Work and family domain stressors and support: Within- and cross-domain influences on work–family conflict

Authors

  • Dora M. Luk,

    Corresponding author
    1. Department of Management, City University of Hong Kong, China
      Correspondence should be addressed to Dr Dora Luk, Department of Management, City University of Hong Kong, Tat Chee Avenue, Kowloon Tong, Hong Kong, China (e-mail: mgdora@cityu.edu.hk).
    Search for more papers by this author
  • Margaret A. Shaffer

    1. School of Business Administration, University of Wisconsin, Milwaukee, USA
    Search for more papers by this author

Correspondence should be addressed to Dr Dora Luk, Department of Management, City University of Hong Kong, Tat Chee Avenue, Kowloon Tong, Hong Kong, China (e-mail: mgdora@cityu.edu.hk).

Abstract

The purpose of this study is to examine the within- and cross-domain influences of work and family domain stressors and support on two forms of work–family conflict (i.e. WIF: work interference with family, and FIW: family interference with work). To test our hypotheses, we collected multi-source data from 248 Hong Kong employees and their spouses. Among the proposed work domain antecedents of WIF, time commitment and work role expectation were significant. Among the proposed family domain antecedents of FIW, parental demands were significant. Direct cross-domain effects included family role expectation and parental demand on WIF and work role expectation and family-friendly policies on FIW. Tests of the moderating effects of work and family support resulted in support for both within-domain and cross-domain interactions. Implications for researchers and human resource managers are discussed.

Work–family conflict has received a lot of attention from researchers and the public recently. Most studies have been conducted in Western countries (e.g. Adams, King, & King, 1996; Duxbury, Higgins, Lee, & Mills, 1992; Frone, Russell, & Cooper, 1992; Greenhaus & Parasuraman, 1999; Gutek, Searle, & Klepa, 1991) with limited investigations in other regions of the world (e.g. Aryee & Luk, 1996a, 1996b; Yang, Chen, Choi, & Zou, 2000). Generally, most of the studies conducted outside the USA have been based on existing frameworks developed from a Western perspective. Underlying this perspective is an implicit assumption that work and family are distinct domains. When role expectations from these two domains are mutually incompatible, inter-role stress occurs (Kahn, Wolfe, Quinn, Snoek, & Rosenthal, 1964). Historically, researchers have highlighted changing socio-economic conditions as an impetus to work–family conflict. Such changes, however, are not limited to Western countries and many individuals in developed and developing countries around the world experience difficulties balancing the work and family domains (Joplin, Shaffer, Francesco, & Lau, 2003).

In this paper, we examine work–family conflict within the context of Hong Kong, which is a modern Chinese city with a workforce of about 2 million. The high cost of living in Hong Kong has forced many married women into the workforce, and consequently, the number of dual-earner families has increased. Based on the Hong Kong Annual Digest of Statistics (1995), in 1994 the rate of participation in the labour force was 77.6% for men and 47.1% for women. In 2003 the rate of participation in the labour force for men and women was 74.4% and 49.4%, respectively (Hong Kong Census and Statistics Department, 2004). At the same time, the participation rates among married women continued to expand, so that by 2001 over 50% had entered the workforce. Other relevant shifts, such as increased numbers of women in managerial and professional positions and greater educational attainments for women, have also occurred in Hong Kong. Corresponding shifts in social roles and responsibilities, however, have not kept pace. Social roles in Hong Kong have remained very traditional, with women assuming primary responsibility of child and home care. Movement toward increased sharing of family responsibilities has been minimal (Aryee, 1993; Yalom, 2000), with working parents relying on the extended family or live-in domestic helpers to care for children.

Although changes in socio-economic conditions across countries are comparable, how employees in different countries experience work–family conflict may be culturally bound. According to Trompenaars and Hampton-Turner (1998), cultures differ in terms of how they segregate life roles such as work and family. Cultures with specific values (e.g. the USA and the UK) keep work and family relationships separate, whereas diffuse cultures (e.g. China and Korea) integrate the two domains. Consequently, Chinese tradition views work as more important than leisure, and as contributing to family welfare instead of competing with it (Redding, 1993; Redding & Wong, 1986). Chinese assign lower importance to sufficient time for personal and family life than do Westerners (Redding, 1993) because the most important function of the individual is in the maintenance and preservation of the household. In addition, Chinese society is influenced by the philosophical traditions of Confucianism, and these play an important part in how Chinese view work and family. The key Confucian principles consist of interpersonal harmony (aspiration toward a conflict-free system of social relations), hierarchy (consciousness of personal position in the social system), and collectivism (rejection of personal aggrandizement as a threat to established group hierarchies; King & Bond, 1985). Together, these distinctive cultural characteristics of Hong Kong Chinese suggest the need for a more complex consideration of work and family domain variables.

