2020;Geist & Brauner-Otto, 2017;Konig & Cesinger, 2015).Prior work has also studied the link between traditional gender ideologies and transitions in German family policies.Until the late 20th century, German policies tended to promote marriage and the male breadwinner ideal (Fleckenstein & Seeleib-Kaiser, 2011;Klammer & Letablier, 2007).Since this time, a series of more egalitarian reforms have passed, such as offering a lengthier parental leave when fathers take part in newborn care (Konig & Cesinger, 2015).
While these reforms are promising from the perspective of gender equality, prior work suggests that German women still complete substantially more housework and childcare than men and have much higher rates of part-time employment (Keller & Haustein, 2012).While some studies have documented a better work-life balance among mothers who work fewer hours (see, e.g., Fagan & Burchell, 2002;Van Rijswijk et al., 2004), findings are not consistent.For example, part-time employed German women actually document greater levels of FTWC than those who work full-time, likely due to their heavy load of unpaid labor (Konig & Cesinger, 2015).Konig and Cesinger (2015, p. 544) called this finding "a high double burden for part-time employed women," and suggest it is evidence of the continued preference for traditional gender role arrangements in Germany.Likewise, Begall and Mills (2011) found no significant relationship between part-time work and fertility intentions for mothers, but a significant, negative relationship between part-time work and fertility intentions among women without children.After conducting additional analyses, they conclude that women without children (particularly those living in countries with lower levels of part-time work) might view part-time employment more negatively or as "constraining" because these jobs are often of a poorer quality.Additionally, Shreffler and Johnson (2013) found there is no clear, linear, negative relationship between increased work hours and fertility.
Thus, as prior research illustrates, the relationship between work-family conflict, women's employment status, and fertility outcomes is unclear.Moreover, to our knowledge, no existing studies have examined the relationship between work-family conflict (in both directions) and partners' agreement on fertility preferences, nor the moderating effects of women's full-time versus part-time employment status on this relationship.However, because German women who work part-time experience more FTWC than full-time working women (Konig & Cesinger, 2015), we expect that the effects FTWC will be more pronounced among couples where women work fewer hours.Accordingly, we predict that the actor and partner effects of work-family conflict (FTWC in particular) will be stronger among modernized male breadwinner couples than among full-time dual-earner couples (Hypothesis 5).

