Taking leave from work after childbirth can be helpful for working parents to deal with the major challenge of caring for a newborn and combining work and family. Mothers and fathers in dual-earner couples do not face this challenge in isolation from each other but together as a couple. Thus, whether to take parental leave, and for how long, is an important decision many (expectant) dual-earner couples must make. Gender role attitudes may help to explain differences in parental leave decisions within couples. Past research has mostly investigated associations between individual gender role attitudes and individual leave decisions (e.g., Duvander, 2014; Hyde et al., 1993). Given that couples and the work and family life courses of both partners are interdependent (“linked lives,” Elder, 1994, p. 6), focusing on only one partner within a committed relationship represents a limited perspective (cf. Stertz et al., 2017). One dyadic longitudinal study showed that mothers’ parental leave decisions are influenced by their partners’ gender role attitudes (Stertz et al., 2017); however, they did not assess attitudes toward the father’s role. This might be the reason why no effects of mothers’ attitudes on fathers’ decisions had been found. Overall, the reciprocal influences between partners are largely neglected in the literature.

The current research aims to make two major contributions to the literature. First, it advances our understanding of the role of partner attitudes toward worker and parental roles in mothers’ and fathers’ parental leave decisions by considering interdependencies between couples both theoretically and methodologically. We use the term parental leave to refer to any period that a mother or father takes off from work after childbirth and do not restrict it to legal leave entitlements (see also Stertz et al., 2017). A parental leave decision is understood here as the decision to take leave or not after childbirth, including the leave length if taken. Second, it extends knowledge about the role of communality, which refers to the orientation toward social partners (Abele & Wojciszke, 2014; Bakan, 1966), in the association of one partner’s attitudes and the other partner’s parental leave decision. We assume that being orientated toward others, which is indicated by strong communal traits, is associated with being more susceptible to partner influences. We therefore propose that partner influences grow with higher communal traits in women and men. We conducted a longitudinal study with heterosexual dual-earner couples spanning from pregnancy until about 18 months after the birth of their first child and analyzed the data using the actor-partner interdependence model (APIM; Kenny & Cook, 1999; Kenny et al., 2006). We examined effects of expectant mothers’ and fathers’ gender role attitudes on their partners’ parental leave decisions, differentiating between attitudes toward (a) women’s parental role, (b) women’s worker role, (c) men’s parental role, and (d) men’s worker role as well as (e) traditional gender ideology about maternal employment. Finally, we explored whether communal traits strengthen the impact of one partner’s attitudes on the other partner’s leave length.

Theoretical Background

Understanding couple dynamics in decision-making requires a systemic perspective. All of us are part of larger social systems that influence our experiences and behavior. Applying Bronfenbrenner’s (1979) ecological system theory to the work-family interface, Voydanoff (2007) conceptualized work and family as microsystems characterized by distinct structures and social roles. At the same time, work and family represent a mesosystem, i.e., interrelated microsystems. Furthermore, exosystems must be considered, i.e., mesosystems including microsystems in which an individual does not participate, such as the partner’s work setting. Lastly, micro-, meso-, and exosystems are embedded in and influenced by the macrosystem of the cultural context. A macrosystem is characterized by opportunity structures and constraints, such as legal parental leave entitlements and gender-specific norms. The latter can be reflected in the attitudes and behaviors of individuals and their partners. In sum, when examining couples’ leave decisions from a systemic perspective, it is important to consider individual attitudes and personality traits, the attitudes and traits of key social partners, and contextual opportunities and constraints. In this study, we focus on the partner, who is also the child’s other parent, as the main person involved in parental decisions related to work and family.

Gender Role Attitudes and Couples’ Parental Leave Decisions

Gender role attitudes represent personal beliefs about the ideal behaviors and responsibilities for members of different gender groups, ranging from egalitarian to traditional. These attitudes coexist with specific roles defined by family relatedness and occupation and include beliefs about gender-segregated spheres. The traditional role for women includes being the homemaker with caretaking responsibilities, whereas the traditional role for men includes being the breadwinner (Eagly, 1987). Individuals occupy different social roles depending on life domains and their context (e.g., wife, mother). Both partners take on the worker role in dual-earner couples. The parental role is added after having a baby. As attitudes toward gender roles act as a lens through which individuals view their social world and make decisions (Davis & Greenstein, 2009) and as attitudes can predict behavior (e.g., Ajzen & Fishbein, 1977), we propose that they may influence the behavior of men and women in their roles in the work and family domains. Because parental leave decisions are associated with both the worker and the parental role, we assume attitudes toward gender roles to be particularly important predictors for both mothers’ and fathers’ parental leave decisions. Past studies with a focus on individuals and not couples confirmed associations between gender role attitudes and mothers’ as well as fathers’ parental leave decisions. Less traditional attitudes toward a mother’s responsibility for the household and child have been found to increase the likelihood that a mother will participate in the labor market after childbirth (Glass & Riley, 1998). In addition, women with less traditional attitudes were found to plan to work later in pregnancy and return to work earlier (Lyness et al., 1999). For fathers, previous research has found that it is more likely fathers with less traditional attitudes will make use of longer parental leave (Duvander, 2014; Hyde et al., 1993).

Less research has applied a dyadic perspective, whereby the impact of both partners’ gender role attitudes on both partners’ leave decisions in dual-earner couples is investigated. Qualitative interview studies on couples’ leave-taking decisions indicate the importance of the other partner’s gendered beliefs about parenthood (e.g., Aunkofer et al., 2018). To the best of our knowledge, only one previous quantitative study with heterosexual dual-earner couples examined the impact of both partners’ gender role attitudes on both partners’ parental leave decisions using a longitudinal design (Stertz et al., 2017). This study found an effect of fathers’ attitudes on mothers’ leave decision, but not vice versa. One reason why Stertz et al. (2017) found no effect of mothers’ attitudes on fathers’ leave decisions could be that attitudes toward men’s roles were not considered. Hence, a more differentiated measurement of gender role attitudes is required in research on couples’ leave decisions.

Attitudes Toward Gender Roles: A Differentiated Conceptualization

Most research on gender role attitudes treats gender ideology as one-dimensional, with more recent research providing evidence of multi-dimensionality (Pepin & Cotter, 2018). We propose a differentiation of attitudes toward the gender of the role occupant and toward roles defined by either family relatedness or occupation to gain a more comprehensive understanding of the impact of gender role attitudes on couples’ parental leave decisions. Adopting a social role theory perspective (Eagly, 1987; Eagly & Wood, 2012), we conceptualize traditional attitudes toward women’s parental role as beliefs that women should be the main caregiver of children, that caretaking is their most important responsibility, and that women’s place is in the home. We conceptualize traditional attitudes toward men’s parental role as beliefs that men are less well-suited to raise and care for children and should stay away from household tasks. Regarding women’s and men’s roles as workers, we conceptualize traditional attitudes toward women’s worker role as beliefs that women should not be the breadwinner or focus on their careers. Traditional attitudes toward men’s worker role are conceptualized as beliefs that men should be the breadwinner, focused on their careers, and that success at work should be the central goal in a man’s life.

A person’s attitudes toward social roles may vary in their degree of traditionalism (cf. Pepin & Cotter, 2018). We assume that a traditional attitude toward one of the four roles may or may not be accompanied by equally traditional attitudes toward the other roles. Regarding men’s roles, for instance, men today are expected not to be exclusively “breadwinners” but also to be involved in caring, and many men themselves aspire to be involved fathers (e.g., Gerson, 2010; Harrington et al., 2016). Accordingly, a traditional perspective on men’s worker role does not necessarily go hand in hand with an equally traditional perspective on men’s parental role. We can only distinguish between these different forms of gender-related attitudes if we assess them independently.

