1 Introduction

During the COVID-19 pandemic, the share of workers working from home increased considerably across industrialized countries (OECD, 2021a, 2021b), a trend researchers expect to last far beyond the pandemic (Barrero et al., 2021; Phillips, 2020). Therefore, the debate on the various consequences of working from home—not only for workers, but also for organizations, society, and the environment—has gained considerable momentum. One potentially strong consequence of working from home is the fit between employees’ work roles and personal lives. Accordingly, this paper explores whether working from home reduces or exacerbates work–family conflict (hereafter “WFC”), that is, incompatibilities between the work and family roles (Greenhaus & Beutell, 1985).

As work and family life overlap spatially when individuals work from home, this work mode may result in less commuting and more schedule flexibility. However, the possibility of caring for one’s household and family at any time during one’s workday may also lead to more role blurring, excessive involvement in either or both roles, and more strain spilling over from one role to the other, thereby increasing WFC. Overall, the evidence on this relationship is mixed: while the literature mostly finds reduced WFC among home workers (Allen et al., 2013; Gajendran & Harrison, 2007; Laß & Wooden, 2023), other studies have documented increased WFC in similar samples (Abendroth & Reimann, 2018; Kim et al., 2020; van der Lippe & Lippényi, 2020).

One question that has not yet been considered in the literature is whether the effect of working from home on WFC depends on individuals’ perceived salience of their work and family roles. In this context, salience describes the relative importance assigned to a specific role, that is, the degree to which one role stands out from the individual’s other roles (Super, 1982). The more salient a specific role is for an individual, the more likely he or she will perform according to that role and view a given situation as an opportunity to perform in that role (Stryker, 1968). Thus, workers whose work roles are more salient than their family roles may be more likely to let their work life interfere with their family life, and thus experience higher work-to-family conflict (hereafter “WTFC”). By contrast, workers who place more importance on their family roles may put more effort into fulfilling family-related demands, thereby increasing family-to-work conflict (hereafter “FTWC”). Given the spatial overlap of work and family life when working from home, the level of role salience may be of particular importance for WTFC and FTWC in this work mode.

In light of this research gap, we use data from the 12th wave of the German Panel Analysis of Intimate Relationships and Family Dynamics (pairfam) to investigate the moderating role of employees’ work-role versus family-role salience in the effect of working from home on WFC. In order to gain a holistic picture, we look at both directions of conflict, WTFC and FTWC. We also go beyond previous studies by considering the frequency of working from home instead of relying on simple binary measures (i.e., whether or not a respondent works from home). Furthermore, we compare the effects for women and men to see whether there are gender differences in these relationships. Finally, we also separately investigate mothers and fathers to see whether these groups, who often have particularly high family demands, are more likely to experience WFC when working from home, and whether the moderating effect of work- versus family-role salience is more crucial for parents.

Germany is an interesting case study. In terms of gender regimes, it is dominated by the modified male breadwinner model, where men act as primary earners and usually work full-time, whereas women work part-time and take primary responsibility for housework and care (Trappe et al., 2015). These gender roles are likely to impact workers’ hierarchy of the work and family roles (Meeussen et al., 2016; Wood & Eagly, 2009). Furthermore, despite recent policy reforms, such as an expansion of subsidized childcare for children under the age of 3, the reconciliation of work and family duties is still challenging for many parents in Germany, partly due to a still-insufficient number of childcare places (Adema et al., 2017; Laß et al., 2023). The question of how flexible workplace arrangements, including working from home, may facilitate or impede a good fit between work and family responsibilities is therefore of great importance in Germany.

However, prior to the pandemic, working from home was practiced only by a small minority of German employees: in 2019, only 5.2% of employed persons aged 15–64 years usually worked from home, and another 7.4% sometimes worked from home, placing Germany below the EU-28 averages of 5.3 and 10.8%, respectively (Eurostat, 2021). Furthermore, in contrast to most prior literature on this topic, studies conducted in Germany have frequently associated working from home with increased WFC (Abendroth & Reimann, 2018; Bellmann & Hübler, 2020; Lott, 2017; Schwarz et al., 2023). Our study thus sheds light on whether the drawbacks of working from home for work & family life fit are spread equally across workers, or whether it depends on workers’ levels of role salience.

2 Theoretical Framework

Prime-aged adults often hold a number of different roles simultaneously, such as worker, spouse, parent, and homemaker, each connected to specific expectations and obligations (Super, 1980). As emphasized by both Role Strain Theory (Goode, 1960) and the Stress Process Model (Pearlin, 1989), it can be difficult to meet the expectations from these different roles simultaneously, creating strain and stress for the individual. In this context, the term work–family conflict designates the form of interrole conflict that arises when demands from one’s work and family roles are mutually incompatible (Greenhaus & Beutell, 1985). This conflict is bi-directional in nature, in that work demands can interfere with family demands (WTFC) and family demands can interfere with work demands (FTWC). This can arise from time and energy constraints, but also when strain stemming from one sphere interferes with one’s role obligations in the other sphere. Furthermore, according to the Job Demands-Resources Model (Bakker & Demerouti, 2007; Demerouti et al., 2001), performance in (multiple) roles is also affected by the resources available to workers, such that higher resources (e.g., job autonomy) may lead to reduced work-home conflict (Bakker et al., 2011).