Our purpose is to develop and test an expanded model of the work–family interface that considers both within- and cross-domain influences on conflict emanating from the work and family domains (i.e. work interference with family and family interference with work). According to role stress theory (e.g. Kahn et al., 1964), stressors in one domain can influence stresses in another. Given the enhanced permeability of boundaries between work and family domains in Hong Kong, our first objective is to explicitly consider the direct effects of work and family demands and resources on two forms of work–family conflict: work interference with family (WIF) and family interference with work (FIW). Role stress theory also asserts that resources may buffer the influence of stressors on work–family conflict and that demands may exacerbate the effects (e.g. Greenhaus & Parasuraman, 1986). Again, considering the less distinct borders between work and family for Chinese and the importance of hierarchy, we look at the interactive relationships of work and family support resources and stressors both within and across domains. For within-domain relationships, we expect support resources to buffer or attenuate the influence of stressors on work–family conflict. However, for cross-domain relationships, we contend that resources from the opposing domain become a demand that exacerbates the influence of stressors on work–family conflict. We test these within- and cross-domain relationships with multi-source data from a matched sample of Hong Kong employees and their spouses.

Development of a model of work–family conflict

The proposed theoretical framework is presented in Fig. 1. This model is based on a bi-dimensional conceptualization of work–family conflict. According to recent studies (Aryee, Fields, & Luk, 1999a; Carlson, Kacmar, & Williams, 2000), work–family conflict is multidimensional, with effects from the work domain influencing WIF conflict and effects from the family domain influencing FIW conflict. Based on the cultural differences between the East and the West and the less distinct boundaries of work and family in Chinese society, we believe that each form of conflict will have both domain-specific (i.e. work stressors and support will influence WIF, and family stressors and support will influence FIW) and domain-spanning antecedents (i.e. work stressors and support will influence FIW, and family stressors and support will influence WIF). Building upon Western studies and taking into consideration different social and cultural conditions in Hong Kong, we expand current conceptualizations of work–family conflict and anticipate complex (within- and cross-domain) interactive effects for work and family domain support resources.

Figure 1.

Model of work–family conflict.

In the next sections, we develop hypotheses for the direct influence of work- and family-domain stressors and support on work–family conflict (i.e. WIF and FIW). The proposed relationships are based on stress theories, such as Hobfoll's (1989) conservation of resources model, which posit that stressors in the form of excessive demands or insufficient resources will result in stressful outcomes such as work–family conflict. In addition, we believe greater demands, regardless of the domain from which they emanate, will lead to greater conflict. We then consider competing arguments for the within- and cross-domain moderating effects of work and family support resources. Within domains, domain-specific resources will buffer the effects of stressors on work–family conflict. In developing hypotheses for cross-domain effects, however, we argue that a resource in one domain may become a demand when considered from the perspective of another domain. Consequently, cross-domain resources will aggravate the impact of stressors on work–family conflict.

Direct effects of work and family influences on WIF and FIW

Work domain stressors and support

In this section, we consider two job demands (time commitment and work role expectations) that have been associated with WIF. Time commitment to work refers to the actual number of hours spent at work. Work role expectations are defined as perceived pressures on an individual to assume work role responsibilities. According to Greenhaus and Beutell (1985), pressures associated with one type of role (e.g. work) make it physically impossible to comply with expectations arising from other roles (e.g. family). This is consistent with Hobfoll's (1989) conservation of resources theory, which argues that time and energy are exhaustible commodities; once spent they are not available for other tasks either within the same domain or other domains. In other words, higher demands and expectations in one role (e.g. work) result in fewer resources for other roles (e.g. family). Research strongly supports the influence of time commitment to work (e.g. Burke, Weir, & DuWors, 1980; Frone, Yardley, & Markel, 1997; Pleck, Staines, & Lang, 1980) and work role expectations (Higgins, Duxbury, & Irving, 1992) on WIF. Although previous researchers have not considered the relationship between work stressors and FIW, we contend that these demands will also affect this form of work–family conflict. As time and energy demands in the work place increase, more opportunities exist for family to intrude on this domain. From a cultural perspective, blurred boundaries between family and work for Chinese enhance the likelihood that high commitment to work, in terms of time and/or expectations, will affect both WIF and FIW. Thus, we propose:

Hypotheses 1a–b. Time commitment to work is a significant positive predictor of (a) WIF, and (b) FIW.