In the present study, a key reason for distinguishing between these types of attitudes is that they might exert differential effects on parental leave decisions. Regarding the mother’s parental leave decision, the father’s role beliefs related to women might be most important whereas for father’s parental leave decision, the mother’s role beliefs related to men might be most important. Theoretically, this can be explained by the principle of compatibility (formerly referred to as the principle of correspondence; Ajzen & Fishbein, 1977; Ajzen et al., 2018). According to this principle, attitudes predict behavior better if the specificity of a measured attitude matches the specificity of the behavior under consideration. In the cases described above, the role occupant’s gender with respect to the attitude that functions as predictor matches the gender of the person whose behavior in terms of leave uptake should be predicted. In addition, the father’s role beliefs related to family might be more important for the mother’s parental leave decision than those related to work because of societal expectations emphasizing the women’s prime responsibility for children’s well-being (Eagly, 1987). Regarding the father’s leave decision, by contrast, beliefs related to work might be more important than those related to family due to societal expectations and widespread gendered divisions of paid and unpaid work (Eagly, 1987). While associations between individual gender role attitudes and parental leave uptake have been repeatedly demonstrated, we focus on the following partner effects in our hypotheses (although actor effects are also modeled in the statistical approach we used):

Hypotheses 1a-d. Mothers with partners who hold more traditional attitudes toward (H1a) women’s parental role, (H1b) women’s worker role, (H1c) men’s parental role, and (H1d) men’s worker role would take longer parental leaves.

Hypotheses 2a-d. Fathers with partners who hold more traditional attitudes toward (H2a) women’s parental role, (H2b) women’s worker role, (H2c) men’s parental role, and (H2d) men’s worker role would take shorter parental leaves.

A schematic illustration of the effects hypothesized is given in Fig. 1. We also explored whether these types of attitudes exert differential effects (see above) and formulated the following research questions (RQs):

RQ1: Are fathers’ attitudes toward women’s parental and worker roles more important for mothers’ leave decisions than fathers’ attitudes toward men’s parental and worker roles?

RQ2: Are mothers’ attitudes toward men’s parental and worker roles more important for fathers’ leave decisions than mothers’ attitudes toward women’s parental and worker roles?

RQ3: Are fathers’ attitudes toward parental roles more important for mothers’ leave decisions than fathers’ attitudes toward worker roles?

RQ4: Are mothers’ attitudes toward worker roles more important for fathers’ leave decisions than mothers’ attitudes toward parental roles?

Fig. 1
figure 1

Schematic Illustration of the Hypothesized Partner and Moderation Effects

Full size image

In addition, we aimed to test whether the findings reported by Stertz et al. (2017; see above) can be replicated and considered attitudes related to an older concept of mother–child ideology based on two pillars: the belief in a special mother–child relationship and the associated irreplaceability of the mother and the belief in the detrimental effect of maternal employment on children’s development (Sieverding, 1990). In the following, we refer to attitudes toward such an ideology as attitudes toward early maternal employment. Attitudes regarding early maternal employment and a mother’s decision to take a leave from paid work for a certain period are corresponding attitudinal and behavioral entities. Consequently, based on the principle of compatibility and on previous empirical findings, we expected the following:

Hypotheses 3. Mothers with partners who hold more traditional attitudes toward early maternal employment would take longer parental leaves.

The Moderating Role of Communal Traits

We investigated the role of communal traits to explain why parents are influenced in their leave decisions to varying degrees by their partners’ attitudes. Communality implies a focus on others and is used to characterize people who are concerned primarily with social functioning and interpersonal relationships (e.g., Abele & Wojciszke, 2014; Bakan, 1966). We assume that other-orientation indicated by higher communal traits is associated with stronger susceptibility to partner influences. Therefore, higher communal traits might strengthen the relationship between one partner’s attitudes and the other partner’s leave decision. To the best of our knowledge, there is no study published that has investigated the influence of communal traits as a moderator in the association between one partner’s attitudes and the other partner’s work-family decisions. Consistent with social role theory (Eagly, 1987), there is meta-analytic evidence that women display higher communal traits than men (Hsu et al., 2021). Being higher in communality might make women generally more susceptible to partner influences than men. This might also help to explain unidirectional partner influences (here: from the man to the woman) shown in previous research on gender role attitudes and couples’ decisions related to work and family (Stertz et al., 2017). Based on these considerations, we explored whether women are more sensitive to partner influences than men when making decisions about parental leave. A schematic illustration of the interaction effects hypothesized can be found in Fig. 1. Here, we expected the following:

Hypotheses 4a. The association between the father’s attitudes and the mother’s length of leave will be moderated by the mother’s communal traits, with a stronger association observed for mothers with higher communality.

Hypotheses 4b. The association between the mother’s attitudes and the father’s length of leave will be moderated by the father’s communal traits, with a stronger association observed for fathers with higher communality.

Pilot Study

Prior to the main study, we conducted a pilot study to develop a multidimensional self-report instrument to measure attitudes toward gender- and domain-specific roles in terms of women’s and men’s roles as a worker and parent that allows us to measure these four roles distinctively. We could not find any published instrument that makes such a distinction for measuring attitudes towards gender roles. We aimed to develop a short scale of 20 items with five items per dimension (i.e., women’s parental role, women’s worker role, men’s parental role, men’s worker role) out of a larger 48-item pool and test its factor structure applying confirmatory factor analysis (CFA) and exploratory structural equation modeling (Asparouhov & Muthén, 2009; Marsh et al., 2014).

Method

Sample and Procedure

We conducted an online study in which participants were asked to indicate how much they agreed with 12 items related to each of the four roles. We paid attention to ensuring that each item referred to either men or women and either the family or the work domain, so that our proposed differentiation of gender- and domain-specific role attitudes could be tested empirically. When selecting items, we drew on those from established scales that matched our conceptualization. Information on the item sources and sample items are presented in Table 1. In the case that items did not refer exclusively to men vs. women or the family vs. work domain, we adapted them accordingly. We refrained from indicating a man’s or woman’s relationship status by not using terms such as “married” or “husband” as is the case in some scales published (e.g., Brogan & Kutner, 1976). Additional items were self-developed within the project group itself if not enough items could be identified in published instruments covering the conceptualization.

Table 1 Attitudes Toward Gender- and Domain-Specific Roles: Information on Item Sources, Number of Items and Items of the Final Version of the Scales (Main Study)
Full size table

We recruited participants through personal networks, social media, student email lists, and by advertising the study to companies, universities, and local institutions. Participants provided informed consent. As an incentive, they could win 3 x € 50 (~ US$ 56), and psychology undergraduates were offered partial course credit. We only included participants for this study who answered all 48 items on men’s and women’s roles as a parent and a worker (N = 1051). Of these participants, we excluded n = 36 participants who either filled out the questionnaire in half the time it took participants on average to complete it or needed less than 144 s to answer the 48 items on men’s and women’s roles as a parent and a worker (i.e., on average less than 3 s per item). We did so because it is questionable whether these participants had paid enough attention to the questionnaire. The analyses are based on N = 1015 participants (43.0% men, n = 436; 56.7% women, n = 436; 0.3% other; n = 3) with a mean age of 29.3 years (SD = 11.4). More than half of the participants (65.7%, n = 658) were in paid work. Most participants (67.6%, n = 665) were in a romantic relationship, and 20.5% (n = 203) had children.

Measures

Participants were asked to indicate their agreement with each of the 48 items on a Likert scale from 1 (= not at all) to 6 (= completely). Higher values represent traditional attitudes. All items were presented in German.