Data and methods
In order to test these research questions, this study uses data from the German Family Panel (pairfam).This data focuses on partnership and family dynamics in Germany and is funded by the German Research Foundation (Brüderl et al., 2019;Huinink et al., 2011).The survey, first launched in 2008-2009, collects data from a national, random sample of 12,402 anchor persons, as well as their spouses and partners, annually.Questions about work-life balance were only asked in Waves 6, 8, and 10.Analyses are based on data from Wave 10 of the German Family Panel (pairfam), release 12.0 (Brüderl et al., 2019). 1 A 1 There are three main reasons why we did not choose to perform a longitudinal analysis of waves 6-10.First, due to our adoption of the APIM framework, our focus would have been to observe the same couples over time, and the attrition rate across waves would have led to an even smaller sample size of couples if we had detailed description of the study can be found in Huinink et al. (2011).Our sample is limited to 1569 heterosexual couples 2 who are either married or cohabiting and from whom we have information from each partner (from the anchor and partner surveys).Among these couples, we dropped couples where either/both partner(s) was/were not in the labor force (N=368). 3Next, we dropped couples who had no valid information on our outcome variable (agreement with partner on fertility preferences) (N=440).This question was only asked to respondents who were not infertile, were not expecting a child currently, and/or who indicated that they would like to have (more) children in the future.Finally, due to the small sample size, we dropped couples where neither partner worked full-time (N=14), couples where only men worked part-time (N=24), and couples with missing information on employment status (N=7).This left us with a final sample of 716 married and cohabiting dual-earner couples.
On average, women in our sample were 35 years old and men were 38 years old.Around 60 percent of women completed higher education, whereas 51 percent of men completed it.Couples in this sample have been married for an average of 8 years.Men in the sample work about 43 hours per week, whereas women work 31 hours per week.Around 33 percent of the couples have a preschool-aged child (< 6 years of age) living in used a longitudinal approach.This is due to some couples being eliminated from the sample because of separation or divorce, or due to them having an incomplete partner survey.Second, in supplemental analyses, we examined the differences between the analytical sample we use in our study and the sample that would have been lost due to attrition if we had employed a longitudinal approach.The results suggest that the sample that would have been lost to attrition is younger, less educated, has lower income, and is less likely to be married or to have a preschool-aged child in the household.By only using the remaining sample, we would have introduced potential bias, since some of these differences would have likely impacted partners' agreement on fertility preferences.Finally, supplemental analysis suggests that there is a very small amount of variation in the outcome variable across time.Specifically, less than four percent of men and three percent of women reported any change in the level of agreement with their partner on fertility preferences between waves 6 and 10.While some prior research using pairfam pooled the data from multiple waves and treated the panel as cross-sectional data (Hajek, 2019), there are also three main drawbacks to using this approach.First, when dealing with dyadic data/the APIM framework, a pooled cross-sectional approach would assume homogeneity of variance in the outcome variable between men and women.Thus, this design would not allow us to account for heterogeneity within couples in our study.Second, this approach cannot accommodate missing data, so couples would be deleted if they did not have data on any variable, leading to a reduction in sample size and the power of statistical tests (Tambling et al., 2011).Third, because some of the variables we use in this study are time-invariant (such as work-family conflict, partners' agreement on fertility preferences, work hours, and the presence of a preschool-aged child in the household), pooling random and different samples from different waves (i.e., time periods) would not be suitable to test our research question, nor to estimate the actor and partner effects as part of the APIM model.When considering all of the drawbacks of adopting these alternative approaches, we chose to only use data from Wave 10 to test our research questions. 2 Due to the small number of same-sex couples in the data and the theoretical limitations that would likely arise if we were to combine both types of couples in our study, we limit our analysis to different-sex couples.3 Our analytical approach focuses on actor and partner effects of each partner's work-family conflict on their own and their partner's level of agreement on fertility preferences.Thus, testing actor and partner effects suggests that we should focus on couples where there is valid information about work-family conflict from both partners.In the pairfam study, work-family conflict questions are only asked to those who are employed at the time of the interview.Thus, our analytical approach forced us to limit our sample to couples where both partners are currently employed.

Dependent variables
This study uses partners' agreement on fertility preferences as the main dependent variable.This is measured with the following question: "Do you and your partner agree about how many children you would like to have together?"The answer categories range from 1 (totally disagree) to 5 (totally agree). 4We measure this item separately for men and women.There is a moderate correlation between the outcomes for men and women (Pearson's r=0.51).

Independent Variables
We use two distinct forms of work-family conflict as our independent variables: WTFC and FTWC.In order to measure WTFC, respondents were asked four questions: To what extent do the following statements apply to you? "(1) Due to my professional, vocational training, or university workload, my personal life suffers.(2) Even when I am doing something with my friends, partner, or family, I often think about work.
(3) After a stressful time at work, I find it difficult to relax at home and/or to enjoy my free time with others.(4) My work prevents me from doing things with my friends, partner, and family more than I'd like."Each item ranged from 1 (not at all) to 5 (absolutely).We ran exploratory factor analysis with varimax rotation on each item in this measure.The factor loadings of these four items ranged from 0.50 to 0.67 and 0.51 to 0.69 for women and men, respectively.Cronbach's alpha scores for women and men were 0.80 and 0.74, respectively, indicating high internal reliability.After summing the four items and taking the average to create a scale, higher scores indicated greater levels of WTFC.
To measure FTWC, respondents were asked: "(1) Because I am often stressed in my private life, I have problems concentrating on my work.(2) Because of my personal schedule, I often lack time to do my work.(3) The time I need for my partner, family, and friends keeps me from being more involved in my job, vocational training, or university education.(4) Conflicts in my personal life reduce my work performance."Each item ranged from 1 (not at all) to 5 (absolutely).We ran exploratory factor analysis with varimax rotation on each item in this measure.The factor loadings of these four items ranged from 0.53 to 0.62 and 0.51 to 0.61 for women and men, respectively.Cronbach's alpha scores were 0.70 and 0.75 for women and men, respectively, indicating high internal reliability.After summing the four items and taking the average to create a scale, higher scores indicated greater levels of FTWC.