2.1 Working From Home and Work–Family Conflict

Whether working from home serves as a resource or just another work demand has been the subject of debate (Abendroth & Reimann, 2018). On one hand, working from home may save commuting time (Melo & Abreu e Silva, 2017; Laß & Wooden, 2023) or yield a more flexible work schedule (Crosbie & Moore, 2004; Hill et al., 1996; Wöhrmann & Ebner, 2021). On the other hand, the lack of physical boundaries between the work and the family spheres may result in one sphere taking up time or energy that was supposed to go toward the other sphere. For example, working from home may lead to long working hours and overtime (Abendroth & Reimann, 2018; Dockery & Bawa, 2014; Peters & van der Lippe, 2007), with working hours extending into evenings, nights, and weekends (Laß & Wooden, 2023).

The results from prior research on this question are far from consistent. On one hand, the three meta-analyses conducted to date (Allen et al., 2013; Beckel et al., 2023; Gajendran & Harrison 2007), which cover mostly small-sample North American studies, found beneficial outcomes of working from home. More precisely, they all found significant negative associations between working from home (or at other remote locations) and WTFC, along with either negative associations or no associations between working from home and FTWC. Also, recent studies based on nationally representative Australian data linked working from home to improved work-life balance and reduced WTFC (Dockery & Bawa, 2014; Laß & Wooden, 2023).

On the other hand, several studies based on German large-scale datasets reported that working from home was associated with both increased WTFC and FTWC (Abendroth & Reimann, 2018; Bellmann & Hübler, 2020; Lott, 2017; Schwarz et al., 2023). Apparently, among German home workers, the drawbacks of working from home, such as blurred boundaries between work and family life, outweigh its potential advantages, such as more flexible work schedules and reduced commuting.

This discrepancy may arise because, as mentioned earlier, working from home was only practiced by a small minority of German workers prior to the COVID-19 pandemic (Eurostat, 2021). These were often workers with high work and/or family demands who worked from home specifically to accommodate these demands. For example, data from the Mikrozensus showed that in 2018, managing directors had the second-highest prevalence of working from home with 37%, only superseded by teachers with 60% (Bujard et al., 2020). Furthermore, Abendroth and Reimann (2018) found that those working from home on average reported more demanding workplace cultures and more frequent overtime hours, both of which were positively associated with WTFC. They also found home workers were more likely to be married, have several children, and have younger children on average—characteristics that are likely to increase FTWC. Correspondingly, Lott (2019) showed that home workers (of either gender) did more overtime than workers only working onsite, and that female home workers additionally invested more time in childcare.

Despite the contradictory evidence from prior research, given the prior findings for German workers, we put forward the following hypotheses:

H1a

Working from home will be associated with increased work-to-family conflict.

H1b

Working from home will be associated with increased family-to-work conflict.

2.2 Role Salience as a Moderator Between Working From Home and Work–Family Conflict

Identity theory (e.g., Stryker, 1968; Stryker & Burke, 2000) posits that the self is constructed from several parts or identities, that is, the meanings individuals attach to the multiple roles they play in society. Whereas social roles include expectations linked to the positions one holds in a network of relationships, identities can be described as internalized role expectations. An individual’s different identities, such as parent, spouse, or employee, are not all equally important, but instead are ordered in a “hierarchy of salience” (Stryker, 1968, p. 560). Similarly, Donald Super defined role salience as the relative importance an individual assigns to a specific role, that is, the degree to which one role stands out from the individual’s other roles (Super, 1982). The importance of each role varies as individuals move through different stages of their life course (Super, 1980).

The salience an individual gives to his or her work and family roles can be expected to affect the level of conflict between these roles. However, the literature provides opposing arguments regarding the direction of this effect. Stryker (1968) emphasized that the more salient a specific role (or identity) is for an individual, the more likely the individual will perform according to that role, view a given situation as an opportunity to perform in that role, and seek out opportunities to perform in that role. Correspondingly, other scholars have argued that the higher the salience of a specific role, the more time the individual will assign to that role and the more strain the individual will experience from the role, leading to more interference between this role and other role obligations (Frone et al., 1992; Greenhaus & Beutell, 1985).This is consistent with the within-domain perspective (Frone et al., 1992), which specifies that work domain antecedents matter mostly for WTFC, whereas family domain antecedents matter mostly for FTWC. This is expected, as WTFC originates in the work domain (i.e., is caused by work-related demands) and FTWC originates in the family domain (i.e., is caused by family demands) (French et al., 2018). Combining these arguments, this perspective would suggest that individuals with higher work-role salience—who invest a great amount of time, energy, and attention in their work roles—would experience higher WTFC, while those with higher family-role salience would experience higher FTWC.

