The existence of dual-earning families (Bellavia and Frone 2005) and the growing presence of men as caregivers and women as breadwinners (Bianchi and Milkie 2010; Council of Economic Advisers 2014; Perrone et al. 2009) suggest that men and women both experience work–life conflict (Rehel and Baxter 2015). Work–life conflict, “a particular type of inter-role conflict in which pressures from the work role are incompatible with the pressures from the [life outside of work] role” (Thomas and Ganster 1995, p. 7) is associated with higher turnover intentions and higher job dissatisfaction (Allen 2001; Thompson et al. 1999). This may be particularly likely for employees who perceive negative career consequences associated with managing family responsibilities and for organizations that lack family-specific workplace support (Anderson et al. 2002).

Family responsibility discrimination reports have increased by 400% over the past decade (Calvert 2010). This presents both practical and ethical issues for organizations, because discrimination is not only legally but also morally wrong (Demuijnck 2009). Discrimination is also associated with increased turnover intentions (Anderson et al. 2002; Dickson 2008; Thompson et al. 1999) which can lead to negative performance effects. Perceived family responsibility discrimination (FRD) occurs when individuals in the workplace perceive that they were denied equality of treatment in the workplace on the basis of their responsibility for the care of their children, elderly parents, or disabled children or relatives (Allport 1954; Scott 2007). As a consequence, they “are put at a disadvantage in the workplace relative to other groups with comparable potential or proven success” (Dipboye and Halverson 2004, p. 131). Research finds a positive relationship between FRD and work–life conflict and turnover intentions (Anderson et al. 2002; Dickson 2008; Thompson et al. 1999).

To better understand why FRD leads to higher levels of work–life conflict, we draw on conservation of resources theory (Hobfoll 1989) which states that individuals strive to protect and enhance themselves through the acquisition and maintenance of resources, and we examine the dependency of this effect on gender and power distance. Conservation of resources theory defines resources broadly as “those objects, personal characteristics, conditions, or energies that are valued in their own right, or that are valued because they act as conduits to the achievement or protection of valued resources” (Hobfoll 2001, p. 339). The theory suggests that environmental conditions can deplete or threaten resources. For example, a poor social relationship (such as with one’s supervisor) can lead to a loss of resources (Hobfoll 1989). Therefore, resources can include being treated with fairness, dignity, and respect. This would make FRD a resource loss because the employee feels they are treated unfairly.

When applied to the work–life literature, conservation of resources theory explains that demands in one domain of life can deplete energies and resources from another domain (Hobfoll 1989). We connect FRD to two important outcomes, emotional exhaustion and work–life conflict. We further build on this by theorizing how the relationship between FRD and these outcomes may differ based on two key moderators: gender and power distance. Women have traditionally fulfilled (and still fulfill) a disproportionately large share of childcare and housework compared to men (Hochschild and Machung 2012), which would facilitate feelings of exhaustion and work–life conflict. Power distance, the degree to which inequalities in power are accepted either as unavoidable or as functional (Clugston et al. 2000; Hofstede 1980), and gender roles complement each other in describing the hierarchical culture of a society and capture underlying inequality between various traditionally hierarchical groups (e.g., superior vs. subordinate, and man vs. woman). This is particularly relevant when considering the role of FRD on emotional exhaustion and work–life conflict, given that inequality and hierarchy between these social groups has traditionally been accepted due to prevalent gender roles of females as homemakers and males as breadwinners (Ridgeway 2009, 2011).

The present study advances theory and research on work–life conflict in the following ways. First, by drawing on conservation of resources theory (Grandey and Cropanzano 1999; Hobfoll 1989), we examine why family responsibility discrimination from a supervisor leads to emotional exhaustion and work–life conflict. By taking a resource loss approach, we add a unique perspective to a stream of literature that has focused more on resource gains associated with family-supportive supervisors (e.g., Bagger and Li 2014; Carlson and Perrewé 1999; Hammer et al. 2009). Additionally, we examine a specific form of FRD (from a supervisor) rather than a general form of discrimination. This adds to research that finds specific work–family support from supervisors is more strongly associated with work-to-family conflict than general supervisor support (Kossek et al. 2011).

Second, our model also answers the call for more research into conditions on which the relationships between antecedents and consequences of work–life conflict vary (Kossek and Lee 2017) by identifying emotional exhaustion as an important mediator and gender and power distance as moderators of the relationship between FRD and work–life conflict. By examining gender and power distance as moderators, our work expands the scope of work–life studies to consider the role of relevant individual differences (Allen 2012; Carlson and Kacmar 2000) and suggests that employees are likely to differ with regard to when their experience of FRD turns into emotional exhaustion and work–life conflict.

Finally, we examine emotional exhaustion as a mediator between FRD and work–life conflict. We add value by exploring why FRD leads to work–life conflict since most models in this research area presume “work-family conflict influences strain, but they do not acknowledge potential influences of strain on work–family conflict” (Nohe et al. 2015, p. 8). The present study also provides insight on one of conservation of resources theory’s mechanisms (resource loss and the experience of emotional exhaustion, Hobfoll 2001) by examining emotional exhaustion as a mediator following a resource loss (FRD), which other researchers have said should be tested in stress models (Halbesleben and Buckley 2004). Practically, organizations should understand individual difference factors that can exacerbate the outcomes of unsupportive supervisors so that they can attempt to lessen emotional exhaustion and work–life conflict.