4
There is evidence from prior research that a Likert scale, when it has five or more answer categories, can be treated as a continuous variable and this causes no harm to the planned analyses (Zumbo & Zimmerman, 1993;Taylor et al., 2006).With this in mind, even though the answer categories are ordinal, we treated our dependent variable as continuous and performed all analyses accordingly.

Control variables
This study also used several other variables that have been shown to be associated with our independent and/or dependent variables (Stein et al., 2014;Shreffler et al., 2010Shreffler et al., , 2019;;Shreffler & Johnson, 2013): men's and women's education, men's and women's age in three categories (<35 years old (reference), 35-44, 45 and older) 5 , relationship status, household income (logged) 6 , men's and women's work hours, relationship duration (logged) 7 , the presence of a preschool-aged child in the household, the number of children in the household 8 , and whether the couple lives in East or West Germany.

Analytical strategy
The analytical approach in this study is similar to some prior research (Yucel & Gassanov, 2010;Yucel & Latshaw, 2020).We test the actor and partner effects of WTFC and FTWC on partners' agreement on fertility preferences using The Actor-Partner Interdependence Model ("APIM") framework.The APIM framework allows us to test the effect of one's own WTFC and FTWC on their own reported agreement with their partner on fertility preferences (i.e., actor effects) and on their partner's reported agreement with them on fertility preferences (i.e., partner effects). 9We run dyadic data analyses using Structural Equation Modeling (SEM) in AMOS 22.0, running nested models to test the aforementioned effects.Maximum likelihood estimation, as part of SEM, is used for imputing missing cases. 10First, we divide the sample into two couple types: full-time dual-earner and modernized male breadwinner couples.For each couple type, we then run the following models to test Hypotheses 1-4.In the first model (Model 1), we test the actor effects of WTFC and FTWC.Next, using chi-square difference tests, we assess whether there are any gender differences in the actor effects (Model 2).We then explore whether the partner effects of WTFC and FTWC improve the model by adding these paths 5 We included age as a categorical variable (instead of as a continuous variable) due to its high correlation with relationship duration.6 Respondents are asked for their total net household income in the last month.Respondents who do not wish to give precise answers to the open-ended questions about income are then asked to select an income category that best fits with their situation.Afterwards, the midpoint of the respective income category was assigned as their household income (Brüderl et al., 2019).7 Both household income and relationship duration have a positively skewed distribution.Therefore, a logarithmic transformation was used to reduce the skewness for both variables.8 Children include any children (i.e., biological, adopted, step, or foster children) living in the household.9 The APIM framework, while testing the actor/spillover and partner/crossover effects, necessitates us to use a unique measurement of the outcome variable from each partner (i.e., male partners' agreement on fertility preferences and female partners' agreement on fertility preferences).Thus, using this approach does not allow us to test a relative comparison in agreement on fertility preferences between partners (i.e., the nature of agreement/disagreement on fertility preferences between a male partner and a female partner).10 We also ran sensitivity analyses where we replicated the analyses using listwise deletion and compared the results to our approach in the paper where missing cases are imputed via maximum likelihood estimation.
Upon doing so, we reached the conclusion that the results were robust across these two different imputation methods.The results of the sensitivity analyses (see Table A) are included in the online appendix.
in the analysis (Model 3).Finally, we test whether there are any gender differences in these partner effects (Model 4).We compare the relative fit indices of these nested models between those that are constrained and unconstrained to decide on the best-fitting model.The unconstrained model is where the coefficients are estimated separately for both women and men within each couple type.The constrained model is where certain coefficients are constrained to be equal for both women and men within each couple type.
If the difference in chi-square between the constrained and unconstrained model is significant, then the unconstrained model (which allows paths to be estimated freely) is preferred.If the difference is not significant, then the constrained models with pooled estimates are preferred (Kenny & Cook, 1999).Finally, to test Hypothesis 5, we use multigroup analyses to assess whether actor or partner effects vary across the two couple types.