However, an alternate perspective predicts the opposite effect: although high role salience may lead an individual to channel more time and energy toward the highly salient role, the individual may simply not perceive this imbalance as problematic. For example, Noor (2004) argued that for workers with high family salience, the intrusion of family demands into the work sphere is acceptable, whereas the intrusion of work demands into family life is not tolerated. Likewise, Carlson and Kacmar (2000) suggested that individuals will be more likely to perceive factors from the less salient sphere as causing the conflict. Accordingly, individuals with higher work-role salience would experience higher FTWC, while those with higher family-role salience would experience higher WTFC. This argument is also consistent with the cross-domain perspective (Ford et al., 2007), which argues that the effects of factors in one domain will spill over into other domain, influencing individuals’ experiences and behavior in the other domain.

Although both perspectives are theoretically plausible, thus far the first perspective has received the most empirical support. Many studies indeed found WTFC to be more pronounced for those with higher work-role salience or involvement (e.g., Beutell & Wittig-Berman, 1999; Chrisangika Perera & Kailasapathy, 2020; Frone & Rice, 1987; Greenhaus & Kopelman, 1981; Wiley, 1987). By contrast, most studies found no effect of family-role salience on FTWC (Beutell & Wittig-Berman, 1999; Chang et al., 2014; Chrisangika Perera & Kailasapathy, 2020; Parasuraman et al., 1996). However, the cited studies usually investigated work-role salience and family-role salience as separate constructs. Far fewer studies have gauged the effects of the relative importance of work versus family roles on WFC, and even that research has generated mixed findings (Cinamon & Rich, 2002; Erdogan et al., 2021; Kuntari et al., 2017). For example, Cinamon and Rich (2002) found that those who value their family role more highly than their work role experience lower WTFC than those who value their work role more highly, or those who value both roles highly. By contrast, Erdogan et al. (2021) found WTFC and FTWC to be relatively low for workers either with predominant family salience or with predominant work salience, whereas those who had dual high salience or dual low salience experienced both higher levels of WTFC and FTWC.

It has also been posited that role salience should moderate the link between work or family demands and WFC. Greenhaus and Beutell (1985) argued that if an individual assigns high salience to the work (family) role, work-related (family-related) demands may be exacerbated because succeeding in that role is particularly important to that person. This suggests that high work-role salience should amplify the positive effect of job demands on WTFC, whereas high family-role salience should amplify the positive effect of family demands on FTWC.

So far, only a few studies have empirically investigated this moderating effect of role salience for WTFC or FTWC, focusing on a number of specific work demands/stressors: quantitative workload, frequency of job stress, schedule irregularity, yingchou (that is, the frequency of socializing for work purposes), job resources (e.g., job control, supervisor support), and family stressors (e.g., marital tensions, lack of child care) (Chang et al., 2014; Day & Chamberlain, 2006; Fox & Dwyer, 1999). To the best of our knowledge, no previous study has investigated role salience as a potential moderator with respect to the link between working from home and WFC. However, there are good theoretical reasons to expect role salience to particularly moderate the effects of working from home.

Specifically, Stryker (1968) highlighted that the hierarchy of salience should matter for individuals’ behavior only in situations of “structural overlap” (p. 560), i.e., when several roles are mutually contingent at the same point in time, and therefore several identities are invoked. Working from home is likely to cause such a structural overlap between an individual’s identities as worker and family member, as work and family life co-occur in space and possibly also in time. We therefore assume that individuals with high work-role salience will be particularly likely to utilize working from home to accommodate work demands. This will lead them to devote more time and energy to the work sphere than they would do if they worked exclusively onsite, resulting in higher levels of WTFC. Similarly, workers with high family-role salience should be more likely to use working from home to accommodate family demands, thereby investing more time and effort into the family sphere than they would otherwise, and consequently experiencing increased levels of FTWC. We thus put forward the following hypotheses:

H2a

The positive association between working from home and WTFC will be stronger for individuals with higher work-role salience compared to individuals with lower work-role salience.

H2b

The positive association between working from home and FTWC will be stronger for individuals with lower work-role salience compared to individuals with higher work-role salience.

2.3 Gender Differences Regarding the Moderating Effect of Role Salience

Germany is characterized by the male breadwinner/female part-time carer model: men usually work full-time regardless of their family situation, whereas women are the primary carers and usually work only part-time if they have family responsibilities (Pfau-Effinger & Sakac Magdalenić, 2009; Trappe et al., 2015). This difference in work hours between men and women further reinforces the gender inequality present in unpaid work; German women spend considerably more time on housework and care than men do (Adema et al., 2017; Altintas & Sullivan, 2017; Hook, 2006; Sullivan et al., 2018). The share of home workers is similar for both genders in Germany, with 11.6% of women and 12.9% of men spending some of their working days at home in 2019 (authors’ calculations based on Statistisches Bundesamt, 2020). Nevertheless, as a consequence of the different gender role expectations, women and men tend to utilize working from home for different purposes: women are more likely to work from home to combine work and family duties, and they may spend more time on housework and care when working from home, whereas men are more likely to work from home for the sake of their work performance (Lott, 2019; Powell & Craig, 2015; Sullivan & Smithson, 2007). The level of WFC experienced when working from home also varies by gender. For example, Yucel and Chung (2023) found working from home to be associated with both increased WTFC and increased FTWC only among women, and Schwarz et al. (2023) found the association between working from home and WTFC to be significantly stronger for women than men.