Theory and Hypotheses

As shown in Fig. 1, we model FRD from one’s supervisor to emotional exhaustion and, ultimately, work–life conflict using conservation of resources theory (Hobfoll 1989, 2001). Work–life conflict is an extension of work–family conflict reflecting the reality that the work role may interfere with individuals’ other personal life roles and interests. Following Siegel, Post, Brockner, Fishman, and Garden (2005), we use the term work–life conflict to reflect the many additional non-work demands in individuals’ lives that are not restricted to those involving the family. We suggest that gender and power distance are essential in understanding when employees feel emotional exhaustion and work–life conflict, as these factors affect what is considered acceptable behavior for supervisors and subordinates (Erdogan and Liden 2002; Lee et al. 2000). Our model suggests that feeling the draining effects of FRD is higher for (a) female employees (regardless of power distance) given gender role expectations of being more devoted caregivers (Hochschild and Machung 2012) and for (b) male employees with high power distance because they place more value on what their supervisor thinks.

Fig. 1
figure 1

Theoretical model

Full size image

Family Responsibility Discrimination, Emotional Exhaustion, and Work–Life Conflict

Research reveals that an unsupportive work–family culture is positively related to work–life conflict (Allen 2001; Behson 2002; Thompson et al. 1999; Voydanoff 2004). Conservation of resources theory suggests that emotional exhaustion, or “feelings of being overextended and depleted of one’s emotional and physical resources” (Maslach and Leiter 2008, p. 498), is likely to occur under resource loss conditions given an individual’s increased sensitivity to resource loss (Hobfoll 2001). An increased sensitivity to resource loss is likely to lead to individual efforts that overcompensate in the amount of resources expended to prevent further loss (Hobfoll and Freedy 1993). For example, an employee who perceives FRD from their supervisor is more sensitive to the loss of that resource, including the loss of fulfilling the social need for relationships or the loss of being able to protect other valued resources such as family time and responsibilities. This increased sensitivity amounts to individuals expending emotional energy (Hobfoll 1989, 2001) on the job in an attempt to prevent further resource loss. However, repeated exposure to the resource loss is likely to lead to feeling emotionally overextended and drained. Therefore, the following hypothesis is presented:

Hypothesis 1a

Perceived family responsibility discrimination will be positively related to emotional exhaustion.

Given limited resources, energy spent at work trying to prevent further resource loss comes at the expense of energy devoted to the non-work role. Experiencing FRD at work may take away from resources that one would spend on the life role, which may include family (e.g., not being able to pick up kids from school) as well as other roles outside of work (e.g., having less time for exercise or for friends). This is because FRD experienced at work depletes resources that could otherwise be spent on non-work activities, whether they include family or not. In line with conservation of resources theory (Hobfoll 2001), previous research has found strong evidence for a lack of workplace social support, emotional exhaustion, and role stress (e.g., Schaufeli and Enzmann 1998) resulting in work–life conflict. Therefore, based on past research, and following conservation of resources theory (Hobfoll 1989), we predict that when an employee encounters high levels of perceived FRD this is likely to lead to a depletion process in emotional energy and increased work–life conflict. Thus, we present the following hypothesis.

Hypothesis 1b

Perceived family responsibility discrimination will be positively related to work–life conflict.

As shown in Fig. 1, our full model hypothesizes the relationship between perceived FRD and work–life conflict through emotional exhaustion. We present the full moderated-mediation model in Hypothesis 4. Before presenting our full moderated-mediation model, we theorize about the moderating roles of gender and power distance on the relationship between perceived FRD and both emotional exhaustion and work–life conflict.

The Moderating Role of Gender

We turn to social role theory (Eagly and Wood 1991) to gain insights into whether societal pressures associated with gender exacerbate the relationship between perceived FRD and emotional exhaustion as well as work–life conflict. Social roles based on gender cue others’ expectations and attributions about men and women (Eby et al. 1999). Social role theory suggests gender-appropriate behaviors are learned and reinforced through society’s power and status structures (Eagly 2013; Eagly and Wood 1991). People internalize defined gender roles and have a higher tendency to behave in ways that conform to the societal expectations associated with these roles (Maccoby 1990). Research suggests the differences between men and women in terms of social norms guide their work and family role behaviors (Duxbury and Higgins 1991; Gutek et al. 1991).

Earlier research suggests that gender differences may exist given the emphasis in female socialization on interdependence between work and family roles as well as the ability to relate meaningfully to others in interpersonal relationships (Greenglass et al. 1988). Research on gender differences in work values supports this notion and finds significant differences in work values consistent with stereotypical male and female value patterns, such as advancement and risk-taking for men and comfortable work environments and congeniality for women (Beutell and Brenner 1986). This also aligns with the stress literature indicating that members of disadvantaged social groups (e.g., women) are especially vulnerable or emotionally reactive to stressors (Thoits 1995). Specifically, women seem to be more vulnerable to events that happen to loved ones in their social networks, while men may be more vulnerable to financial and job-related stressors (Gore and Colten 1991).