Hypotheses 2a–b. Work role expectation is a significant positive predictor of (a) WIF, and (b) FIW.

Various work resources have also been associated with WIF. In this study, we consider two forms of support resources: instrumental and social. Instrumental support includes suggesting ways, or actually lending a hand, to complete a task. Organizations that offer family-friendly policies provide this form of support. Social support in the work domain may come from a number of sources, such as peers or supervisors, who create a more positive work environment (Roskies & Lazarus, 1980). Social support also includes caring and listening sympathetically as well as providing empathy. Applying Hobfoll's (1989) conservation of resource theory to work–family conflict, Grandey and Cropanzano (1999) argued that those who have more resources, such as support, will experience less stress and reduced levels of work–family conflict. Within the work domain, both family-friendly policies (i.e. instrumental support such as child-care facilities and flexitime) and supportive supervisors (i.e. social support) have been related to work–family conflict (e.g. Allen, 2001; Aryee & Luk, 1996b; Kim & Ling, 2001; Nielson, Carlson, & Lankau, 2001; Thomas & Ganster, 1995; Wiersma, 1990). Insofar as these supportive workplace policies and practices often target the family domain, they should also help to alleviate the extent to which family interferes with work. Thus, we propose that work domain instrumental support and social support are associated with lower levels of WIF and FIW.

Hypotheses 3a–b. A family-friendly policy is a direct negative predictor of (a) WIF, and (b) FIW.

Hypotheses 4a–b. Supervisor support is a direct negative predictor of (a) WIF, and (b) FIW.

Family domain stressors and support

Similar to stressors in the work domain, we expect family time commitment (i.e. hours spent on family matters) and role expectations (i.e. perceived pressures to fulfil family role responsibilities) to act as demands that reduce resources available for the work domain (Hobfoll, 1989) and that trigger FIW and WIF. Employees who spend a great deal of time with their families and who have greater family role expectations find it difficult to spend sufficient time and energy at work (Greenhaus & Beutell, 1985; Higgins et al., 1992). With the increased blending of work and family domains within a Chinese context, it is likely that commitment to the family domain will also spill over and affect the work domain. That is, as a result of strong attachment to family, work demands may be perceived as being more intrusive to the family domain. Thus, we propose:

Hypotheses 5a–b. Time commitment to family is a significant positive predictor of (a) WIF, and (b) FIW.

Hypotheses 6a–b. Family role expectation is a significant positive predictor of (a) WIF, and (b) FIW.

Another source of stress within the family domain is parental demands. According to Bedeian, Burke and Moffett (1988), younger children are more dependent on their parents than are older children. In particular, the years before school require a greater commitment of parental time and energy. Drawing on Hobfoll's (1989) conservation of resource theory, Grandey and Cropanzano (1999) pointed out that the number of children living at home relate to family-role stress and work–family conflict. Having more children at home means less time and energy can be devoted to work. This phenomenon is particularly prominent in Chinese society. Asian parents have the highest expectations on their children (Crystal et al., 1994). Parents in Hong Kong are concerned about their children, especially their academic achievement and upbringing, and this concern may eventually lead to heavier parental demands. As they become more involved in their parental role and spend more time in this role, they need to make sacrifices in other areas of their lives, such as work or other family responsibilities. Therefore, we propose:

Hypotheses 7a–b. Parental demand is a significant positive predictor of (a) WIF and (b) FIW.

There are some special features of life in Hong Kong that may help to ease the problem of work family conflict. One of these is the employment of domestic helpers. In a survey conducted by the Census and Statistics Department of Hong Kong in 1993, it was estimated some 69,600 households employed one or more domestic helpers. This represented 4.4% of all domestic households in Hong Kong. From the 2001 population census (Hong Kong Census and Statistics Department, 2001), there were 160,527 households employing one or more domestic helpers; this accounts for 7.8% of all domestic households in Hong Kong. With domestic help, family members need to perform less housework, which results in more time to concentrate on families and work. With this aspect of Hong Kong life in mind we propose:

Hypotheses 8a–b. Domestic helper support is a direct negative predictor of (a) WIF, and (b) FIW.