Furthermore, it is reasonable to expect that the moderating effect of role salience in linking working from home and WFC may also differ between the genders. According to Social Role Theory (Eagly & Wood, 2011), behavior that is consistent with gender norms is typically rewarded by others, whereas behavior that is inconsistent with gender norms tends to be penalized. Prioritizing a traditional role is thus easier to justify than prioritizing a non-traditional role (Erdogan et al., 2021). In this case, we may expect working from home to facilitate engagement in the traditional role, especially for those workers for whom the traditional role is the more salient role. Accordingly, men with high work-role salience may invest a particularly large share of their time and energy into paid work when working from home, and women with high family-role salience may invest particularly strongly in their family duties, thereby reducing the time and energy available for their respective non-traditional roles. We therefore assume that the association between working from home and WTFC is particularly strong for men with high work-role salience, whereas the association between working from home and FTWC is particularly strong for women with high family-role salience. This leads us to the following hypotheses:

H3a

The moderating effect of work-role salience on the association between working from home and WTFC will be stronger for men than for women.

H3b

The moderating effect of work-role salience on the association between working from home and FTWC will be stronger for women than for men.

3 Method

3.1 Data and Sample

We use data from Wave 12 (2019–2020) from the German Family Panel (Brüderl et al., 2021; pairfam Group, 2021).Footnote 1 This annual survey started in 2008–2009, when data was collected from a nationwide, random sample of 12,402 respondents. These respondents came from three age cohorts: 1971–1973, 1981–1983, and 1991–1993. Starting in Wave 11, a refreshment sample was drawn to increase the sample size. Specifically, the goal of this refreshment sample was to increase the sample size of the two younger age cohorts (1981–1983 and 1991–1993) by 1,500 respondents each, and to add a new birth cohort (2001–2003) with a sample of 3000 respondents. The oldest cohort (1971–1973) was not refreshed (Brüderl et al., 2021). By Wave 12, 7630 respondents were in the survey. For our analysis, we first dropped 72 individuals who were younger than 18 years old. Second, we dropped 2317 individuals who were not employed. Next, we excluded 377 self-employed individuals and 569 individuals who reported changing work locations. We performed listwise deletion where we only kept the sample with complete information on all variables, and we dropped the cases with missing data (N = 228).Footnote 2 These restrictions led to a final sample size of 4067 employees (N = 1806 men and 2261 women).Footnote 3

3.2 Measures

3.2.1 Dependent Variables

We use WTFC and FTWC as the outcome variables. We sum and average the following four items to measure WTFC: “(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.” Respondents who were missing information on any of these four items were excluded. Each item ranges from 1 = not at all to 5 = absolutely. A higher number on the scale indicates higher levels of WTFC. Cronbach’s alpha for the whole sample is 0.77 (0.75 for men and 0.79 for women).

We sum and average the following four items to measure FTWC: “(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.” Respondents who were missing information on any of these four items were excluded. Each item ranges from 1 = not at all to 5 = absolutely. A higher number on the scale indicates higher levels of FTWC. Cronbach’s alpha for the whole sample is 0.72 (0.74 for men and 0.70 for women). Both scales were originally developed (in English) by Carlson et al. (2000), then translated into German and validated by Wolff and Höge (2011).

3.2.2 Independent Variables

Our key independent variable is the frequency of working from home. This item is measured by combining the following two questions: (1) “Some people always work at the same location, while others change their working location, and yet others work directly from home. How is this for you, where do you work most of the time? (1 = Always working from home, 2 = Unchanging work location with the possibility of working from home (home-office), 3 = Unchanging work location without possibility of working from home)” and (2) “How often do you work from home? (1 = Daily, 2 = Several times per week, 3 = Once per week, 4 = 1–3 times per month, 5 = less often).”

Using a strategy from prior research (Schwarz et al., 2023; van der Lippe & Lippényi, 2020), we created a measure that reflects the proportion of total working hours that respondents spend working from home per month. Precisely, we assigned a proportion of zero to respondents who did not have the possibility of working from home, 0.02 to those who could work from home but reported working from home “less often than 1–3 times per month,” and 0.09 to those who reported working from home “1–3 times per month.” Those who reported working from home “once a week” were assigned 0.18, those who reported working from home “several times per week” were coded as 0.55, and those who either always work from home or who reported working from home “daily” were coded as 1. This transformation allowed us to treat the original ordinal variable as a continuous measure in our analyses.Footnote 4

3.2.3 Moderating Variables

Our analysis tests two moderating variables: gender (1 = male, 0 = female) and work-role salience. Work-role salience is measured by the following question: “How important are your job and your family in your daily life? Choose the response that best describes you and your daily priorities.” The answer categories range from 1 = My family is my highest priority to 5 = My job is my highest priority. We recoded this variable so that the answer categories range from 0 to 4, where a higher number on the scale indicates more work-role salience.