Given these value differences, conservation of resources theory (Hobfoll 2001) would suggest that individuals are more likely to protect and maintain the resources they value most. Based on social role theory, women are likely to value, protect, preserve, and maintain relationships (Eagly and Wood 1991; Hochschild and Machung 2012). This includes relationships in both work and non-work domains and, as such, women are suggested to experience an increased sensitivity to perceiving a supervisor that discriminates based on family responsibilities. This increased sensitivity may be attributed to increased efforts to protect the resources they value most, including the need to maintain relationships and at the same time protect their family responsibilities. Therefore, women who experience FRD are suggested to experience work stress, partly from stereotypes regarding appropriate feminine gender-role behaviors (Greenglass et al. 1988). Research also shows that married women with children do a disproportionately large amount of childcare and housework compared to their husbands (Hochschild and Machung 2012). Data from the year 2000 show that married fathers spent almost twice as many hours per week working in the labor market (42.5 h) compared to married women (23.8 h), while married women spent almost twice as many hours on non-market work including childcare and housework (41.1 h) as married men (21.5 h) (Bianchi et al. 2006). Investing more time in housework and childcare, on average, combined with having a supervisor who is unsupportive and exhibits FRD toward employees should cause a particularly strenuous reaction with respect to emotional exhaustion and work–life conflict from those who have the most obligations at home: women. Therefore, the following hypotheses are posited:

Hypothesis 2a

The positive relationship between family responsibility discrimination and emotional exhaustion will be stronger for women than for men.

Hypothesis 2b

The positive relationship between family responsibility discrimination and work–life conflict will be stronger for women than for men.

The Moderating Impact of Power Distance

Next, we turn to a discussion of power distance values as an explanatory variable which can further clarify the interaction effects presented in Hypotheses 2a and 2b. Power distance represents an individual value characterized by acceptance of hierarchy and status differences (Hofstede 2001). Power distance influences employees’ relationships with their supervisors (Clugston et al. 2000; Erdogan and Liden 2002) as well as their perceptions, attitudes, and reactions to organizational injustice (Farh et al. 2007; Lam et al. 2002; Lian et al. 2012; Liu et al. 2013). People who believe that superiors should have a great deal of power over subordinates are high on power distance while people who believe that a smaller degree of power is appropriate are low on power distance. Thus, people with high power distance find unequally distributed power more acceptable than do those with low power distance (Clugston et al. 2000). Research supports this point, finding that individuals with high power distance are more likely to accept and be satisfied with unfair supervisors (Taras et al. 2010). They are also less likely to react adversely to injustice (Lee et al. 2000) and will be more tolerant of supervisor criticism and insults (Bond et al. 1985; Leung 2001). High power distance employees tend to believe that bypassing their bosses is insubordination and, as such, tend to behave submissively around managers, show respect and reverence to supervisors, avoid disagreements, and are less likely to be involved in direct arguments with their superiors (Liu et al. 2013). Individuals with low power distance are more likely to react negatively when authorities treat them unfairly because they view such treatment as violating relational bonds between them and their supervisors (Tyler et al. 2000).

Power distance provides insight into status differences that may arise between employees and their supervisors, particularly for male employees. For women, we propose that regardless of their power distance values, the experience of FRD results in added emotional energy expended at work and, in turn, emotional exhaustion and work–life conflict. Men, however, are more likely to value status rather than relationships as a resource (Eagly and Wood 1991; Halrynjo 2009). Specifically, male employees high on power distance are more sensitive to status signals (Ng and Van Dyne 2001). Employees high in power distance are bound by role-constrained interactions and are more likely to comply with role prescriptions (Leung 2001) to conform to a supervisor’s expectations. Moreover, an increased sensitivity to pleasing an unsupportive supervisor can result in additional emotional energy expended at their job. Instead of conserving resources, male employees who score high on power distance values may overcompensate and spend more energy at work in an attempt to please their supervisor (House et al. 2004; Tyler et al. 2000). This is likely to lead to more emotional exhaustion and higher levels of work–life conflict.

However, male employees who score low on power distance should respond the least to FRD. Hobfoll (1989) explains that the burden of a poor social relationship can be made lighter by reducing the importance one places on that relationship. Men who are low on power distance are less concerned about their relationship with their supervisor and are less likely to overcompensate. As such, male employees with low power distance are predicted to expend the least amount of emotional energy in response to FRD because they do not have self-imposed pressures to conform to their supervisors (i.e., compared to those high in power distance) and are less bound by social pressures to conform to domestic or caregiving roles (i.e., compared to females). Therefore, we predict the following three-way interaction between FRD, gender, and power distance:

Hypothesis 3

The effect of perceived family responsibility discrimination on emotional exhaustion (H3a) as well as work–life conflict (H3b) will be stronger for women, regardless of power distance, and for men who score high on power distance compared to men who score low on power distance.

The Moderated Mediation Model

Taken together, we propose that FRD leads directly to work–life conflict and also indirectly to work–life conflict via emotional exhaustion. Further, we hypothesize a three-way moderation effect on both the direct effect as well as the first-stage of the indirect effect. The direct effect of FRD on work–life conflict as well as the three-way moderation mechanisms follow the same logic described above. Conservation of resources theory explains why we expect an indirect effect through emotional exhaustion. Employees who experience FRD have less control of work–life demands and perceive they have fewer resources to manage work–life conflict. The draining of resources caused by dealing with an unsupportive supervisor creates emotional exhaustion. In turn, a higher level of emotional exhaustion is likely to impede the non-work domain, interfere with non-work responsibilities (Greenhaus and Beutell 1985), and ultimately result in increased work–life conflict (Dickson 2008).