Descriptive findings
Descriptive findings for all variables are displayed in Table 1.The results show whether there are significant differences in each item between full-time dual earner couples and modernized male-breadwinner couples, as estimated by independent t-tests (for continuous variables) and chi-square tests (for categorical variables).First, there are no differences in partners' agreement on fertility preferences (for either men or women) between full-time dual-earner couples and modernized male breadwinner couples.In addition, there are no differences in men's WTFC and FTWC between these two couple types.While there are also no differences in women's FTWC between these two couple types, women's level of WTFC is significantly stronger among full-time dual-earner couples compared to modernized male breadwinner couples (p<.001).Modernized male breadwinner couples have also been married for a longer time than full-time dual-earner couples (9 versus 7 years).There are no significant differences in men's work hours between full-time dual-earner couples and modernized male breadwinner couples.In addition, modernized male breadwinner couples are significantly more likely to have a preschool-aged child living in the household (47 versus 24 percent) and to have more children living in the household (2 children versus 1 child) compared to full-time dualearner couples.Men and women in full-time dual-earner couples are also more likely to have completed higher education compared to modernized male breadwinner couples (54 versus 47 percent for men, 64 versus 55 percent for women, respectively).Moreover, fulltime dual-earner couples are more likely to live in East Germany when compared to modernized male breadwinner couples (47 versus 22 percent).Finally, full-time dualearner couples have a higher household income compared to modernized male breadwinner couples.

Multivariate analyses
The results from full-time dual-earner couples are presented in Table 2, while the results from modernized male breadwinner couples are displayed in Table 3.

Full-time dual-earner couples
First, Model 1 shows that among full-time dual-earners, an increase in one's own FWTC (but not WTFC) for both men and women is associated with their own reports of lower levels of agreement with their partner on fertility preferences, and this effect is significant (p<.01 for men and p<.10 for women, respectively).Model 2 constrains the paths for actor effects to be equal for men and women.According to the chi-square difference test comparing Models 1 and 2, the effects are not significantly different (Δχ 2 =2.12, df=2, p>.10), and thus, we conclude that there are no gender differences in actor effects.Model 2 is preferred compared to Model 1.This suggests that the negative association between one's own FTWC and one's own reported level of agreement with their partner on fertility preferences is the same for men and women.Model 3 adds the paths for partner effects.
The results show that the chi-square difference is not statistically significant when comparing Models 2 and 3 (Δχ 2 =2.06, df=4, p>.10).Thus, adding the paths for partner effects does not significantly improve the model.Therefore, Model 2 is preferred and is the best-fitting model.In this model, there are significant actor effects for FTWC (but not WTFC) for men and women (with no gender differences).Thus, Hypothesis 1 is partially supported, Hypothesis 2 is not supported, and because partner effects do not improve the model, Hypotheses 3 and 4 are not supported.Specifically, for both men and women, higher levels of FTWC are associated with their own lower levels of reported agreement with their partner on fertility preferences (b=-.176,SE=.053, p<.001).From the control variables, being married is positively associated with both men's and women's own reports of agreement with their partner on fertility preferences (p<.05).Having a preschool-aged child living in the household is negatively associated with women's reports of agreement with their partner on fertility preferences (p<.10).The full table for this bestfitting model is included in the online appendix (see Table B, the first column).Figure 1 displays the unstandardized coefficients from the best-fitting model (i.e., Model 2) for fulltime dual-earners.