3.2.4 Control Variables

Following prior research (Laß & Wooden, 2023; van der Lippe & Lippényi, 2020; Yucel & Chung, 2023), we control for the following variables that may be related to both the frequency of working from home and to overall WFC: age, educational attainment, the number of children living in the household, the presence of a preschool child (age < 6) living in the household, relationship status and partners’ employment status, residence, household net income (logged), high-skilled occupations, working in the public sector, and work hours.

Given that pairfam is based on a sample of four cohorts, age is treated as a categorical variable. The three dummy variables indicate ages 18–29 (reference), ages 37–39, and ages 47–49. Education is measured as the highest level of schooling: “lower education” is equivalent to less than a high school degree, “intermediate education” is equivalent to a high school diploma or some college, and “upper education” is equivalent to having earned at least a college degree. We control for the total number of children living in the household (0, 1, 2, 3 or more), and we also have a dummy variable indicating whether there is a child younger than 6 years old living in the household.

We also created three categories combining the presence of a partner in the household and their employment status: having a partner in the household who is employed (reference), having a partner who is not employed, and having no partner in the household. Residence is measured by a dummy variable indicating whether the respondent works in East Germany. Respondents’ household monthly net income is measured as a continuous variable. This variable is skewed, so we take the natural logarithm before entering it into our models. High-skilled occupations are indicated by a dichotomous variable distinguishing those who report being in professional, managerial, or technical occupations from others. Public sector is a dummy variable indicating whether the respondent works in the public sector. Finally, work hours measures respondents’ usual hours worked per week.

3.3 Analytic Approach

Table 1 presents the descriptive findings for all variables used in our analysis, first for the whole sample and then for men and women separately. We use t-tests and chi-square tests to determine whether there are any significant gender differences in the continuous and categorical variables, respectively. The multivariate results are estimated by running a series of linear regression analyses. We use Stata 14.0 as a software to estimate these models.

Table 1 Descriptive statistics for all the variables
Full size table

Tables 2 and 3 present the main effects of the key predictors and interaction terms from the regression models predicting WTFC and FTWC, respectively (the full models with the controls are available in the Appendix as Tables 4 and 5). The analytical steps are as follows. First, we estimated the main effect of the frequency of working from home on each outcome (Model 1). Second, we added an interaction term between the frequency of working from home and role salience, to determine whether the main association between working from home and each outcome variable differs by the level of role salience (Model 2).

Table 2 Regression Analyses Predicting Work-to-Family Conflict: Unstandardized Coefficients and SEs
Full size table
Table 3 Regression Analyses Predicting Family-to-Work Conflict: Unstandardized Coefficients and SEs
Full size table

Next, we stratified the sample by gender and tested separately whether the main effect of the frequency of working from home on each outcome variable exists among men and women (Models 3 and 5). Then, we tested whether the interaction effect between the frequency of working from home and role salience exists among men and women separately (Models 4 and 6). Finally, we added a three-way interaction between working from home, work-role salience, and gender to the joint gender model, to see whether the effects differ significantly between men and women (Model 7). There was no evidence of or concern about multicollinearity among the variables used in the analyses (all variance inflation factors were less than 5). In all the analyses, unstandardized coefficients are reported.

We also conducted some sensitivity analyses to show the robustness of the main findings. Specifically, we ran analyses based on the subsamples from two groups with relatively high family demands: mothers and fathers. Tables 6 and 7 in the Appendix show the results from these separate models for fathers and mothers in predicting WTFC and FTWC, respectively. For both outcome variables, we first tested the main effect of the frequency of working from home on WTFC and FTWC separately (Model 1), and next we tested whether this association varied by work-role salience, by adding an interaction term between the frequency of working from home and work-role salience (Model 2). Furthermore, Tables 8 and 9 in the Appendix show the results from models predicting WTFC and FTWC using an alternative measure for role salience.Footnote 5

4 Results

4.1 Descriptive Findings

On average, respondents report low levels of WTFC and FTWC (2.31 and 1.63 on a scale from 1–5 respectively). There is no gender difference in the reports of WTFC or FTWC. Women report working from home more frequently than men (an average of 0.13 for women versus 0.11 for men, p < 0.01). In addition, men report higher work-role salience compared to women (an average of 1.12 for men versus 1.01 for women, p < 0.001). More men report having no children in the household (60% for men versus 47% for women, p < 0.001). More women report having an employed partner living in the household (64% for women versus 46% for men, p < 0.001). More women work in high-skilled occupations than men (58% for women versus 49% for men), but men work significantly more hours compared to women (an average of 39 h per week for men versus 32 h for women, p < 0.001). For all the other descriptive findings, please refer to Table 1.