Consistent with our earlier argument regarding the three-way interaction, the effects will be the strongest for women, who experience higher demands with respect to housework and childcare compared to men (Bianchi et al. 2006; Hochschild and Machung 2012), as well as for men who are high on power distance values and emphasize pleasing their supervisors. Therefore, we present the following comprehensive moderated mediation hypothesis.

Hypothesis 4

Perceived family responsibility discrimination relates indirectly to work–life conflict through emotional exhaustion. The direct effect as well as the indirect effect of family responsibility discrimination on work–life conflict will be moderated by both gender and power distance such that women (regardless of power distance) and men who are high in power distance experience more emotional exhaustion, and subsequently, work–life conflict compared to men who are low in power distance.

Method

Procedures and Participants

We collected our data through StudyResponse, a panel service based in the U.S. To be considered for the sample participants had to be employed adults, United States residents, and willing to participate in two surveys. This study was conducted as part of a larger data collection. Aside from demographic variables, there is no overlap between this study and others from the same data collection. To gather a diverse sample, a participation request was sent to all of the African–American and Hispanic individuals that met the criteria as well as to a random selection of Caucasian individuals that met the criteria until a total number of 350 were invited.

In Time 1, employees were asked questions about family responsibility discrimination, emotional exhaustion, power distance orientation, demographics, and other work characteristics unrelated to this study. All participants received a $13 Amazon.com gift card in exchange for their participation in Time 1. The response rate for employees who completed Time 1 was 95% or 334 employees. Employees who completed the second phase at Time 2 (which took place four months after Time 1) answered questions about their work–life conflict and were given a $10 Amazon.com gift certificate. Collecting data across multiple points in time helps establish the temporal order of the variables in the study, thereby reducing common method bias concerns (Podsakoff et al. 2003). The final response rate was 74%, or 258 employees that completed the first phase and completed the second phase. Of those 258, thirty-four had missing values on one or more of the variables of interest in our study, taking our final sample to 224. The employee sample was on average 50% male, 39 years old, 67% Caucasian, and worked for about 8 years. Checks for non-response bias were performed and we found no difference between those who did respond and those who did not respond in terms of demographic characteristics such as: sex [χ2(1) = 1.58, p > 0.05], racial majority/minority status [χ2(1) = 0.27, p > 0.05], or age, [t(348) = − 1.60, p > 0.05].

Employees were also asked to provide contact information for a significant other, defined as someone (18 years or older) who knows the employee the best or really well and, as such, would be in a good position to assess the employee’s work and personal life (i.e., a spouse, friend, relative, coworker) (Boswell and Olson-Buchanan 2007). These surveys were sent electronically using the contact information provided by the employee. We matched all of the significant other surveys to our final sample of 224 employee surveys. Data were collected from significant others as a way to address the common method bias that arises from a single source respondent (Podsakoff et al. 2003). Given previous research that has used both self-reports and significant other reports of work–family conflict (Breaugh and Frye 2008), we chose to corroborate the relationship between family responsibility discrimination as reported by the employee with reports provided by significant others of employee work–life conflict. We present these findings in the supplemental analysis section.

Measures—Time 1 Survey

Family Responsibility Discrimination

The research team adapted a two-item measure from published research that examines perceived age discrimination from a supervisor (Triana et al. 2017). We modified the two items to reflect family responsibility discrimination from a supervisor by replacing the word “age” with the words “family responsibilities.” Participants were asked to answer these questions about the direct supervisor they work with the most to get their job done. The items were: “To what extent do you believe that your supervisor has discriminated against you on the basis of your family responsibilities?” and “To what extent do you believe that your supervisor has treated you differently from others based on your family responsibilities?” Employees answered the questions on a Likert scale from 1 = to a small extent to 5 = to a large extent. Spearman–Brown coefficient of reliability was equal to 0.93 and the correlation between item 1 and item 2 was 0.87.

Emotional Exhaustion

We used three items of the Maslach Burnout Inventory (Maslach et al. 1986) to represent emotional exhaustion. A sample item was: “I feel emotionally drained from my work.” Responses were scored on a 7-point frequency scale, anchored at 1 = never and 7 = very often (Cronbach alpha α = 0.94).

Gender

One item on the survey asked participants their gender. This variable was coded 0 = female and 1 = male.

Power distance

We used Maznevski and DiStefano’s (1995) scale which was used by Kirkman and Shapiro (2001). Sample items include, “A hierarchy of authority is the best form of organization” and “People at lower levels in the organization should not have much power in organizations.” Responses were scored on a 7-point agreement scale, anchored at 1 = strongly disagree and 7 = strongly agree (α = 0.82).

Controls

We controlled for variables that have been shown to relate to work–life issues (Bagger and Li 2014). We controlled for whether the employee had any children living at home (0 = no children living at home, 1 = children living at home), the employee’s relationship status (0 = single, 1 = spouse/partner), full-time employee status (0 = part-time, 1 = full-time), professional occupation (0 = non-professional, 1 = professional), employee age (in years), and organizational tenure (in years). We coded type of occupation to reflect either professional or non-professional occupations. Occupations were coded as professional if the occupation required an education and/or work took place in an office. We coded professional jobs because they typically provide more benefits (e.g., vacation time, schedule flexibility) that can help employees deal with work–life conflict.