Statistical Analyses and Results

Before testing the factor structure, we determined the item difficulty Pi of each item (possible value range: 0 to 100; Moosbrugger & Kelava, 2020). The item difficulty of the items ranged from 2.4 to 62.8. We eliminated five items with an item difficulty below 5, indicating an extremely low level of agreement with these items among the participants. Based on the remaining 43 items, we conducted a CFA and a series of exploratory structural equation models (see Table 2 for information on the models and their fit indices). We used Mplus 8.7 (Muthén & Muthén, 1998–2017) for the analyses and maximum likelihood estimation with robust errors (MLR) for the specification of measurement models. In the comparisons of models, the model fit was assessed by the comparative fit index (CFI) and the Tucker-Lewis index (TLI), with higher values indicating a better fit, and the root-mean-square error of approximation (RMSEA), with lower values indicating a better fit. The first four models considered all 43 items and aimed at specifying and comparing different measurement models. We specified (1) a one-factor model with all items loading on a single factor; (2) a two-factor model separating (a) work- and (b) family-related statements on women and men; (3) a two-factor model separating work- and family-related statements (a) on women and (b) on men; and (4) a four-factor model separating (a) work- and (b) family-related statements on women, and (c) work- and (d) family-related statements on men. We specified the factors as correlated using an oblique target rotation in Model 2–4. According to the fit indices (see Table 2), Model 1, 2, and 3 did not fit the data well. Model 4 revealed a good fit. We continued with the specification of a four-factor model with correlated factors (i.e., Model 4) in the following models to optimize this model and reduce the number of items.

Table 2 Gender- and Domain-Specific Role Attitudes (Pilot Study): Measurement Models
Full size table

In the first optimization step, only items with a target loading > .3 and low non-target loadings (i.e., smaller than the target loadings) were kept. This led to the elimination of 11 items. The next optimization step was based on that model with 32 items (i.e., Model 5). We aimed at reducing the number of items to five items per factor and selected those five items per factor with the highest target loading (i.e., 0.37 – 0.77 [average loading of 0.62]), resulting in Model 6. We relied on the results of this scale development procedure in the main study, which we present below.

Main Study

We conducted a longitudinal study with heterosexual dual-earner couples during the transition to parenthood. This research was conducted in Germany, where mothers are generally not allowed to work during the first eight weeks after giving birth for health protection reasons. Both parents have the individual legal right to take an optional dismissal-protected parental leave of up to three years (Schober et al., 2023). With two major amendments, the German policy changed in 2007 to encourage more fathers to take leave. Firstly, greater financial compensation for salary losses for 12 out of 36 months (referred to as basic parental allowance) and, secondly, the introduction of the so-called partner months. That is, the parental allowance is paid for up to 14 months if both parents take at least 2 months of the basic 12 months (Schober et al., 2023). As intended, the reform has significantly increased the number of fathers taking leave. Currently, 44% of fathers take paid parental leave (Statistisches Bundesamt, 2022), but this also means that most fathers in Germany still decide against parental leave. By contrast, almost all mothers make use of paid parental leave in Germany (Statistisches Bundesamt, 2022) and, on average, take a leave of 14.6 months (see Schober et al., 2023). In addition, fathers are not only less likely to take parental leave than mothers, but fathers’ leave is generally shorter. Most leave-taking fathers (more than 70%) make use of only the two paid partner months (Statistisches Bundesamt, 2022).

Method

Participants and Procedure

To take part, women had to be pregnant, both partners had to be without previous children, and in paid work with a minimum of 18 weekly working hours. Participants provided informed consent. Participants were recruited at baby fairs, at information events for expectant parents, via advertisements on online platforms, and by advertising the study to people who are involved with expectant parents in their jobs (e.g., midwives) and asking them to hand out leaflets on the study to expectant parents. Measurement occasions took place during pregnancy (T1) and 3 (T2), 6 (T3), 12 (T4), and 18 (T5) months after childbirth. Participants could fill out the surveys online or in a paper–pencil version (which included postage-paid envelopes) and were instructed to answer the surveys on their own. Participating couples had the opportunity to take part in a lottery with ten cash prizes of € 150 (~ US$ 180) each at every measurement occasion.

We ran a Monte Carlo simulation (Muthén & Muthén, 2002) to determine the required sample size. Data in this simulation was generated from a population with predefined parameter values (here: assumed sizes of partner, actor, and moderator effects). We performed the estimation for a basic APIM (Cook & Kenny, 2005) with two independent variables (IVs) and two dependent variables (DVs), and one moderator variable. We assumed the following standardized effects to predict mothers’ and fathers’ parental leave length: partner effect on father’s decision = -.30; partner effect on mother’s decision = .30; actor effect mother = .30; actor effect father = -.30; communal-trait-interaction effect on mother’s decision = .20; and communal-trait-interaction effect on father’s decision = -.20. We estimated 10,000 models. A sample size of 250 couples turned out to be sufficient to keep the power constantly above the commonly accepted size of 0.80. Because some sample attrition is to be expected in longitudinal studies, we decided to recruit at least 300 couples to participate in the study.

N = 373 couples filled out the first survey. Information on parental leave length could be determined for at least one partner based on the data from the postnatal surveys for N = 365 couples; the analyses are based on these couples. Nearly all couples lived in Germany (98.4%, n = 359; other countries of residence: Belgium, Netherlands, Switzerland, n = 6) and 95.9% (n = 350) of both men and women had German citizenship. Expectant mothers were an average of 30.6 years old (SD = 3.7) and expectant fathers were an average of 33.1 years old (SD = 4.9). More than half of the women (59.2%, n = 216) and men (58.1%, n = 212) had an advanced university degree. The expectant mothers were, on average, in their 26th week of pregnancy (SD = 7.8) at T1. Couples had been in a relationship for an average of 7.5 years (SD = 7.0), 73.7% (n = 269) were married, and almost all were living together (98.9%, n = 361). Women’s average number of contractual work hours before childbirth was 37.8 h/week (SD = 4.6) and men’s was 39.1 h/week (SD = 3.5). The median of couples’ annual household income before childbirth fell within the range of € 80,000 – 120,000 [US$ 90,000 – 135,000]. Men contributed more to the household income before childbirth in 63.0% (n = 230) of couples, women earned more in 20.8% (n = 76) of couples, and both partners contributed about the same to the household income in 16.2% (n = 59) of couples.

Measures

Descriptive statistics and bivariate correlations of main study variables (all based on manifest variables) are displayed in Table 3. We used Mplus 8.8 (Muthén & Muthén, 1998–2017) and MLR for the specification of measurement models.

Table 3 Descriptives and Bivariate Correlations of Main Study Variables
Full size table

Gender- and Domain-Specific Role Attitudes

Participants’ attitudes toward men’s and women’s roles as a parent and a worker were assessed during pregnancy (T1) with the 20-item instrument developed in the pilot study. Items and information on the item sources are presented in Table 1. The items had to be rated on a Likert scale from 1 (not at all) to 6 (completely). Higher scores indicate more traditional attitudes.

We conducted a series of CFAs to examine the factor structure of the instrument. To improve the model fit, we inspected the standardized factor loadings and removed the items with the lowest factor loadings for each factor. The results of these analyses are presented in the Results section. In sum, tests of the factor structure indicated that models separating (a) work-related statements on women, (b) family-related statements on women, (c) work-related statements on men, and (d) family-related statements on men provided the best fit to the data. In the final model, the factor “attitudes toward women’s parental role” consisted of 4 items (αmothers = .89; αfathers = .84), the factor “attitudes toward women’s worker role” of 4 items (αmothers = .69; αfathers = .71), the factor “attitudes toward men’s parental role” of 3 items (αmothers = .75; αfathers = .81), and the factor “attitudes toward men’s worker role” of 5 items (αmothers = .87; αfathers = .87).

For replication purposes, we administered the same four-item instrument as Stertz et al. (2017) to measure attitudes toward early maternal employment (i.e., the “mother–child ideology”-scale by Sieverding, 1990) at T1 (αmothers = .80; αfathers = .77). The items had to be rated on a Likert scale from 1 (not at all) to 6 (completely). A higher score indicates more traditional attitudes. Sample item are “A pre-school child is likely to suffer if the mother works,” and “Mothers of small children, who pursue their professions full-time, do so at the expense of their children’s development.” The former item is also typically used in scales aimed at measuring gender roles in panel studies such as the International Social Survey Programme (www.issp.org). In addition to items referring to maternal employment and contrary to the “mother–child ideology”-scale, these scales usually also contain items referring to women and work, gendered separate spheres, and the (primacy) of men’s breadwinner role (cf. Braun, 1999).