4.2 Predicting Work-to-Family Conflict

As shown in Table 2, Model 1, working from home more frequently is associated with higher levels of WTFC in the joint sample (b = 0.23, SE = 0.05, p < 0.001). Thus, Hypothesis 1a is supported. More work-role salience is also associated with higher WTFC (b = 0.09, SE = 0.02, p < 0.001). Model 2 adds the interaction term between the frequency of working from home and work-role salience. The first thing to note is that the main effect of the frequency of working from home declines markedly and becomes statistically insignificant. This means that for individuals with a work-role salience value of zero (i.e., workers with high family-role salience) working from home more frequently is not associated with an increase in WTFC. In line with our expectations, the interaction term is positive (b = 0.11, SE = 0.06, p > 0.05), which indicates that among workers with higher work-role salience, the positive association between the frequency of working from home and WTFC becomes stronger. However, since the interaction term is not statistically significant, we ultimately cannot confirm Hypothesis 2a.

Next, we divide the sample by gender. Model 3 in Table 2 shows that working from home more frequently is not associated with higher levels of WTFC among men, and there is a positive but not significant moderating effect of work-role salience among men (see Model 4). Model 5 in Table 2 shows that working from home more frequently is associated with higher levels of WTFC among women (b = 0.30, SE = 0.07, p < 0.001). As shown in Model 6, there is also a positive but not significant moderating effect of work-role salience among women (b = 0.11, SE = 0.08, p > 0.05). Figure 1 presents the predicted values of WTFC based on the estimates from the interaction model. Figure 1 shows that the upward slope, indicating the positive effect of working from home more frequently on WTFC, is steepest for both men and women with high work-role salience.

Fig. 1
figure 1

The effect of the frequency of working from home on work-to-family conflict among men (upper part) and women (lower part). WFTC: Work-to-Family Conflict. Low Work-Role Salience is coded as those who respond “My family is my highest priority”. Moderate Work-Role Salience is coded as those who respond “My job and my family are equally important to me”. High Work-Role Salience is coded as those who respond “My job is my highest priority”

Full size image

We also formally test whether the moderating impact of work-role salience differs by gender, by means of a three-way interaction between the frequency of working from home, work-role salience, and gender (Model 7). The results show that the moderating effect of work-role salience on the association between working from home and WTFC does not vary between men and women (b = − 0.01, SE = 0.12, p > 0.05). Hypothesis 3a is therefore not supported.

4.3 Predicting Family-to-Work Conflict

Table 3 presents the results for the main effect of the frequency of working from home on FTWC and the moderating effect of work-role salience within the full sample, as well as among men and women separately. Model 1 in Table 3 shows that working from home more frequently is associated with higher levels of FTWC (b = 0.21, SE = 0.04, p < 0.001). Thus, Hypothesis 1b is supported. Work-role salience, however, is not associated with FTWC. Model 2 shows a negative and significant interaction term between more frequently working from home and work-role salience (b = − 0.09, SE = 0.04, p < 0.05). This suggests that the positive association between more frequently working from home and FTWC is weaker among those with higher work-role salience. Hypothesis 2b is thus supported.

Next, we divide the sample by gender. Model 3 in Table 3 shows that working from home more frequently is not associated with higher levels of FTWC among men. As shown in Model 4, the small and insignificant interaction term suggests that among men, work-role salience does not moderate the effect of the frequency of working from home on FTWC. Model 5 in Table 3 shows that working from home more frequently is, however, associated with higher levels of FTWC among women (b = 0.30, SE = 0.05, p < 0.001). Model 6 reveals that among women, the positive association between the frequency of working from home and FTWC is weaker among those with higher work-role salience (b = − 0.14, SE = 0.06, p < 0.01). As Fig. 2 shows, the upward slope, indicating the positive effect of the frequency of working from home on FTWC, is steepest for women with lower work-role salience. Moreover, this figure shows that the biggest differences in predicted FTWC levels between the three role salience groups are found among those women who always work from home. For men, however, FTWC levels are very similar regardless of their frequency of working from home or their work-role salience.

Fig. 2
figure 2

The effect of the frequency of working from home on family-to-work conflict among men (upper part)  and women (lower part). FTWC: Family-to-work conflict. Low Work-Role Salience is coded as those who respond “My family is my highest priority”. Moderate Work-Role Salience is coded as those who respond “My job and my family are equally important to me”. High Work-Role Salience is coded as those who respond “My job is my highest priority”

Full size image

We also formally test whether the moderating impact of work-role salience differs by gender, by means of a three-way interaction between the frequency of working from home, work-role salience, and gender (Model 7). The significant three-way interaction coefficient (b = 0.11, SE = 0.05, p < 0.01) confirms that the moderating effect is stronger for women: more precisely, it can only be found among women, since for men, the sum of the two-way interaction between working from home frequency and work-role salience (b = − 0.14) and of the three-way interaction (b = 0.11) is close to zero. Hypothesis 3b is therefore supported.