Measures—Time 2 Survey

Work–Life Conflict

Following Boswell and Olson-Buchanan (2007) we assessed work-to-life conflict using the four-item Work Interference With Family Scale (Gutek et al. 1991) since the original items reflect an individual’s personal life more generally. This scale was used to assess the extent to which employees agreed or disagreed with four statements about how much work interferes with life. Employees reported the extent of agreement on a Likert-type scale from 1 = strongly disagree to 7 = strongly agree. Sample items include: “After work, I come home too tired to do some of the things I’d like to do” and “My work takes up time that I’d like to spend with family/friends.” A higher score reflects more conflict. We assessed employee work–life conflict as reported by the employee at Time 2, or four months following the Time 1 survey (α = 0.84). As part of our supplemental analysis, we also assessed employee work–life conflict as reported by the employee’s significant other. The items on the significant other survey were worded to elicit the individual’s perspective regarding the employee’s work–life conflict (α = 0.91). The correlation between the employee’s report of work–life conflict and the significant other’s report of work–life conflict was 0.60.

Results

To provide evidence concerning the discriminant validity of our self-reported measures, we conducted a confirmatory factor analysis (CFA) in LISREL testing a series of nested models. A four-factor solution (FRD, emotional exhaustion, power distance values, and work–life conflict) was an acceptable fit for the data (χ2 = 389.38, p < 0.01, df = 98, CFI = 0.92, IFI = 0.93, SRMR = 0.08; Kline 2005). The hypothesized four-factor model provided a significantly better fit to the data than: a three-factor model with work–life conflict and emotional exhaustion as factor 1, family responsibility discrimination as factor 2, and power distance as factor 3 (χ2 = 633.34, p < 0.01, df = 101, CFI = 0.86, IFI = 0.86, SRMR = 0.12; ∆ χ2 = 243.96, df = 3, p < 0.00), a two-factor model with work–life conflict and emotional exhaustion as factor 1 and family responsibility discrimination and power distance as factor 2 (χ2 = 936.27, p < 0.01, df = 103, CFI = 0.78, IFI = 0.78, SRMR = 0.14; ∆ χ2 = 546.88, df = 5, p < 0.00), and a one-factor model (χ2 = 1382.06, p < 0.01, df = 104, CFI = 0.67, IFI = 0.67, SRMR = 0.15; ∆ χ2 = 992.68, df = 6, p < 0.00). In sum, results of the CFA support our proposed four-factor model in which work–life conflict, emotional exhaustion, family responsibility discrimination, and power distance are four distinct variables.

Means, standard deviations, and correlations for study variables are reported in Table 1. Given that significant intercorrelation was evident among some variables, multicollinearity was further investigated using variance inflation factors (VIFs). The VIFs for independent variables were all below 10 (2.886 for family responsibility discrimination, 2.962 for gender, 2.453 for power distance, and 7.613 for emotional exhaustion). Therefore, there was no evidence of substantial multicollinearity (Cohen et al. 2003).

Table 1 Means, standard deviations, and correlations
Full size table

We used regression analysis to test the main effects of family responsibility discrimination on emotional exhaustion (Hypothesis 1a) and work–life conflict (Hypothesis 1b). As can be seen in Table 2, higher levels of FRD were associated with higher levels of emotional exhaustion (b = 0.69, p < 0.01), supporting Hypothesis 1a. Hypothesis 1b was also supported; higher levels of FRD were associated with higher levels of work–life conflict (b = 0.44, p < 0.01).

Table 2 Regression analysis: family responsibility discrimination on emotional exhaustion and work–life conflict
Full size table

We used the Hayes (2013) PROCESS macro to test our moderation and mediation hypotheses. PROCESS uses bootstrapping, which provides greater statistical power and smaller Type 1 and Type 2 errors than the traditional Sobel test for complex models tested with low- to moderate-sized samples (Preacher and Hayes 2008). Hypotheses 2a and 2b examined the moderating effect of gender on the relationships between FRD and emotional exhaustion and work–life conflict, respectively. As shown by Model 1 in Table 3, the moderating effect of gender on emotional exhaustion was not significant (b = 0.05, NS), failing to support Hypothesis 2a. Results for Hypothesis 2b can be found in Model 3 of Table 3. The moderating effect of gender on work–life conflict was not significant (b = − 0.12, NS). We next tested the three-way interaction effect of gender, power distance, and FRD on emotional exhaustion (Hypothesis 3a) and work–life conflict (Hypothesis 3b). The coefficients on the three-way interaction shown in Model 2 of Table 3 and Model 4 of Table 3 are significant for both emotional exhaustion (b = 0.42, p < 0.05) and work–life conflict (b = 0.40, p < 0.01), providing support for Hypotheses 3a and 3b.

Table 3 Moderated analysis conducted using PROCESS macro
Full size table

To facilitate interpretation, we plotted the three-way interaction at one standard deviation above and below the mean for power distance and at zero and one for gender (Aiken and West 1991) for both emotional exhaustion (see Fig. 2) and work–life conflict (see Fig. 3). We mean centered FRD and power distance prior to constructing interaction variables to avoid multicollinearity, following Cohen et al. (2003).