Communality

We used the positive expressive/communal traits scale of the German version (GEPAQ; Runge et al., 1981) of the Personal Attributes Questionnaire (Spence & Helmreich, 1978) to measure participants’ communal traits at T1 (αmothers = .77; αfathers = .76). The scale measures the degree to which a person can be classified according to communal characteristics and involved eight item pairs (e.g., not at all understanding of others – very understanding of others; not at all helpful to others – very helpful to others; not at all aware of feelings of others – very aware of feelings of others; not at all emotional – very emotional). Participants indicated the degree to which they can be characterized regarding each item pair on a five-point scale from 1 to 5, representing the two extremes (e.g., not at all understanding of others = 1, very understanding of others = 5). High values represent high self-ascribed communal traits.

We examined possible differential item functioning between the subsamples of men and women in the measurement model of the communal traits scale by testing weak and strong invariance compared to configural invariance (i.e., no parameters constrained). Setting the factor loadings equal to test weak invariance showed that a model with weak invariance (CFI = 0.89; TLI = 0.88; RMSEA = .005) provided a similar fit to the data as a model with configural invariance (CFI = 0.89; TLI = 0.88; RMSEA = 0.06). Constraining loadings and intercepts of the items to test strong invariance worsened the model fit significantly (CFI = 0.83; TLI = 0.81; RMSEA = 0.07). Based on these comparisons, we included invariance constraints of the weak invariance model for modeling communal traits in the subsequent analyses.

Parental Leave Length

Mothers indicated their employment status at each postnatal survey. If a mother indicated for the first time in one of these surveys that she was working, the mother was asked about her first day back at work. We determined the length of a mother’s leave based on the information on her child’s date of birth and the date of her first day back at work. The leave length for women who had not yet returned to work at the last measurement point about 18 months after childbirth was calculated based on the information they provided about their planned first workday.

Fathers in Germany do not necessarily take leave immediately after the birth of their child. Therefore, they were asked if they took leave at each postnatal measurement and, if so, for the date of their first and last leave day. German legislation allows leave-taking parents to work part-time for up to 32 h per week (Schober et al., 2023). For this study, we only counted the periods when fathers did not work part-time while they were on parental leave. We calculated fathers’ total leave length based on the information they provided about their leave periods. The value zero in the leave length variable used in our analyses indicates that a father did not take any parental leave (i.e., 0 days of leave).

Control Variables

For the sake of comparability, we controlled for variables included as control variables in the study by Stertz et al. (2017), i.e., mothers’ and fathers’ number of working hours before childbirth, their educational level (i.e., holding a university degree: 0 = no; 1 = yes), couples’ financial situation, and main earner status before childbirth. In contrast to Stertz et al. (2017), we did not control for experiencing first-birth because our sample consisted only of first-time parents and not for the country of residence because almost all our participants lived in Germany. Regarding the financial situation pre-birth, we assigned couples to a low vs. average/high household income category depending on the median of the pre-birth income in our sample. Income categories at the median (i.e., € 80,000 – 120,000) and those above the median were classified as average/high income (= 1), incomes below the category in which the median fell were classified as low income (= 0). Participants were asked to indicate whether their annual gross income was equal to, higher, or lower than their partner’s income at T1. The main earner status was determined based on this information. For this purpose, we included two dummy variables (D1: 1 = higher paternal income, 0 = equal income, 0 = higher maternal income; D2: 1 = higher maternal income, 0 = equal income, 0 = higher paternal income).

Analytic Approach

We analyzed our dyadic data using the APIM (Kenny & Cook, 1999; Kenny et al., 2006) in a structural equation modeling framework with latent variables and MLR in Mplus 8.8 (Muthén & Muthén, 1998–2017) to test our hypotheses. We applied the latent moderated structural equations approach (Klein & Moosbrugger, 2000) to test multiple latent interaction effects. Continuous sociodemographic control variables were centered at the mean.

The APIM estimates the extent to which the DV for an individual is influenced by the score of the partner on the IV (i.e., partner effect). The extent to which the DV of an individual is influenced by the score of the same individual on the IV (i.e., actor effect) is simultaneously estimated. Consequently, applying the APIM is appropriate to test our hypothesized partner effects (H1a-d, H2a-d, H3) by considering the interdependence of the dyadic data. The APIM allows for the integration of moderator variables (Cook & Snyder, 2005), which enables us to test our hypotheses on the moderating effect of communal traits (H4a-b).

We specified the following models: in APIM 1, we tested the impact of mothers’ and fathers’ attitudes toward women’s parental role (IV 1a-b), women’s worker role (IV 2a-b), men’s parental role (IV 3a-b), men’s worker role (IV 4a-b), and toward early maternal employment (IV 5a-b) on their partners’ leave decisions. More precisely, in APIM 1a, we included all ten IVs (i.e., IV 1a-b – IV 5a-b), mothers’ and fathers’ length of leave (i.e., DVs), their communal traits (i.e., moderator variable), and the control variables and tested ten partner effects (i.e., the effects of ten IVs) in a first step. The model also included a total of 12 actor effects, i.e., the effects of mothers’ attitudes toward the five types of gender roles and of mothers’ communal traits on mothers’ own leave length and the effects of fathers’ attitudes toward the five types of gender roles and of fathers’ communal traits on their own leave length. We applied the Wald test of parameter constraints to explore differences in the strength of two significant partner effects. The models specified next based on APIM 1a included interaction terms. These models (ten in total) were used to test whether a partner effect is moderated by communal traits. Due to the complexity of APIM 1a, we chose not to test an even more complex model with a total of 10 interaction terms, but instead tested the hypothesized interactions in separate models. The same applies to the testing of interaction effects on the basis of APIM 3a (see below).

In ancillary analyses, we tested the influence of attitudes toward early maternal employment on couples’ parental leave decisions in models without IV 1a-b – 4a-b to ensure greater comparability of our analyses with those of Stertz et al. (2017). In APIM 2a, we used mothers’ and fathers’ attitudes toward early maternal employment as IVs and included their length of leave (i.e., DVs) and the control variables. In the subsequent model (APIM 2b), we added mothers’ and fathers’ communal traits (i.e., moderator variables) and the interaction terms to test whether partner effects are moderated by communal traits. Furthermore, we analyzed models with attitudes toward all types of gender roles as predictors except attitudes toward early maternal employment to determine the explanatory power of expectant mothers’ and fathers’ attitudes toward men’s and women’ worker and parental roles. APIM 3a included eight IVs (IV 1a-b – 4a-b) and the control variables. The models specified next based on APIM 3a included mothers’ and fathers’ communal traits (i.e., moderator variables) and interaction terms. These models (eight in total) were used to test whether a partner effect is moderated by communal traits.

Results

Gender- and Domain-Specific Role Attitudes

We investigated the factor structure of the self-report instrument measuring gender- and domain-specific role attitudes in our dyadic data set with one row for every couple with a series of CFAs. We started with separate analyses for the subsamples of women and men, followed by analyses considering the total sample. We tested the models regarding their fit to the data and compared their model fit indices under consideration of the CFI, the TLI, and the RMSEA. We tested four models each in the separate analyses for men and women: (1) a one-factor model; (2) a two-factor model separating (a) work- and (b) family-related statements on women and men; (3) a two-factor model separating work and family-related statements (a) on women and (b) on men; and (4) a four-factor model separating (a) work- and (b) family-related statements on women, and (c) work- and (d) family-related statements on men. We tested similar models in the analyses based on the total sample but, in each case, we considered participating men’s and women’s data in one model, resulting in twice as many factors in each model than in the analyses based on either the men’s or women’s subsample data. According to the model fit indices, the models separating (a) work-related statements on women, (b) family-related statements on women, (c) work-related statements on men, and (d) family-related statements on men (i.e., Model 4) provided a better fit to the data than all other models that have been specified in the subsample of women (CFI = 0.93; TLI = 0.92; RMSEA = 0.06), men (CFI = 0.90; TLI = 0.88; RMSEA = 0.07) and the total sample (CFI = 0.89; TLI = 0.88; RMSEA = 0.05). Model fit indices and χ2 difference tests statistics of all models tested are provided in Table 4.