Modernized male breadwinner couples
Table 3 shows the same steps for modernized male breadwinner couples.Model 1 results show that there are significant actor effects.Specifically, higher levels of FTWC for men and higher levels of WTFC for women are both associated with their own lower levels of reported agreement with their partner on fertility preferences (p<.10 and p<.001, respectively).Model 2 constrains these actor effects to be equal between men and women.
According to the chi-square difference test comparing Models 1 and 2, we find that the difference is not significant (Δχ 2 =3.68, df=2, p>.10), suggesting that these constraints improve the model.Thus, Model 2 is preferred (compared to Model 1), and we can conclude that there are no gender differences in these actor effects.Next, we add the partner effects in Model 3. Adding these paths significantly improved the prior model (Δχ 2 =11.10, df=4, p<.05), and thus, adding partner effects does improve the model.Specifically, the results suggest that the partner effect of FTWC for men is associated with lower levels of agreement on fertility preferences reported by female partners (p<.05).Thus, Model 3 is preferred over Model 2. Finally, in Model 4, we add the constraints where the partner effects of WTFC and FTWC for men and women are constrained to be equal.Once we add these constraints, the chi-square difference test, comparing Model 3 and 4, is not significant (Δχ 2 =4.74, df=2, p>.10).This suggests that the constrained model (Model 4) is preferred over Model 3. Overall, Model 4 is the best-fitting model for modernized male breadwinner couples.This model suggests that there are significant actor effects for both WTFC and FTWC for men and women (p<.01 and p<.10, respectively), but there are no gender differences in these actor effects.Thus, Hypothesis 1 is supported, while Hypothesis 2 is not supported.In addition, we find that there are significant partner effects for FTWC (but not WTFC) for both men and women (p<.05).Therefore, Hypothesis 3 is partially supported.However, there are no gender differences in these partner effects between men and women.Thus, Hypothesis 4 is not supported.Specifically, for both men and women, higher levels of WTFC and FTWC are both associated with their own lower levels of reported agreement with their partner on fertility preferences SE=.051,p<.01;SE=.065,p<.10,respectively).In terms of partner effects, higher levels of FTWC (but not WTFC) for both men and women are associated with lower levels of agreement on fertility preferences as reported by their partners (b=-.156,SE=.065, p<.05).Longer relationship duration and being married are also both positively associated with women's own reports of agreement with their partner on fertility preferences (p<.01 and p<.10, respectively).The full table for this best-fitting model is included in the online appendix (see Table B, the second column).Figure 2 displays the unstandardized coefficients from the best-fitting model (Model 4) for modernized male breadwinner couples.
Table 3: Predictive model of partners' agreement on fertility preferences among modernized male breadwinner couples Note: WTFC=work-to-family conflict; FTWC= family-to-work conflict.Higher scores in WTFC and FTWC indicate higher levels of WTFC and FTWC, respectively.The indicators of the dependent variable are coded so that higher scores indicate more agreement with one's partner on fertility preferences.This study also controls for relationship duration (logged), relationship status, men and women's work hours, men and women's age (in categories), men and women's education, household income (logged), living in East or West Germany, having preschool children living in the household, and number of children living in the household.Higher scores indicate higher levels of agreement between partners on fertility preferences.

Multi-group analyses
Finally, as the last step, we test whether actor and partner effects vary across the two couple types.In order to do this, nested models are estimated, such that the actor and partner effects of WTFC and FTWC are constrained to be equal across groups in consecutive steps (Byrne, 2013).Then, using chi-square difference tests, the model fit between each of these constrained models is compared to the model fit of the baseline unconstrained model (Model 1).According to chi-square difference test results, adding constraints for actor effects for WTFC and FTWC, for both men and women, improves the fit of the baseline model (Model 1).Thus, findings suggest that actor effects for WTFC and FTWC do not differ between full-time dual-earner and modernized male breadwinner couples.On the other hand, when we add the constraints for the partner effects for WTFC and FTWC for men and women, the chi-square difference also suggests that there is no difference in the partner effect for WTFC between these two couple types (for both men and women).We find, however, that the partner effect for FTWC, for both men and women, is significantly different between these two couples (Δχ2=5.50,df=1, p<.05).Specifically, the partner effect for FTWC for both men and women is significantly different (and in the negative direction) among modernized male breadwinner couples.Thus, Hypothesis 5 is partially supported.The fit indices of nested models for the multigroup analyses are shown in Table 4 (the models which are significantly different between two groups are shown in bold font).Step 4: actor effect of women's WTFC is equal across groups Step 5: actor effect of women's FTWC is equal across groups Step 6: partner effect of men's WTFC is equal across groups Step 7: partner effect of men's FTWC is equal across groups Step 8: partner effect of women's WTFC is equal across groups Step