Fig. 2
figure 2

Interaction of family responsibility discrimination, power distance, and gender predicting emotional exhaustion

Full size image
Fig. 3
figure 3

Interaction of family responsibility discrimination, power distance, and gender predicting work–life conflict

Full size image

We tested the differences of the simple slopes from zero. For emotional exhaustion as our dependent variable, we found the female low power distance slope was positive and significant (b = 0.62, p < 0.01), the female high power distance slope was positive and significant (b = 0.70, p < 0.01), the male low power distance slope was not significantly different from zero (b = − 0.09, NS), and the male high power distance slope was positive and significant (b = 0.92, p < 0.01). For work–life conflict as our dependent variable, we found the female low power distance slope was positive and significant (b = 0.59, p < 0.01), the female high power distance slope was positive and significant (b = 0.48, p < 0.01), the male low power distance slope was not significantly different from zero (b = − 0.30, NS), and the male high power distance slope was positive and significant (b = 0.47, p < 0.01). As predicted, the slopes of the lines were positive and significantly different from zero for all women, regardless of power distance, and for men who scored high on power distance. These results provide support for Hypotheses 3a and 3b.

Hypothesis 4

tests the full model in which emotional exhaustion acts as a mediator between FRD and work–life conflict. We hypothesized a first-stage three-way interaction on emotional exhaustion as well as a three-way interaction on the direct effect of FRD to work–life conflict. We used Model 12 of the PROCESS macro to test these effects (Hayes 2013). These results can be viewed in Table 4. Taken together, the first-stage three-way interaction on emotional exhaustion is significant (b = 0.42, p < 0.05) and the three-way interaction on the FRD to work–life conflict direct effect is also significant (b = 0.26, p < 0.05). Moreover, emotional exhaustion mediates the effect (b = 0.33, p < 0.001). We calculated the strength of the indirect effect at one standard deviation below and above the mean of power distance and at zero and one of gender, using the procedures suggested by Preacher, Rucker, and Hayes (2007). For Hypothesis 4, the bootstrapped indirect effects were significant for females with low power distance (b = 0.21, [95% confidence interval (CI) 0.10, 0.34]), females with high power distance (b = 0.23, [0.12, 0.39]), and males with high power distance (b = 0.30, [95% CI 0.18, 0.46]). The bootstrapped indirect effect was not significant for males with low power distance (b = -0.03, [95% CI − 0.21, 0.14]). The conditional direct effects were significant for females with low power distance (b = 0.38, [95% CI 0.15, 0.62]) and marginally significant for females with high power distance (b = 0.24, [95% CI 0.00, 0.49]). The conditional direct effects were not significant for males with high power distance (b = 0.16, [95% CI − 0.07, 0.39]) or for males with low power distance (b = − 0.27, [95% CI − 0.64, 0.10]). These results provide support for the three-way moderated mediation of the indirect effect of Hypothesis 4 and partial support for the three-way moderated mediation of the direct effect.

Table 4 3-Way moderated mediation analysis conducted using PROCESS macro
Full size table

Supplemental Analyses

We conducted a series of supplemental analyses to examine our hypotheses. First, we replicated our results using significant other perceptions of work–life conflict. Second, we performed reverse causality tests. Finally, we replicated our results using structural equation modeling techniques.

Replication with Perceptions from a Significant Other

First, past research has used significant other perceptions of work-to-family conflict in order to not rely solely upon self-report data (Breaugh and Frye 2008), thereby reducing common method bias concerns (Podsakoff et al. 2003). We provide results of our analyses to test if the general relationships in our model are consistent when the employee’s work–life conflict is reported by a significant other. We included the same control variables as before and added two additional controls to account for the gender of the significant other and whether the significant other is a spouse of the employee. The main effect of FRD on work–life conflict was replicated when we used the significant other’s report of the employee’s work–life conflict (b = 0.77, p < 0.01). The non-significant findings for the moderating effect of gender on the FRD—work–life conflict relationship were also reproduced (b = 0.09, NS). Examining the three-way interaction on the direct effect of FRD to work–life conflict as reported by the significant other was as follows (b = 0.29, p = 0.05). Finally, the full model three-way interaction effects were replicated for the indirect effect of emotional exhaustion (b = 0.40, p < 0.05) but not the direct effect on work–life conflict (b = 0.12, NS). On the whole, the reports on employees’ work–life conflict as provided by the significant other were consistent with the employees’ self-reported work–life conflict.

Reverse Causality Tests

Second, since FRD and emotional exhaustion were measured at the same time, it is possible that the order of our model may be reversed such that employees who experience emotional exhaustion may perceive more FRD from their supervisor and, ultimately, more work–life conflict. Work–life conflict could also drive future emotional exhaustion, which in turn could cause higher reports of FRD. We conducted a mediation test of the hypothesized model:

(FRD → emotional exhaustion → work–life conflict) as well as an alternative model with the variables reversed (work–life conflict → emotional exhaustion → FRD).