Table 4 Gender- and Domain-Specific Role Attitudes (Main Study): Measurement Models
Full size table

Based on this model with the best fit (i.e., Model 4) and considering the total sample, we inspected the standardized factor loadings (SFLs) and removed the items with the lowest factor loadings for each factor (i.e., < 0.55 for a certain factor in either the men’s or women’s data) to further improve the model fit. This led to the deletion of one item assigned to the factor “attitudes toward women’s parental role” and the factor “attitudes toward women’s worker role” each, and two items assigned to the factor “attitudes toward men’s parental role” and resulted in the final model (i.e., Model 5). In Model 5, the factor “attitudes toward women’s parental role” consisted of 4 items (SFL: 0.65 – 0.92; Ø 0.79), the factor “attitudes toward women’s worker role” of 4 items (SFL: 0.50 – 0.68; Ø = 0.62), the factor “attitudes toward men’s parental role” of 3 items (SFL: 0.56 – 0.80; Ø = 0.73), and the factor “attitudes toward men’s worker role” of 5 items (SFL: 0.72 – 0.90; Ø 0.79). As intended, deleting these items resulted in an improved model fit (CFI = 0.93; TLI = 0.92; RMSEA = 0.05; ΔCFI = 0.036, ΔTLI = 0.036, ΔRMSEA = 0.002).

In our newly developed scales, attitudes towards women’s roles as parents and employees are not formulated in relation to a specific life phase, but more generally (see Table 1). By contrast, attitudes toward early maternal employment refer to the period when children are young. We conducted CFAs, including the items of the mother–child ideology scale and the items of the newly developed gender- and domain-specific roles scale, to test empirically whether the scale measuring attitudes toward early maternal employment can be distinguished from the scales measuring attitudes toward women’s parental role and women’s worker role. A ten-factor (i.e., [A] mothers’ attitudes toward (1) women’s parental role, (2) women’s worker role, (3) men’s parental role, (4) men’s worker role, (5) early maternal employment, and [B] fathers’ attitudes toward (1) women’s parental role, (2) women’s worker role, (3) men’s parental role, (4) men’s worker role, (5) early maternal employment) model provided a better fit to the data (CFI = 0.92; TLI = 0.91; RMSEA = 0.43) than eight-factor models with the items of the mother–child ideology scale loading on the same factor as the items of the attitudes toward women’s parental role scale (CFI = 0.74; TLI = 0.73; RMSEA = 0.08) and the items of the attitudes toward women’s worker role scale (CFI = 0.85; TLI = 0.84; RMSEA = 0.06).

We examined possible differential item functioning between the subsamples of men and women in the final ten-factor model by testing weak and strong invariance compared to configural invariance (i.e., no parameters constrained). Setting the factor loadings equal to test weak invariance showed that a model with weak invariance (CFI = 0.98; TLI = 0.97; RMSEA = 0.05) provided a similar fit to the data as a model with configural invariance (CFI = 0.98; TLI = 0.96; RMSEA = 0.05). Constraining loadings and intercepts of the items to test strong invariance worsened the model fit significantly (CFI = 0.95; TLI = 0.93; RMSEA = 0.08). Based on these comparisons, we favored the more parsimonious model with weak invariance and included corresponding invariance constraints in the subsequent analyses when applying the model.

Our analyses of the factor structure showed that the different types of gender role attitudes represent factors that can be distinguished from each other. In terms of construct validity, this is also supported by the pattern of bivariate correlations (see Table 3). Moderately positive correlations of attitudes toward early maternal employment and traditional attitudes toward women’s worker role, men’s parental role, and men’s worker role (rmothers = .25 – .30; rfathers = .25 – .34) indicate which attitudes are not covered by the older concept of mother–child ideology. By contrast, positive and rather strong correlations of attitudes toward early maternal employment and traditional attitudes toward women’s parental role (rmothers = .57; rfathers = .55) are indicative of convergent validity.

Descriptive Information on the Sample

Our participants generally endorsed egalitarian attitudes (see Table 3). T-tests indicated that men had more traditional attitudes toward women’s worker role, (t[364] = 2.97, p = .005, d = 0.15), men’s parental role (t[364] = 5.51, p < .001, d = 0.29), and early maternal employment (t[364] = 3.08, p = .002, d = 0.65) than women, whereas there were no significant gender differences regarding attitudes toward women’s parental role (t[364] = 1.28, p = .202, d = 0.07) and men’s worker role (t[364] = 1.49, p = .138; d = 0.08).

We analyzed in further t-tests whether there were mean differences in the manifestation of attitudes toward the different types of gender roles within individuals (see Table 5). Taken together, the findings indicate that attitudes toward different roles vary in their degree of traditionalism within individuals. Of the various attitudes, those toward men’s worker role and early maternal employment were the most traditional in relative terms in most comparisons. Attitudes toward men’s parental role were the least traditional in relative terms in all comparisons. In addition, the largest effect sizes (Cohen’s d ≥ 1.0) were found for comparisons including attitudes toward men’s parental role. This indicates that the difference in the degree of traditionalism is most pronounced when comparing this type of attitude with other types (e.g., Mmothers’ trad. attitudes toward men’s parental role = 1.27 (SD = 0.54), Mmothers’ trad. attitudes toward men’s worker role = 2.49 (SD = 1.14), d = 1.14). Regarding communality, an independent t-test indicated that, on average, women reported higher communal traits than men, t(363) = -7.88, p < .001; d = 0.41.

Table 5 Analyses of Mean Differences in Gender- and Domain-Specific Role Attitudes (Main Study)
Full size table

Mothers returned to work, on average, after 15.58 months (SD = 9.56; Mdn = 12.85 months). A total of 23.2% of fathers did not take any leave. In the group of fathers who took leave, their leave length ranged from one day to slightly more than 15 months (i.e., 464 days). The average leave length in this group was 82.17 days (SD = 75.93; Mdn = 60.00).

Hypothesis Testing

The bivariate correlations (see Table 3) indicate that mothers’ more traditional attitudes toward women’s parental role, men’s worker role, and early maternal employment were significantly and negatively related to fathers’ leave length, and that fathers’ more traditional attitudes toward the same types of gender roles were significantly and positively related to mothers’ leave length.

The hypothesized partner effects were tested in APIM 1a (i.e., a model without interaction terms). In addition, actor effects were analyzed in this model. Table 6 displays unstandardized and standardized coefficients, standard errors, p-values of IVs and moderator variables of APIM 1a.