4.4 Sensitivity Analyses

We also ran sensitivity analyses where we tested the effects of the frequency of working from home and the moderating effect of work-role salience among two subsamples: mothers and fathers. Most of the findings from these tests (see Tables 6 and 7 in the Appendix) are consistent with our main findings, with one exception. Specifically, while working from home more frequently is not significantly associated with higher WTFC among men based on the whole sample, the same effect is significant among fathers (b = 0.34, SE = 0.12, p < 0.01). The magnitude of this effect is more than twice as large among fathers compared to all men in the whole sample. In terms of similarities, working from home more frequently is also associated with higher WTFC among mothers (b = 0.33, SE = 0.09, p < 0.001) as well as among the women in the whole sample, and despite the positive coefficient for the interaction term, the moderating effect of role salience on WTFC is also not significant for fathers or mothers. Moreover, the results from the supplementary analyses are consistent with the main findings, in that working from home more frequently is associated with higher FTWC among mothers only (b = 0.36, SE = 0.07, p < 0.001), which corresponds to a significant main effect among women only in the whole sample. Finally, despite the negative coefficient for the interaction term on FTWC among fathers, the moderating effect of role salience is only significant among mothers, corresponding to a significant interaction effect only among women in the whole sample.

In other supplemental analyses, we use an alternative measure for role salience. Building on identity theory, our preferred measure assumes a tradeoff between the salience of work and family roles in the form of a hierarchy of salience. Prior researchers have challenged this assumption, however, by arguing that individuals might experience low or high levels of role salience in work and family spheres simultaneously (Cinamon & Rich, 2002; Erdogan et al., 2021). We therefore created a categorical measure that distinguished among “dual high salience” (i.e., high salience for both family and work roles), “dual low salience” (i.e., low salience for both family and work roles), “predominant work-role salience” (i.e., low salience for the family role and high role salience for the work role), and “predominant family role salience” (i.e., high family role salience and low work role salience). Please see footnote # 5 for the construction of the scales and Tables 8 and 9 in the Appendix for the model results.

Consistent with our main findings, the alternative measure for role salience does not moderate the association between working from home more frequently and WTFC in the full model, for either gender. However, individuals with dual low salience and those with predominant family salience both report lower levels of WTFC and FTWC than those with dual high salience. Moreover, the positive association between working from home more frequently and FTWC is weaker among those with dual low salience and those with a predominant work salience, compared to those with dual high salience, both in the full sample and in the sample of women.

5 Discussion

Social scientists have long been concerned with the effects of working from home on the fit between work and family life, and this research field has even gained more momentum in the wake of the COVID-19 pandemic. Nevertheless, to the best of our knowledge, the literature has not yet considered whether the effect of working from home on WFC may depend on the salience workers assign to their work and family roles. Therefore, using data from Wave 12 of the German pairfam study, this paper not only analyzed the main effects of the frequency of working from home on WTFC and FTWC, but also tested the moderating role of work-role salience in linking these phenomena, and examined whether these moderating associations differ between men and women. Four main findings emerged.

First, in line with expectations, the frequency of working from home is positively associated with both WTFC and FTWC in the general sample. This finding corresponds with much of the previous research on Germany, which found increased WTFC and reduced work-life balance among home workers (Abendroth & Reimann, 2018; Bellmann & Hübler, 2020; Lott, 2017; Schwarz et al., 2023). It may suggest that the drawbacks of working from home, such as boundaryless working hours or rising family demands, more than offset the possible advantages such as reduced commuting time and increased control over one’s working schedule. Alternatively (or additionally), this result may suggest that German employees working from home pre-COVID were a select group with particularly large work and/or family demands. This explanation is supported by previous studies showing that home workers were likely to work in high-demand workplaces, work frequent overtime, have younger children, and spend more time on childcare than those working exclusively onsite (Abendroth & Reimann, 2018; Lott, 2019). While we partly account for such work and family demands in our analyses by controlling for key socio-demographic and job characteristics, home workers might have additional (unobserved) characteristics that further drive WFC up, such as specific reasons for working from home. For instance, those working from home involuntarily due to physical constraints might experience higher levels of WTFC. Furthermore, working from home pre-COVID likely was often the result of high time pressure at work, leading workers to bring home work that they did not manage to finish during the regular work day.

Second, separate analyses by gender showed that the positive associations between working from home and WTFC/FTWC are only significant among women; for men, the direction of the effects was similar, but the estimates were smaller and did not reach statistical significance. Again, this result aligns with previous research finding these effects to be restricted to or stronger for women (Schwartz et al., 2023; van der Lippe & Lippényi, 2020; Yucel & Chung, 2023), even though it contradicts other studies that did not find any significant gender difference (Lott, 2017). This finding may be attributed to the fact that women are usually the primary caregivers in their households, which may lead to stronger competition for time and energy between work and family for women, compared to men, when working from home. When focusing only on parents, however, we did find working from home to be associated with increased WTFC for both mothers and fathers. Again, this finding may be explained by the relatively high family commitments of these sub-groups.

Third, we found that the salience people assign to their work and family roles affects the degree of conflict between the two spheres. Precisely, placing higher priority on one’s work role (as opposed to one’s family role) is associated with an increase in WTFC in the general sample, as well as for both genders. This result reflects previous findings (Beutell & Wittig-Berman, 1999; Chrisangika Perera & Kailasapathy, 2020; Frone & Rice, 1987; Greenhaus & Kopelman, 1981; Wiley, 1987) and is in line with the notion that workers who place more importance on their job will invest more time and/or energy in that job, and are thereby more likely to see job demands encroach on their family life.