Moderating effects of work and family domain support on WIF and FIW

Most models of job stress agree that various psychological resources and demands may have important buffering or exacerbating influences (e.g. Greenhaus & Parasuraman, 1986; Kahn et al., 1964). Cohen and Wills (1985) postulated that support plays a stress-buffering role; adequate support intervenes between the experience of stressors and stress. Several researchers have explored work–family conflict and the effectiveness of stress-buffering variables and coping behaviour for reducing work–family conflict (Adams et al., 1996; Aryee, Luk, Leung, & Lo, 1999b; Thomas & Ganster, 1995). Most of this research, however, has focused on within-domain effects. After elaborating more on these domain-specific influences, we present arguments for cross-domain effects.

Within-domain moderating effects

Considerable evidence exists for the moderating effects of work and family support on relationships involving stressors and strains within their respective domains. Carlson and Perrewe (1999) suggested that the work stress–burnout relationship was ameliorated by supportive relationships in the work environment (e.g. co-workers, supervisors) and that family support buffers the effects of family stressors on work–family conflict. According to Hobfoll's (1989) conservation of resources theory, the resources available to an individual will affect how that individual reacts to stress (Grandey & Cropanzano, 1999). Although results for the buffering effects of work and family resources have been inconsistent, the evidence tends to support this relationship. Instrumental support in the work domain, such as family-friendly policies, has played an important role in moderating the effects of work stressors and work–family conflict (Carlson & Perrewe, 1999; Eisenberger, Huntington, Hutchison, & Sowa, 1986). Also, given similar work stressors, lower work–family conflict was found for employees who had supportive supervisors (Goff, Mount, & Jamison, 1990; Jones & Butler, 1980). Within the family domain, the linkage of parental demands to work–family conflict was moderated by spouse social support (Matsui, Ohsawa, & Onglatco, 1995). Some research (e.g. Fu & Shaffer, 2000) has also investigated the moderating effect of instrumental support provided by domestic helpers. Consistent with past research, we propose:

Hypotheses 9a–b. Family-friendly policies will moderate the relationships between (a) work time commitment, and (b) work role expectation and WIF, such that the relationships will be weakened.

Hypotheses 10a–b. Supervisor support will moderate the relationships between (a) work time commitment, and (b) work role expectation and WIF, such that the relationships will be weakened.

Hypotheses 11a–c. Domestic helper support will moderate the relationships between (a) family time commitment, (b) family role expectation, and (c) parental demand and FIW, such that the relationships will be weakened.

Cross-domain moderating effects

Although previous research has examined social support as a promising coping mechanism, questions as to how social support affects work–family conflict remain unanswered (Carlson & Perrewe, 1999). As noted previously, researchers have tended to focus on relationships among variables within one domain, such as the effect of work domain support on the relationship between work domain stressors and WIF. Despite the fairly strong evidence for the moderating influence of within-domain support, several studies have failed to corroborate these findings or have produced inconsistent results (Frone, Russell, & Cooper, 1995; Parasuraman, Greenhaus, & Granrose, 1992). In some cases, reverse-buffering effects have been examined, but these have been restricted to situations in which the source of the stressor and support are from the same domain (e.g. Beehr, Farmer, Glazer, Gudanowski, & Nair, 2003). Again, mixed support for these effects has been reported (Beehr & Glazer, 2001; Glaser, Tatum, Nebeker, Sorenson, & Aiello, 1999).

Although support within a domain normally has a buffering or weakening effect on relationships among variables within that same domain (Aryee et al., 1999a, 1999b; Carlson & Perrewe, 1999; Greenhaus & Parasuraman, 1986; Matsui et al., 1995; Nielson et al., 2001; Thomas & Ganster, 1995), the moderating effect may be enhanced (i.e. a reverse-buffering effect) when two different domains are involved (e.g. Shaffer, Harrison, Gilley, & Luk, 2001). According to the conservation of resources model (Hobfoll, 1989), excessive demands, insufficient resources or both within a particular role domain or between domains may result in negative affective and behavioural outcomes. A loss of resources is especially threatening when one domain is dependent on resources from another (Burke, 1991). From a cognitive dissonance perspective (Festinger, 1957), when support from another domain is received, an implicit obligation to reciprocate may result in increased anxiety and tension rather than reduced stress.