Discussion
This study uses data from 716 dual-earner couples to test the effect of each partner's WTFC and FTWC on their own and their partner's reported level of agreement on fertility preferences.In addition, we test whether there are gender differences in these actor and partner effects.Finally, we analyze whether the main effects vary for full-time dual-earner versus modernized male breadwinner couples.The results show that there are significant actor effects among both types of couples.Partner effects exist only among modernized male breadwinner couples.However, neither actor nor partner effects differ by gender.
Results also suggest differences in the partner effect of FTWC (but not WTFC) between these two couple types, such that the negative effect is found to be significantly stronger among modernized male breadwinner couples than full-time dual-earner couples.
Our results can be interpreted in light of several key theoretical frameworks, and are, at times, both consistent and divergent from prior research on work-family conflict and fertility-related outcomes.First, when we revisit the theoretical expectations outlined by the matching (Amstad et al., 2011) versus the cross-domain (Frone et al., 1992) hypotheses, our findings are more aligned with the former than the latter.In accordance with the matching hypothesis and as supported by prior work (Amstad et al., 2011;Nohe et al., 2015), FTWC (which emerges from the family domain) appears to have a stronger negative effect on partners' agreement on fertility preferences (a family domain outcome) than WTFC (which emerges from a different domain -the workplace).The predictions of the matching hypothesis are also supported by several of our key findings.For example, we find that for men and women in full-time dual-earner couples, higher levels of their own FTWC (but not WTFC) are associated with their own lower levels of reported agreement on fertility preferences.Likewise, among modernized male breadwinner couples, higher levels of FTWC (but not WTFC) for both men and women are associated with their partners' lower levels of reported agreement on fertility preferences.Thus, it appears that when considering partners' fertility preferences, lower levels of agreement are more strongly related to work-family conflict that travels in the direction of FTWC (and is rooted in the family domain).
Our findings also support the predictions of the "role-scarcity hypothesis" (Edwards & Rothbard, 2000).Specifically, we document that greater levels of FTWC are associated with one's own lower levels of reported agreement on fertility preferences in both couple types.This aligns with the predictions of this hypothesis, because when one has a finite number of resources, the strain experienced when family roles interfere with work roles can lead to greater levels of perceived conflict with one's partner over shared fertility plans.These findings also support the theoretical framework Pearlin calls the "stress process model" (1999).As this model proposes, it does appear that primary stressors (e.g., having a sick child) can lead to secondary stressors (e.g., missing an important meeting at work), which in turn, can lower one's perceptions of agreement and heighten one's perceptions of conflict with their partner on fertility preferences.While this contrasts with some prior work that found no relationship (Shreffler et al., 2010) and a positive relationship (Begall & Mills, 2011) between work-family conflict and fertility intentions, our findings do align with a more recent study suggesting there is a negative relationship between work-family conflict and subsequent birth/adoption (Burch, 2020).Likewise, it supports a vast array of prior work documenting the negative spillover effects of workfamily conflict (in both directions) on a wide variety of outcomes (Allen et al., 2000;Amstad et al., 2011;Bakker et al., 2008;Frone et al., 1996;Michel et al., 2009), as well as the negative impact work-family conflict can have on relationships, specifically (Burch, 2020;Judge et al., 2006).Moreover, the fact that we find these effects across both couple types suggests that the negative effects work-family conflict has on individuals are present in dual-earner couples, regardless of whether women are employed full-time or part-time.
While we expected to find gender differences in both the spillover and crossover effects of work-family conflict, neither actor nor partner effects differed for men versus women in either type of couple.One of the few existing studies we highlighted did document gender differences in work-family conflict on fertility intentions, but it was not in the manner the authors necessarily expected (Shreffler et al., 2010).This prior work found that men's perceptions of their wives' work-family conflict was related to men's fertility intentions (but their own work-family conflict was not) and neither women's perceptions of their husbands' work-family conflict nor their own work-family conflict were related to women's fertility intentions.While we can only speculate, perhaps our lack of significant gender differences in the effect of work-family conflict on agreement on fertility preferences could be due to our sample, which only contains dual-earner couples.Dual-earner couples tend to be more egalitarian in terms of the gendered division of labor than are more traditional male breadwinner/female homemaker couples (Gerson, 2010).Thus, for dual-earner couples, the experiences men and women have in the workplace and home could be more alike than they are different (Young et al., 2014).
Finally, when it comes to crossover effects, we do find significant results, but it is only for the partner effect of FTWC (not WTFC) and it is only significant for modernized male breadwinner couples (not full-time dual-earner couples).The presence of this significant partner effect does confirm Westman's assertion (2001Westman's assertion ( , 2006) ) that crossover can occur due to partners' shared empathy and interconnectedness and/or a decline in effective couple communication occurring when conflict arises.Here, it seems likely that the negative effects of FTWC are crossing over to partners due to poorer communication among couples (as evidenced by the greater disagreement couples report on their fertility preferences).We can also speculate several plausible theories explaining why we only find this among modernized male breadwinner couples.First, perhaps it is due to a lack of symmetry in the work and family roles of these couples.While one might assume that having two full-time workers in a household could generate more conflict, perhaps, instead, it is a lack of parity in men's and women's work hours that might result in greater conflict between partners.For example, perhaps women in modernized male breadwinner couples have part-time jobs, but are still expected to complete the grand majority of unpaid housework and childcare.Alternatively, perhaps the crossover effects of FTWC are more pronounced in these couples because it amplifies the resent part-time working women might already feel about their work and family roles.For example, these women might prefer to work full-time, be unhappy working part-time (because they would rather stay home), have part-time jobs that require their full-time attention, or be working parttime due to a lack of available or affordable childcare.Regardless of the reason, if the parttime working women in modernized male breadwinner couples are unhappy with their present work-family arrangement, experiencing FTWC could heighten doubts and change perceptions of partners' agreement on fertility preferences.Moreover, the presence of FTWC could prompt part-time working women to fear losing even more connection to the labor force if they were to expand their families.Given the fact that the underlying mechanisms are not clear, we encourage scholars to continue exploring the role women's employment status plays in moderating the relationship between work-family conflict (FTWC in particular) and fertility outcomes.