The results of the mediation analysis in SPSS using the Hayes PROCESS macro Model 4, which tests for mediation, show that the indirect effect for the hypothesized model is 0.22 (95% confidence interval from 0.15 to 0.31) and the b coefficients were as follows: FRD → work–life conflict b = 0.28; FRD → emotional exhaustion b = 0.63; emotional exhaustion → work–life conflict b = 0.35. For the alternative model where the variables were in reverse order, the indirect effect was b = 0.17 (95% confidence interval from 0.16 to 0.42) and the b coefficients were as follows work–life conflict → FRD b = 0.29; work–life conflict → emotional exhaustion b = 0.71; emotional exhaustion → FRD b = 0.24. Therefore, both models are plausible and show a statistically significant indirect effect. The indirect effect is somewhat stronger for the hypothesized model compared to the alternative model.

Replication with Structural Equation Modeling

Finally, we examined our full model (shown in Fig. 1) with structural equation modeling to determine if the results from the PROCESS model were replicated. We modeled power distance, emotional exhaustion, and work–life conflict as latent variables and FRD as an observed variable since having only two indicators per construct can be problematic (Little et al. 1999). We used a parceling approach for the power distance variable given the larger number of items, small sample size, and complexity of the model (Little et al. 2002). Based on the full model results, we found support for a main effect of FRD on emotional exhaustion (b = 0.65, p < 0.05) and work–life conflict (b = 0.28, p < 0.05), lending support to Hypotheses 1a and 1b. We did not find support for a two-way interaction between FRD and gender on emotional exhaustion (b = − 0.17, p > 0.05) as was the case with the PROCESS macro results. However, we did find a two-way interaction for work–life conflict (b = − 0.22, p = 0.05), partially supporting Hypothesis 2b. We also found support for our hypothesized three-way interaction of FRD, power distance, and gender on the indirect effect of emotional exhaustion (b = 0.46, p < 0.05) and on the direct effect of work–life conflict (b = 0.28, p < 0.05), offering support in line with Hypotheses 3a and 3b. Furthermore, we found that emotional exhaustion mediates the relationship between FRD and work–life conflict (b = 0.28, p < 0.05), offering support for Hypothesis 4. The conditional indirect effects were significant for females with low power distance (b = 0.17, p < 0.01), females with high power distance (b = 0.20, p < 0.01), and males with high power distance (b = 0.28, p < 0.01), supporting Hypothesis 4. The conditional indirect effects were not significant for males with low power distance (b = − 0.02, NS). The conditional direct effects were significant only for females with low power distance (b = 0.34, p < 0.01) and were approaching statistical significance for females with high power distance (b = 0.21, p = 0.056). Thus, the results of the structural equation modeling approach are very similar to the results of the PROCESS macro approach using ordinary least squares regression, which strengthens our confidence in the relationships found in this study.

Discussion

While research has suggested that supervisors play an important role in creating family-supportive environments by helping employees manage work–family demands (e.g., Hammer et al. 2009; Kossek and Lee 2017), the effects of family responsibility discrimination (FRD) from a supervisor on emotional exhaustion and work–life conflict are less well known. Using a sample of employees, we found a positive association between family responsibility discrimination and reports of emotional exhaustion and work–life conflict. Gender did not moderate these relationships. However, when gender and power distance were considered together with FRD, we observed the highest reports of emotional exhaustion and work–life conflict from women (regardless of power distance) and from men who scored high on power distance. In other words, the relationship was stronger for employees who experienced gender role expectations (i.e., females) and power distance expectations (i.e., high power distance employees) compared to employees who did not (i.e., males with low power distance). By examining variation in the experience of emotional exhaustion and work–life conflict based on power distance and gender, this study provides a nuanced understanding of how gender norms and cultural values influence employees at work.

Theoretical Implications

We contribute to the work–life literature to illuminate when employees feel the draining effects of FRD. We consider how gender norms along with individual values to appease a supervisor play a role in this experience. Employees who are constrained by values associated with gender and hierarchy are more likely to experience emotional exhaustion following FRD. We suggest that contextual values should be considered when theorizing about how gender differences emerge in the experience of work–life conflict. Our study sheds light on inconsistent findings regarding the role of gender on experiences of emotional exhaustion (Purvanova and Muros 2010) and points to the importance of individual differences in values such as power distance. In our case, the traditionally less encumbered workers (i.e., men) were more likely to feel emotional exhaustion when they valued hierarchical principles (i.e., power distance). This was in contrast to women, who experienced emotional exhaustion regardless of power distance values.

We extend the work of Bagger and Li (2014) who uncovered how a resource gain affects outcome variables by testing how a resource loss affects outcome variables. Specifically, we suggest the indicators of resource loss (i.e., an unsupportive supervisor) and indicators of resource gain (i.e., a supportive supervisor) are differentially related to how individuals experience emotional exhaustion (Lee and Ashforth 1996). Further, although past research suggests that individuals high in power distance are more likely to accept unfairness and inequality (Daniels and Greguras 2014), our research suggests that these employees still experience negative effects of FRD by overextending themselves emotionally in negatively charged situations. This study provides support for a double-edged nature of power distance (Lian et al. 2012) by demonstrating that it can both weaken or strengthen other relationships, depending on the examined outcome. In our case, high power distance strengthens levels of experienced emotional exhaustion as a result of FRD. Thus, while on the surface high power distance employees may appear to be more accepting of unjust behaviors, underneath the surface these employees may suffer in terms of feeling emotionally exhausted and, in our context, experience more work–life conflict.