Table 6 Actor–Partner Interdependence Model 1a: Actor and Partner Effects of Mothers and Fathers Attitudes and their Communal Traits on their Leave Length (N = 365 Couples)
Full size table

Partner Effects

We found a significant effect of fathers’ attitudes toward early maternal employment on mothers’ leave length (b = 3.56, p < .001). Mothers with more traditional partners took longer leaves. Fathers were influenced by the attitudes of their partners toward men’s worker role (b = -9.88, p = .026) and early maternal employment (b = -14.66, p = .012). Fathers with more traditional partners took shorter leaves. Thus, H1a-d, which predicted that mothers with partners who hold more traditional attitudes toward (H1a) women’s parental role, (H1b) women’s worker role, (H1c) men’s parental role, and (H1d) men’s worker role take longer leaves, were not supported (see Table 6). By contrast, the findings by Stertz et al. (2017) that mothers with partners who hold more traditional attitudes toward early employment take longer leaves were replicated by our findings and H3 was supported. Regarding partner effects, our findings indicated that not only mothers are influenced by their partners’ attitudes toward early maternal employment but also fathers. In addition, our findings supported H2d, suggesting that fathers with partners who endorse more traditional attitudes toward men’s worker role take shorter parental leaves. H2a-c suggesting that traditional attitudes by the partner toward (H2a) women’s parental role, (H2b) women’s worker role, (H2c) and men’s parental role would predict shorter paternal leaves were not supported. We tested for differences in the strength of both significant partner effects from mothers to fathers. The Wald test indicated no statistically significant differences (p = .513).

Actor Effects

Regarding the influence of mothers’ and fathers’ individual attitudes toward gender roles and their individual leave length, only the effect of mothers’ attitudes toward early maternal employment on their leave length reached significance (b = 2.19, p = .012). Mothers with more traditional attitudes took longer leaves. In terms of main effects of communal traits, we found a significant effect on mothers’ leave length (b = -3.68, p = .007), with mothers scoring higher on communal traits taking longer leaves. Fathers’ leave length was not influenced by their communal traits (b = 8.97, p = .442).

Effects of Control Variables

In terms of bivariate correlations, we identified significant associations between the control variables and both parents’ leave length (see Table 3). However, none of the control variables in the APIM had a significant influence on either mothers’ or fathers’ leave length.

Interaction Effects

Testing the interaction effects in further models showed that the partner effect of mothers’ attitudes toward men’s worker role and fathers’ leave length was moderated by fathers’ communal traits (b = -31.78, p = .010). This model explained 21.7% of the variance in fathers’ leave length and 29.3% of the variance in mothers’ leave length. All other interaction effects failed to reach significance (bs = -13.88 to 15.99, ps = .105 to .974; see Table 7).

Table 7 Results of the Tested Interaction Effects based on Actor–Partner Interdependence Model 1a (N = 365 Couples)
Full size table

We have illustrated the interaction graphically and calculated simple slopes (Aiken & West, 1991) to better understand this significant interaction effect. Figure 2 displays the interaction effect of mothers’ attitudes toward men’s worker role and fathers’ communal traits on fathers’ leave length. The levels of mothers’ gender role attitudes are depicted on the x-axis. A value of one standard deviation (SD) below the mean (= -1.14) indicates a lower level of mothers’ attitudes toward men’s worker role (i.e., more egalitarian attitudes), the mean value indicates a medium level, and a value of one SD above the mean (= 1.14) indicates a higher level (i.e., more traditional attitudes). Separate lines are used to denote lower (-1 SD; = -0.50), medium (= 0), and higher communal traits (+ 1 SD; = 0.50). The simple slope test reached significance (b = -22.20, p = .001) at a higher level of communal traits. The negative coefficient indicates a decrease in fathers’ leave length with mothers’ attitudes becoming more traditional. The test was not significant at a medium level (b = -8.83, p = .050) and a low level (b = 4.53, p = .529), indicating that for fathers with average and low communal traits, their leave length is not significantly affected by their partners’ attitudes toward men’s worker role.

Fig. 2
figure 2

Interaction Effect Between Mothers’ Attitudes Toward Men’s Worker Role (Centered at the Sample Mean) and Fathers’ Communal Traits on Fathers’ Leave Length

Full size image

Our results support H4b, which proposed that the relationship between mothers’ attitudes and fathers’ leave decisions is stronger when fathers’ communal traits are more pronounced, but only regarding mothers’ attitudes toward men’s worker role. Our assumption that the relationship between the father’s attitudes and the mother’s leave decisions is moderated by the mother’s communal traits (H4a) was not supported.

Ancillary Analyses

We conducted two sets of ancillary analyses. The first set aimed at ensuring greater comparability of our analyses with those of Stertz et al. (2017). The second set aimed at investigating the influence of couples’ attitudes toward gender- and domain-specific roles, measured with the newly developed scales, on their leave decisions independently of their attitudes toward early maternal employment. By and large, the ancillary analyses supported the main findings. When considering only attitudes toward early maternal employment as IVs (i.e., APIM 2a), we observed a partner effect from fathers’ attitudes on mothers’ leave length (b = 2.52, p = .001), a partner effect from mothers’ attitudes on fathers’ leave length (b = -15.02, p < .001), and an actor effect for mothers (b = 2.91, p < .001). The actor effect for fathers (b = 0.93, p = .847) and the interaction effects (b = 3.35, p = .813; b = 0.29, p = .888) did not reach significance. APIM 2a explained 12.8% of the variance in fathers’ leave length and 23.4% in mothers’ leave length. APIM 2b, which included mothers’ and fathers’ communal traits (i.e., the moderator variables) and interaction terms, explained 13.4% of the variance in fathers’ leave length and 25.7% of the variance in mothers’ leave length.

In APIM 3a, which included attitudes toward all types of gender roles except early maternal employment, we could confirm a partner effect from mothers’ attitudes toward men’s worker role (b = -9.76, p = .025). Contrary to the main analysis, we found a significant actor effect from mothers’ attitudes toward women’s parental role on their leave length (b = 2.84, p = .001), with more traditional mothers taking longer leaves. APIM 3a explained 15.5% of the variance in fathers’ leave length and 13.6% in mothers’ leave length. Testing the interaction effects in further models showed that, again, the partner effect of mothers’ attitudes toward men’s worker role and fathers’ leave length was moderated by fathers’ communal traits (b = -29.45, p = .016). The interaction followed the same pattern as in the main analyses. Again, the simple slope test reached significance at a higher level of communal traits (b = -22.97, p = .002) and not at a medium (b = -8.24, p = .063) and a low level (b = 6.48, p = .405). This model explained 18.8% of the variance in fathers’ leave length and 17.6% of the variance in mothers’ leave length. All other interaction effects failed to reach significance (bs = -.13.16 to 14.33, ps = .131 to .712).

Discussion

We examined in this study how dual-earner couples in heterosexual relationships influence each other’s parental leave decisions through their gender role attitudes, taking interdependencies between couples theoretically and methodically into account. Regarding these interdependencies, we found that both mothers and fathers were influenced in their parental leave decisions by selected gender role attitudes held by their partners. In addition, we found for fathers that particularly those with higher communal traits were influenced in their decisions by the attitudes of their partners toward men’s worker role. Mothers, by contrast, were not more susceptible to the influences of their partners’ attitudes as a function of the strength of their communal traits. Overall, our findings indicate that when considering the different types of attitudes we examined, attitudes toward early maternal employment seem to be most important for mothers’ leave decisions. Mothers’ attitudes toward early maternal employment also played a role in fathers’ leave decisions. In addition, fathers’ leave decisions were affected by mothers’ attitudes toward men’s worker role. Lastly, when considering the different types of attitudes, except for attitudes toward maternal employment, we found that mothers’ leave length was predicted by their attitudes toward women’s parental role.

RQ1 to RQ4 addressed the importance of different types of roles that can be classified in terms of gender (i.e., men, women) and life domain (i.e., work, family) for leave decisions of mothers and fathers. According to our results, fathers’ role beliefs related to women, family, and work seem to be relevant for mothers’ leave decisions. However, this seems to be particularly true if they are linked and relate to the early family phase, as reflected in attitudes toward early maternal employment. Contrary to our initial assumptions, mothers’ beliefs about men and the worker role do not appear to be more important for fathers’ leave decisions than beliefs about women and early maternal employment.

Theoretical Implications

The Interplay of Couples’ Gender Role Attitudes and Parental Leave Decisions

According to our analyses, the different types of gender role attitudes we considered can be distinguished from each other and vary in their degrees of traditionalism within individuals. Our study is thus in line with previous findings showing that gender role attitudes are multi-dimensional and that attitudes held toward different roles are not equally traditional (Pepin & Cotter, 2018).