By contrast, and contrary to our expectations, the level of work-role salience is generally not significantly associated with FTWC (with the exception of the subgroup of fathers, where we see a positive association). Workers with higher family-role salience thus appear no more likely than others to have their family life demands encroach on their work life. This finding, which mirrors previous research (e.g., Beutell & Wittig-Berman, 1999; Chang et al., 2014), might be explained by the fact that around three-fourths of the sample reported either that family is their highest priority or that they tend to prioritize family more than their jobs. Therefore, it might be that most people in this sample perceive their work role as the “extra role” (Noor, 2004, p. 392), which would make it acceptable for family demands to interfere with work responsibilities. Consequently, these workers would not experience any role conflict when family responsibilities interfere with their work (i.e., FTWC) (Noor, 2004). Moreover, social desirability may also play a role, in that individuals with high family-role salience might not want to admit experiencing family responsibilities interfering with their work demands.

Fourth, and most importantly, our analysis provides insights into the potential moderating role of work-role salience on the link between working from home and conflicts between work and family. Contrary to our expectations, we did not find any moderating effect of role salience on the link between working from home and WTFC. The results, however, showed that work-role salience moderates the effect of working from home on FTWC. In addition, separate analyses by gender revealed work-role salience to negatively moderate the link between working from home and FTWC among women. Precisely, working from home more frequently is associated with the steepest increase in FTWC among those who prioritize their family role over their work role. By contrast, among those who prioritize their work over their family, an increase in the frequency of working from home is not at all associated with an increase in FTWC. As a result, the predicted values showed that among women who work from home never or rarely, the level of FTWC is independent of the level of work-role salience, whereas among those working from home on a daily basis, FTWC levels strongly depend on the level of work-role salience. This outcome is intuitive considering that family commitments have less opportunity to affect the work role when workers are at their workplace, regardless of role salience. By contrast, when at home, women with high family-role salience appear to have greater difficulties blocking out family demands.

Among men, FTWC levels are very similar regardless of their frequency of working from home or their work-role salience levels. This finding can potentially be attributed to the fact that men use working from home much less to accommodate family demands and more so to accommodate work demands and work overtime hours (Lott, 2019). With respect to WTFC, the interaction effect between the frequency of working from home and work-role salience was, although sizeable and positive, not statistically significant. Nevertheless, the predicted values, as shown in our figures, suggest that the positive association between the frequency of working from home and WTFC may be restricted to workers with high work-role salience. This would be in line with our assumption that these workers are more likely to shift time and energy to their work role when working from home and thereby create higher WTFC. However, since the effect is not statistically significant and thus lacks precision, future studies will be required to uncover whether this effect holds beyond our analysis sample.

Our study also has some limitations. First, the cross-sectional nature of our data implies that we cannot rule out the possibility that unobserved factors, correlated with both working from home and WTFC/FTWC, could bias our results. Second, given that we use wave 12 of a panel study, there is the risk of bias from selective attrition. Third, even though our overall sample was sizeable with several thousand cases, certain characteristics (such as very high work-role salience) are relatively rare, potentially causing some of our effects to be statistically insignificant. Finally, our key independent variable in this study (i.e., the frequency of working from home) was only asked of the main respondent, but not their partners. This prevented us from using couple-level analysis to test the possible spillover and crossover effects of couples' flexible work arrangements on agents’ and partners' reports of work–family conflict. Given the theoretical argument that stress and strain are transmitted across partners due to empathy among them (Westman, 2001, 2006), future studies could benefit from replicating these analyses using longitudinal and couple-level data.

Overall, our findings suggest that the link between working from home and the fit between work and family is heterogeneous, varying not only by gender but also partly by the importance workers assign to their different life roles. Whereas some prior research tested how role salience moderates the effect of work and family demands on the work-family interface (Chang et al., 2014; Day & Chamberlain, 2006; Fox & Dwyer, 1999), to our knowledge, our study is the first to consider role salience in relation to working from home and with a large, nationally representative dataset. It will be an interesting question for future research whether our findings will also hold post-COVID, in a context where working from home has become accessible for a much larger share of employees. Furthermore, our study highlights the need for work and family researchers to consider the influence of both role salience and gender (as well as parenthood) in their studies. Finally, given the fact that cultural background impacts role salience hierarchies (Greer & Egan, 2012), future research could benefit from comparative analyses using cross-national data.

Given our finding that working from home can lead to increased WFC, combined with the fact that a much larger segment of the workforce is expected to work from home post-COVID, workplaces would be well-advised to implement HR policies that facilitate a better balance between work and family responsibilities for employees working from home. Such policies might include agreed-upon work hours, after which workers are not expected to respond to any work requests or complete any work tasks. Furthermore, employees should be provided with the flexibility to take breaks at their discretion to deal with personal and family demands. Moreover, flexible and high-quality childcare options should be made available to all workers, including those who work from home. Overall, the beneficial potential of working from home for employees can only be realized within a more supportive work-life culture in and outside of the workplace.