Limitations and contributions
There are some limitations to our study.First, our analyses are based on a relatively small number of couples, which could negatively influence the reliability and significance of our results, as well as explain cases in which our findings deviate from existing work on this topic.A longitudinal study with a larger sample of the same couples, observed over time, would be better suited to explore the long-term and causal actor and partner effects of work-family conflict on partners' agreement on fertility preferences.We encourage future research to explore this possibility.In addition, due to data limitations, this study could not examine the effects of any positive spillover and/or crossover between work and family domains (i.e., work-family enrichment --see, e.g., Greenhaus & Powell, 2006).It would be useful for future research to investigate the effect of work-family enrichment (in both directions) on partners' agreement on fertility preferences.Likewise, because pairfam only asks work-family conflict questions to individuals and partners who are currently employed, we could not include "traditional male breadwinner" or "female breadwinner" couples as comparison groups in our analyses.We encourage scholars to explore other data sources to more deeply examine whether the relationship between work-family conflict and fertility varies among couples where one partner is not in the labor force.Moreover, while our main contribution is establishing that the previously unexplored connection between work-family conflict and partners' agreement over fertility preferences exists and is significant, we encourage future scholars to assess in more detail the moderating and mediating mechanisms (e.g., decreased couple-level communication, worsened mental health, and/or changes in perceptions of the value of children) of this connection.For example, the quantitative approach presented in this paper could also be complemented by future qualitative work to shed further light into the mechanisms and reasons as to how and why work-family conflict can lead to lower levels of partner agreement on fertility preferences.Due to data limitations, we were also unable to include same-sex couples in our analysis.Prior work suggests that the complexities of work-family conflict (Allen & Eby, 2016), as well as fertility intentions (Hank & Wetzel, 2018), might operate differently among same-sex couples.Thus, we encourage future scholars to explore this topic among couples who do not identify as heterosexual.Moreover, due to our use of the APIM model and dyadic data analysis as our analytical strategy, this study could not test the effect of work-family conflict on a relative comparison in the level of reported agreement on fertility preferences between partners.Disentangling the nature of disagreement between partners might have different implications, such as being able to more precisely determine whether male partners perceive more disagreement than female partners, or vice versa.We encourage future scholars to explore this relationship between work-family conflict and partners' fertility-related disagreement using a different analytical approach to capture the more nuanced, relative differences between male and female partners.Finally, our findings cannot be generalized beyond the German context.Future research could attempt to replicate our results across a wider array of countries and/or integrate macro, country-level variables (such as GDP, state policies, etc.) into the study design.
Despite these limitations, this study makes significant contributions to prior literature.First, to our knowledge, this is the first study that uses couple-level data and dyadic data analysis to empirically investigate the effect of one's own WTFC and FTWC on their own and their partner's reports of agreement on fertility preferences.In addition, this study extends prior work by testing gender differences in the effects of work-family conflict on partners' agreement on fertility preferences.Moreover, by testing the moderating effect of the employment composition of dual-earner couples, we uncover how working women's employment status impacts the association between work-family conflict and partners' agreement on fertility preferences.Finally, we add confirmation that despite a movement toward more dual-earning couples in Germany, traditional gender ideologies and a preference for the breadwinner-homemaker family form persist (Florean & Engelhardt, 2020;Geist & Brauner-Otto, 2017;Konig & Cesinger, 2015).Hence, we contribute by exploring the role women's employment status plays in the relationship between work-family conflict and fertility within this unique country context.
Perhaps most importantly, we contribute by uncovering a previously unknown source of partner disagreement on fertility preferences: work-family conflict.This link between work-family conflict and fertility disagreement should not be underestimated, since the plausible negative outcomes resulting from it are diverse and far reaching.Because incongruent fertility intentions are associated with pregnancy avoidance (Testa, 2012;Testa et al., 2014;Testa & Bolano, 2021), while shared fertility intentions are linked to greater birth outcomes (Shreffler et al., 2019), the link between greater work-family conflict and fertility disagreement might further diminish Germany's low birth rate.Moreover, if work-family conflict continues to prompt greater fertility-related discord, couples could face higher amounts of reproductive coercion (Grace & Andersson, 2018) or even higher risk pregnancies (Hohmann-Marriot, 2009).This is especially problematic for working women, who face a larger opportunity cost associated with childbearing (Cavalli, 2012;Stein et al., 2014).Likewise, greater fertility disagreement due to incompatible work and family demands might impact the overall satisfaction, stability, and longevity of relationships, potentially resulting in greater separation and dissolution.Thus, the negative outcomes that could transpire when couples have greater disagreement on fertility preferences reaffirm the import of uncovering the correlates of this disagreement, as we do here.