When simultaneously considering (a) all types of attitudes and (b) the impact of both partners’ attitudes on each partner’s leave length, only a selection of attitudes predicted couples’ leave decisions, i.e., mothers’ and fathers’ attitudes toward early maternal employment and mothers’ attitudes toward men’s worker role. More traditional attitudes toward early maternal employment on the part of fathers resulted in their partners taking longer leave. More traditional attitudes toward early maternal employment and men’s worker role on the part of mothers resulted in their partners taking shorter leave.

Our results are in line with those by Stertz et al. (2017) by highlighting the importance of the father’s attitudes toward the harmfulness of early maternal employment for the mother’s leave decision. Both previous and present findings underline that this type of attitude on the part of the father is crucial for the mother’s leave. Women whose partners were more traditional in this respect acted in line with this attitude and took longer leaves than women with less traditional partners, who reentered the labor market earlier after giving birth. When looking at the beta coefficients of the effect, it is striking that they were almost identical in the two studies (beta = .34 vs. .35), suggesting that the effect is robust across two cohorts of parents.

Our findings differ from those of Stertz et al. (2017) regarding whether fathers are influenced by their partners, as this was not case in this earlier study. In the present study, fathers were influenced in their parental leave decisions by their partners’ attitudes, namely, by mothers’ attitudes toward early maternal employment and men’s worker role (the latter was not measured by Stertz et al., 2017). To better understand this seemingly contradictory results, it is important to highlight some key differences. There are approximately eight years between the data collection of the two studies (start of data collection: 2010 vs. 2018). In Stertz et al.’s (2017) study mainly couples from Switzerland (with very different statutory leave regulations) participated, followed by couples from Germany, whereas the sample of the present study consisted almost exclusively of couples from Germany.

As outlined in the beginning, since the 2007 parental leave reform in Germany, the uptake of parental leave by fathers has increased more than tenfold. There is greater overall variability in fathers’ leave decisions today than in previous decades. Furthermore, the lively public debate about parental leave for fathers has probably created a climate in which almost all expectant couples weigh up the pros and cons of both partners taking parental leave in view of their specific life situation. In other words, parental leave has increasingly become a partnership issue that affects both parents’ work-family decisions. The notion that parental leave in Germany has become more of a couple’s decision is empirically supported by the finding that fathers justify not having taken paid parental leave or having taken only a rather short paid parental leave by indicating that the mother wanted to take the larger share of the paid parental leave entitlement (Samtleben et al., 2019). Moreover, in our study, couples’ leave length was negatively correlated (r = -.26, p < .001), indicating that longer leaves by the mother go along with shorter leaves by the father and vice versa. This had not been the case (r = .04, p = .685) in the study by Stertz et al. (2017).

Taken as a whole, the significance of expectant mothers’ and fathers’ attitudes toward early maternal employment for couples’ parental leave decisions demonstrated in the current study can be explained by the principle of compatibility (Ajzen & Fishbein, 1977; Ajzen et al., 2018). According to this empirically proven principle (for a meta-analysis see Kraus, 1995), attitudes and behavior must match in terms of specificity to increase the predictor-criterion relationship (Ajzen & Fishbein, 1977). This study demonstrates the utility of the compatibility principle (a) for gender role attitudes and specific behavior of parents targeted to a bounded phase of the parental life course, i.e., taking parental leave (or not) in the early family phase, and (b) for investigating partner effects. The present study and previous research (Stertz et al., 2017) showed that a mother’s and her partner’s attitudes toward early maternal employment are essential for the mother’s behavior at the work-family interface in the first years of her child’s life, more precisely, for her decision on the length of her leave from paid work to care for her child. In addition, this study revealed that the mother’s attitude toward early maternal employment is also crucial for the father’s parental leave decision, highlighting the mutual interdependence of partners and of parents. Initially, we thought that attitude-behavior compatibility also applies to the relations between attitudes toward women’s parental and worker roles and couples’ parental leave decisions. On closer inspection, however, the more general conceptualization of these roles and the fact that they accordingly do not refer to the early family phase (see Method section and Table 1) represents a poorer correspondence between attitudes and target behavior.

For fathers, attitudes toward the role traditionally associated with men, i.e., the breadwinner role, predicted their work-family decisions in the early family phase in terms of fulfilling this role (by not taking leave) vs. not fulfilling it (by taking leave). The principle of compatibility is also clearly recognizable here. Although attitude-behavior compatibility is also evident in the relationship between attitudes toward men’s parental role and uptake of parental leave, attitudes toward this type of gender role were not related to either fathers’ or mothers’ behavior. This demonstrates the persistent and consequential nature of attitudes regarding men’s particular “fit” for the breadwinner role (regardless of what people think of men as caregivers). It also ties in with previous findings that indicate the “stickiness” of conventional ideologies about gender as outlined by Pepin and colleagues (Dernberger & Pepin, 2020; Pepin & Cotter, 2018). Our study and the study by Pepin and Cotter (2018) have also shown that attitudes toward early maternal employment remain relevant and relatively traditional. This, in turn, speaks in favor of the persistence of attitudes related to women’s comparative advantage as parents – particularly in the early family phase – despite growing egalitarianism regarding women’s worker roles.

Understanding the Role of Communal Traits

We explored whether communal traits strengthen interindividual relationships of gender role attitudes and parental leave decisions and hypothesized that communal traits would increase the susceptibility to partner influences. We detected such an effect, but only for the influence of mothers’ attitudes toward men’s worker role on fathers’ leave decisions. If the mother held egalitarian attitudes toward men’s worker role and the father scored higher on communal traits, fathers acted more in line with their partners’ attitudes. These fathers dedicated more time to the family at the expense of dedicating time to paid work by taking longer leaves. Fathers in such couple constellations took the relatively longest leaves (see Fig. 2). By contrast, if the mother held more traditional attitudes toward men’s worker role and the father scored higher on communal traits, fathers acted in line with their partners’ attitudes and took shorter leaves. Fathers in such couple constellations took the relatively shortest leaves (see Fig. 2). If fathers’ communal traits were less pronounced, however, their leave decisions did not depend on their partners’ attitudes toward men’s worker role.

Regarding the significant influence of mothers’ attitudes toward early maternal employment on fathers’ decisions, we did not find a stronger susceptibility as a function of fathers’ communal traits. This means that fathers generally refrained from taking (longer) parental leave and fulfilled their provider role if their partners believed that early maternal employment was harmful to children’s development. It might be that partners’ attitudes toward men’s worker role in contrast to their attitudes toward early maternal employment had a more nuanced influence on fathers’ decisions because they are more closely associated with themselves as men and their breadwinner role, which represents a central aspect of masculinity (Eagly, 1987; Thébaud, 2010). According to social role theory (Eagly, 1987), communal traits reflect femininity. When fathers score high on communal traits, suggesting that they are more emotional, empathic, and oriented toward others, the way they perform the worker role is shaped by their partners’ preferences about this role. For these fathers, it may be less relevant to shield themselves from the influence of others in decisions that might affect their masculinity.

We did not find a significant interaction between mothers’ communal traits and fathers’ attitudes toward early maternal employment. This means that the main effect of fathers’ attitudes on mothers’ leave length was evident for all mothers (i.e., independent of how communal a mother was). All mothers were responsive to partner influences and their leave length increased with the degree of traditionality of the father’s attitudes on a continuum from more egalitarian to more traditional. In other words, mothers with less traditional partners regarding early maternal employment acted in line with their partners’ attitudes by reentering the labor marker earlier after giving birth, while mothers with more traditional partners stayed at home longer. Both longer and shorter leaves can be interpreted as selfless and thus in line with gendered ideals if mothers view their decision as a decision they made for the family – which mothers typically do (Damaske, 2011). This might explain why no interaction effect between mothers’ communal traits and fathers’ attitudes could be detected. Nevertheless, there were constant level differences throughout the continuum in mothers’ leave lengths depending on mothers’ communal traits. More communal mothers always took the longest leaves on average, and less communal mothers always took the shortest leaves on average.