Finally, this study is also unique in that it addresses FRD, a relatively unexplored form of discrimination (Dickson 2008) that has received increasing attention in the legal system. Despite recent legal and policy-oriented attention due to the increasing number of lawsuits and their associated costs (Scott 2007), research on FRD has been described as exploratory since few researchers have examined this topic (Dickson 2008). This illustrates an opportunity that this paper addressed to help understand boundary conditions and consequences of FRD as a form of family-unsupportive supervision.

Practical Implications

This study has both practical and ethical implications that suggest the importance of understanding how FRD may evoke ineffective coping responses (overextending oneself on the job) and ultimately lead to increased levels of emotional exhaustion and work–life conflict. This suggests that if management can send positive messages to employees and supervisors to invest in both their work and non-work roles, that may help prevent emotional exhaustion caused by high demands from one role that take away from another role. Such practices might include fostering employees’ and supervisors’ positive attitudes toward roles outside of work, dispelling strict gender role expectations that may push workers to overcompensate and burnout, embracing socialization practices that increase the use of family-friendly policies, or developing strong cultures where family responsibilities are not penalized.

Our results also suggest that employees high on power distance experience higher levels of emotional exhaustion because of their values to appease their supervisor. Companies can address this with training on egalitarianism and power effects to help reduce biases toward employees with family or outside responsibilities. Moreover, this study reveals that male employees, especially those who value hierarchy, are also more likely to experience emotional exhaustion. In this case, organizations should look for signs of emotional exhaustion to help determine whether employees need interventions to lower work–life conflict. This is important because directly asking employees may not suffice given that employees high in power distance may evaluate a wide range of supervisory behaviors as appropriate when, in fact, their supervisor’s behaviors may be causing stress-based pressures, emotional exhaustion, and work–life conflict.

Limitations and Future Research

The findings of this study should be interpreted in light of its limitations. We captured variables at two time points, which limits theoretical insights on how the resource loss cycle and depletion unfolds over time. A recent meta-analysis highlights the importance of studying strain and work–family conflict over time to better understand which comes first, the chicken or the egg (Nohe et al. 2015). While our study provided insight on the potential influences of strain on work–life conflict, we were not able to empirically test reciprocal relationships between emotional exhaustion and work–life conflict. As such, future research could conduct a longitudinal study to test if a history of resource loss at work may be of importance when employees encounter current or future resource loss. For example, researchers could collect data on previous supervisor FRD (as opposed to a current supervisor) and test whether having a previous family-unsupportive supervisor makes one more or less likely to experience the negative effects of emotional exhaustion on work–life conflict. This research would help extend theory on the relationship between past resource loss and current resource loss as it relates to the need to acquire and maintain resources of value (Hobfoll and Shirom 1993).

The results of our supplemental analysis testing an alternative model with the variables in reverse order shows a somewhat stronger indirect effect in the hypothesized model compared to the alternative model. However, both models showed a significant indirect effect and were clearly plausible. For the link between FRD and emotional exhaustion, the relationship is stronger with FRD leading to emotional exhaustion rather than the other way around. With respect to the link between emotional exhaustion and work–life conflict, the relationship is stronger when work–life conflict leads to emotional exhaustion rather than the other way around. As is true of most psychological variables, these variables likely have recursive relationships between them such that they mutually affect each other at different points in time. Future research which measures variables at more than two points in time should investigate whether FRD is associated with emotional exhaustion and work–life conflict which, subsequently, leads to more emotional exhaustion.

We also recognize that our measure of family responsibility discrimination from one’s supervisor is based on modifying an existing measure that examined age discrimination from one’s supervisor (Triana et al. 2017). Future research could develop a validated scale of perceived family responsibility discrimination from one’s supervisor. Moreover, while we used a significant other’s report of the employee’s work–life conflict as a way to reduce common method bias, future research could theorize how another person’s report of work–life conflict may differ from one’s own report and why.

In line with previous research on work–life topics (Allen 2001; Eby et al. 2005) our sample was highly educated and employed in white-collar occupations, thus limiting the generalizability of the findings. We encourage researchers to examine diverse samples in terms of education and occupation to understand whether blue-collar workers experience similar results as white-collar professionals. Some research has suggested the results may differ with blue-collar workers because their needs are different and they may experience additional conflict due to scheduling complexity (Stanczyk et al. 2016). As suggested by an anonymous reviewer, the experience of FRD may differ by job level. Perhaps men who are at the top of the organization may not experience as much FRD due to a heavier emphasis on work than on family. Future research may test this possibility.

Conclusion

As gender norms evolve and men take on more caregiving roles while women take on more breadwinning roles in addition to their caregiving roles, it is important to consider both the ethical and practical ramifications of these demographic shifts in the workplace. The present study finds that perceiving family responsibility discrimination (FRD) from one’s supervisor is associated with emotional exhaustion, and ultimately, work–life conflict. The relationship between FRD and emotional exhaustion was always positive for women, and it was also positive for men with high levels of power distance values. Results suggest both ethical and practical issues for organizations, because having family-unsupportive supervisors at work can result in work–life conflict for employees. This is an important ethical question to consider for both organizations and societies, especially for countries like the U.S. which have no federally mandated parental leave requirements to help both men and women balance the demands of work and family.