Conclusion
In conclusion, the finding that work-family conflict is associated with greater partner disagreement on fertility preferences highlights the negative impact that challenges balancing work and family responsibilities have on couples' family planning.Considering the already low fertility rate in Germany, this research and its findings might indicate further difficulties in reaching the desired replacement level.Given that prior work documents a positive link between partners' agreement on fertility intentions and actual birth outcomes (Shreffler et al., 2019), future scholars might extend our findings by assessing whether the lower agreement resulting from work-family conflict results in fewer births as well.In addition, the stronger negative impact of one's own FTWC on their partner's reports of agreement on fertility preferences is only found among modernized male breadwinner couples.This suggests that dissymmetry in work hours between partners might create a stressful environment, such that the main burden of childcare and housework still falls on part-time working women and can be detrimental.Because Germany has a high percentage of women working part-time, this finding has important implications for policymakers.Future interventions should concentrate on quality, affordable, available childcare when encouraging German women to be more active in the labor force.Ultimately, by reducing work-family conflict, couples might establish more harmonious reproductive plans and lower fertility-related disagreement, thereby lessening the risk of experiencing potential negative outcomes associated with it.

Note:
WTFC=work-to-family conflict; FTWC= family-to-work conflict.Higher scores in WTFC and FTWC indicate higher levels of WTFC and FTWC, respectively.The indicators of the dependent variable are coded so that higher scores indicate more agreement with one's partner on fertility preferences.This study also controls for relationship duration (logged), relationship status, men and women's work hours, men and women's age (in categories), men and women's education, household income (logged), living in East or West Germany, having preschool children living in the household, and the number of children living in the household.The best-fitting model is highlighted in bold.+p<.10, *p < .05,**p < .01,***p < .001(two-tailed tests).

Figure 1 :
Figure 1:Unstandardized coefficients for the best fitting models in SEM for partners' agreement on fertility preferences among full-time dual-earner couples

Figure 2 :
Figure 2:Unstandardized coefficients for the best fitting models in SEM for partners' agreement on fertility preferences among modernized male breadwinner couples

Table 1 :
Descriptive statistics for all variables -WTFC=work-to-family conflict; FTWC= family-to-work conflict.The indicators of the dependent variable are coded so that higher scores indicate more agreement with one's partner on fertility preferences.Higher scores in WTFC and FTWC indicate higher levels of WTFC and FTWC, respectively.The asterisks indicate significant differences in each variable between full-time dual earner couples and modernized male-breadwinner couples, as estimated by independent t-tests (for continuous variables) and chi-square tests (for categorical variables).+ p<.10, * p<0.05, ** p<0.01, *** p<0.001 (two-tailed tests)

Table 2 :
Predictive model of partners' agreement on fertility preferences among full-time dual-earner couples

Table 4 :
Fit indices for nested models in multi-group analyses for partners' agreement on fertility preferences comparing full-time dual-earner couples and modernized male breadwinner couples