According to our findings, fathers’ communal traits per se were not associated with fathers’ parental leave decisions. This finding is not consistent with results from experimental studies on perceptions of fictitious fathers by others, which show that taking parental leave leads to the attribution of more pronounced communal traits (Fleischmann & Sieverding, 2015; Rudman & Mescher, 2013). In accordance with social role theory (Eagly, 1987), the authors of these experimental studies assume that fathers who take parental leave are seen as more communal than fathers who do not take leave and, thus, suffer a femininity stigma for gender-role deviance. In such studies, participants evaluate men based on vignettes or CVs, and the use of parental leave is experimentally manipulated. Our study shows that communal traits are not reflected that easily in the parental leave decisions made by “real” fathers. And, depending on partners’ attitudes toward men’s worker role, more communal fathers might even take shorter leaves if this corresponds with their partner’s preferences. This underlines how important it is to complement experimental studies on perceptions by others with longitudinal field studies bringing fathers’ actual experiences and behaviors to the fore.

Study Strengths

The present study has several strengths. First, the data was collected using a longitudinal design. In contrast to retrospective studies, this prospective approach allows us to capture attitudes and decisions as they unfold over time, thereby ensuring accuracy and external validity. Second, the study design is not only longitudinal but also dyadic. Third, the analyses are based on a large sample of 365 couples (= 730 individuals) during the transition to parenthood. These methodological strengths ensure the high external validity of this field study. Fourth, the use of the APIM (Kenny et al., 2006) as the appropriate method for analyzing the dyadic data and fifth, the conceptualization and operationalization of gender role attitudes as multidimensional, allowed us to investigate the predictive power of couples’ attitudes toward different types of gender roles on both parents’ leave decisions, contributing to a deeper understanding of the significance of attitudes toward specific roles for dual-earner couples’ work-family decisions and the (stalled) gender revolution. It also allowed us to examine the impact of communal traits on mothers’ and fathers’ leave decisions and their potential moderating role in the form of strengthening partner effects. Finally, examining the role of psychological influences for couples’ leave decisions beyond socio-structural influences is another strength to be emphasized.

Limitations and Future Research Directions

We focused on heterosexual couples in this study to tie in with the study by Stertz et al. (2017) and expand the knowledge about the impact of attitudes toward gender roles and communal traits on heterosexual couples’ parental leave decisions. The restriction to heterosexual couples limits our study because we cannot generalize to same-gender or other mixed-gender couples. Therefore, future research on attitudes toward different types of gender roles and couples’ parental leave decisions should consider other couples to broaden the knowledge about couple dynamics. In addition, our study considered only first-time parents. Further research is needed to investigate whether the effects found in the present study are also seen in couples who are already parents and have another child. These mothers and fathers know what it is like to parent a newborn from their previous experiences and their attitudes might have changed over time in response to childbirth and their external realities (e.g., employment status as a parent; Zhou, 2017). Consequently, the significance of gender role attitudes for work-family decisions of experienced parents may be different.

Due to the research design, the last measurement point took place 18 months after childbirth. Parents can take parental leave beyond this period in the country in which the study was conducted (i.e., Germany), and some of the mothers participating in our study did so. In these cases, only information on the expected total leave length could be considered in our analyses. Macrosystemic characteristics (e.g., legal parental leave regulations, gender-specific norms) can affect the use of parental leave by mothers and fathers. Such contextual influences may also translate into gender-specific partner effects (i.e., one-sided from mothers to fathers or vice versa). Accordingly, future research should examine the influence of couples’ attitudes toward gender roles on their parental leave decisions in contexts other than Germany.

Our study revealed that attitudes toward early maternal employment and men’s worker role are particularly important for couples’ decisions targeting a specific phase of the parental life course, i.e., the early family phase, as they show a particularly high attitude-behavior consistency. When it comes to other decisions related to work and/or family, which target different life phases (e.g., deciding who stays at home when a sick child cannot attend school, deciding which partner’s career is given priority, deciding whether or not to retire at a certain age), other specific gender role attitudes (e.g., attitudes toward men’s parental role, women’s parental role, and women’s worker role) may gain in importance. This should be examined in future research. Furthermore, we focused on communality as personality trait that implies a focus on others in order to explain why parents are influenced in their leave decisions to varying degrees by their partners’ attitudes. According to our findings, only the relationship between mothers’ attitudes toward men’s worker role and fathers’ leave length was moderated by fathers’ communal traits. Further research should investigate the role of communality as a moderator for the association between one partner’s specific gender role attitudes and the other partner’s work- and/or family-related decisions apart from parental leave decisions to obtain a wider picture of the role of communality for the susceptibility to partner influences. Finally, besides communal traits, other traits (e.g., agentic traits) but also other types of attitudes (e.g., benevolent and hostile sexism) related to power dynamics within couples and household decision-making are important in this context and should be considered as possible moderators in future research.

Practice Implications

This research contributes findings of practical relevance for couples, society, career and family counselors, and political decision-makers. Parental leave entitlement can help dual-earner couples to maintain balance between work and family during the transition to parenthood (Wong, 2023). Our findings show that one partner’s attitudes matter for the other partner’s leave decision in heterosexual dual-earner couples, thereby, highlighting that actual parental leave uptake should be regarded from a couples’ perspective. This is true at least for contexts where mothers and fathers are entitled to parental leave, and parental leave of fathers is an increasingly important and socially discussed issue. Of relevance for couples’ parental leave decisions are attitudes with a clear correspondence between the attitude and the behavior that results from these decisions, namely attitudes toward early maternal employment and men’s role as worker. In terms of traditional attitudes, these attitudes simultaneously reflect the classical roles of women and men, namely the caregiver and the breadwinner role. Expectant mothers’ and fathers’ individual attitudes and, therefore, couples’ leave decisions, are influenced by beliefs embedded in society. According to social role theory (Eagly & Wood, 2012), cultural meanings associated with biological sex for men and women influence their self-concept and, thus, form their gender identities. In addition, others’ stereotypic expectations and being rewarded for conforming to gender roles shape men’s and women’s behaviors. Consequently, socially shared beliefs on men and women can contribute to a more traditional but also more equal division of paid work and unpaid work in terms of caregiving among couples. This very much depends on whether the dominance of classic roles, which suggest a certain advantage of women over men and vice versa in the caregiver and breadwinner role, persists.

Career and family counselors should know about interdependencies between couples and the importance of certain attitudes for couples’ parental leave decisions. These interdependencies must be considered, even if the counseling can only be done with one of the partners. Finally, our findings are relevant for policymakers who need to understand parents’ decisions regarding the use of public family policies. When evaluating and adapting parental leave legislation, it is important to keep in mind that the leave decisions of both partners are interrelated.

Conclusion

Our study confirmed earlier research that mothers are influenced in their leave decisions by their partners’ attitudes toward early maternal employment. Further, we have shown that fathers nowadays are also influenced in their parental leave decisions by their female partners’ attitudes toward early maternal employment and men’s worker role, and that both parents’ leave decisions are interrelated. Another important takeaway from this research is the usefulness of the principle of compatibility (e.g., Ajzen et al., 2018) for predicting behavior through gender role attitudes not only on the within-person level, but also across partners. Influences on mothers’ and fathers’ parental leave decisions are complex and multifaceted. This study contributes to a better understanding of couples’ parental leave decisions by adopting a psychological perspective and considering interdependencies between couples. Studying couples and doing so longitudinally is a challenging endeavor, both theoretically and methodologically, but this study shows that it is worth taking this path.