We believe that it is possible that support from an opposing domain becomes an additional demand when it operates within the context of another domain. For example, having a domestic helper at home is generally a resource that helps to reduce involvement in housework and child care. However, when the domestic helper is sick or has personal problems, this resource becomes a demand that drains time and energy away from other activities (i.e. work). Similarly, support from a supervisor may become a demand when an individual is struggling to deal with family problems. Cognitively, the support of the supervisor requires reciprocation in the form of work commitment. When faced with increased demands within the family domain, such cognitive work obligations may exacerbate the experience of FIW. Based on this argument, we propose that cross-domain moderators will have an exacerbating effect on work–family conflict.

Hypotheses 12a–c. Family-friendly policies will moderate the relationships between (a) family time commitment, (b) family role expectation, and (c) parental demand and FIW, such that the relationships will be strengthened.

Hypotheses 13a–c. Supervisor support will moderate the relationships between (a) family time commitment, (b) family role expectation, and (c) parental demand and FIW, such that the relationships will be strengthened.

Hypotheses 14a–b. Domestic helper support will moderate the relationships between (a) work time commitment, and (b) work role expectation and WIF, such that the relationships will be strengthened.

Method

A field study was conducted to collect primary data by distributing questionnaires to Hong Kong Chinese couples with children. Employee and spouse surveys were sent to each household so that measures of work–family conflict could be obtained from two sources. This was done to mitigate problems associated with common method variance. Questionnaire packages were delivered to five local schools (from kindergartens to secondary schools), which distributed them to parents. To reduce the chances of couples sharing their opinions and biasing their responses, we asked them not to discuss their answers with each other until after they each completed the questionnaires. We also provided separate self-addressed stamped envelopes for them. For dual-income couples (i.e. when both partners worked), we requested they randomly select which partner would complete the employee survey. Drawing upon techniques used by experimental researchers (Sekaran, 2000), we asked the partners to toss a coin. The one who chose and got heads was instructed to complete the spouse survey; the one who chose and got tails was instructed to complete the employee survey. If they were not a dual-income couple (traditional families), we asked them to let the primary breadwinner answer the employee survey and the other partner to complete the spouse survey. In this way, we tried to avoid any possible gender biases. To encourage participation in this research, we included a self-addressed, stamped postcard for respondents to return for entry into the drawing of a tea-buffet at one of the leading hotels in Hong Kong.

Of the 2,050 distributed questionnaire packages, 341 were returned by employees and 320 were returned by spouses. Schools were responsible for distributing the surveys; therefore, we are not sure whether surveys actually reached the intended respondents. Follow-up efforts revealed that several schools did not distribute all our surveys to their students' parents. Thus, our response rates of 16.6% for employees and 15.6% for spouses are probably underestimates of actual responses. We created a combined sample of data from employees and spouses who both completed our surveys, resulting in a matched sample of 248. This matched sample represents 73% of the employees who responded.

Sample

Among the employee respondents, 79% were men and 21% were women; all were married. The mean age of respondents was 41.9 years old. A total of 79% were from traditional families (83% males and 17% females), and 21% were from dual-earner families (67% males and 33% females). The average number of children was 2; 26.4% of families had 1 child, 55.3% had 2 children, and 18.8% had more than 2 children. The percentage that had at least one domestic helper was 31.1%; over 50% of the local families earned in the range of 100,000 and 300,000 Hong Kong dollars per year. About 80% were educated at diploma level or below, and 13.7% had a bachelor's degree. Of these, 50.6% worked at managerial or professional levels, 28.5% were at the supervisory level, and the remaining 20.9% worked in non-managerial or frontline positions. Respondents worked an average of 39.2 hours per week. The characteristics of this sample are comparable to the population in Hong Kong. According to the Hong Kong Statistics Report (2004), the average number of children is 1.6 and the average domestic household size is 3.3. Out of all the working population, 56.2% are male and 43.8% are female, with 45.1% earning an average domestic income in the range of 100,000 and 300,000 Hong Kong dollars. About 79.7% have been educated at diploma level or below, and 13.4% have a bachelor's degree; 5.5% work at managerial or professional levels, 10.7% are at the supervisory level, and the remaining 64.1% work in non-managerial or frontline positions.

Measures

Operationalizations of the variables are described below. Unless otherwise indicated, measures were on a 1–5 disagree/agree scale.

Work–family conflict

To assess work–family conflict, we used eight items devised by Gutek et al. (1991). Four items were used to measure WIF, and the other four were used to measure FIW. An example of an item from the WIF scale was, ‘I have so much work to do on the job that it takes time away from my personal interests.’ An example of an item from the FIW scale was, ‘I'm often too tired at work because of the things I have to do at home.’ To minimize the problem of common method variance (Podsakoff & Organ, 1986), both the employee and the spouse provided ratings of the WIF dimension of work–family conflict. The correlation for these ratings was r=.46, suggesting a fairly strong correspondence between employee and spouse perceptions. Because spouses are less likely to know the extent to which family is interfering with work, we did not ask them about FIW.

Work and family stressors

Work time commitment was assessed with a single item that asked respondents to report the total number of hours devoted to work-related activities per week (Frone et al., 1997). Work role expectation was measured by using a 4-item scale from Cooke and Rousseau (1984). A sample item was, ‘I take on work-related duties and responsibilities, even though these activities may interfere with my free time.’ Family time commitment was represented by a single item that asked about time commitment to family responsibilities (Frone et al., 1997). Family role expectation was measured with two questions from Quinn and Staines (1979). The respondents had to indicate who has the main responsibilities for household chores and care of the children. These two positions were dummy coded to indicate whether or not the person was responsible for family chores or taking care of children (0 = not responsible, and 1 = responsible). These two questions were then added together. Parental demand was assessed using a scale developed by Bedeian et al. (1988). There were all together four groups depending on the ages of children: (1) one or more children older than the age of 22 but none under 22; (2) one or more children between 19 and 22 but none under the age of 19; (3) one or more children between 6 and 18 but none under 6; (4) one or more children under 6. The total number of children was added to this coding scheme to obtain an overall measure of parental demands.

Work and family support resources

Family-friendly policies were measured by a 5-item scale developed by Judge, Boureau, and Bretz (1994). An example of the questions asked was, ‘my organization provides programmes to assist in balancing demands of dual-earner couples.’ Supervisor support was measured with si items based on the work of Fernandez (1986), and Greenberger, Goldberg, Hamill, and O'Neil (1992). A sample item includes, ‘my supervisor is willing to listen to my personal problems’. Domestic helper was assessed with a single item that asked respondents to report the number of domestic helpers they employed. Responses ranged from 0 to 2.

Results of analyses

Means, standard deviations, zero-order correlations and estimated reliabilities for the measures used in this study are presented in Table 1. We included age and education as control variables at the initial stage of analysis; however, the results showed that they had no effect, so to conserve power we excluded them from the regressions. We used multiple hierarchical (moderated) regressions to test the hypotheses of direct and indirect effects (see Tables 2 and 3). To test the moderator hypotheses, we first created interaction terms by multiplying each of the independent stressor variables with the support variables. We then entered these into each regression equation after the main effects. The results are presented in Table 3.

Table 1.  Means, standard deviations and correlationsThumbnail image of
Table 3.  Regression results of moderating effects on WIF and FIWThumbnail image of Thumbnail image of
Table 2.  Regression results of direct effects on WIF and FIWThumbnail image of

Work domain stressors and support

According to Hypotheses 1a–b and 2a–b, two work domain stressors will affect WIF and FIW. As predicted by Hypothesis 1a, work time commitment (β=.18 and .21, p<.01 for employee and spouse ratings of WIF, respectively) had a positive influence on WIF. To support Hypothesis 2a, work role expectation (β=.16, p<.05 and .20, p<.01 for employee and spouse ratings) was a significant positive predictor of WIF. Contrary to our hypothesis, work role expectation (β=−.14, p<.05) was a significant negative predictor of FIW (Hypothesis 2b). No significant effects were found for work time commitment (Hypothesis 1b) on FIW. None of the hypotheses (3a–b, 4a–b) predicting the effects of work domain support on WIF and FIW were supported. Although we found a significant effect of family-friendly policies (Hypothesis 3b) on FIW, it is a positive predictor (β=.23, p<.05), which is contrary to our hypothesis.

Family domain stressors and support

Hypotheses 5–7 tested for the effects of family domain stressors on WIF and FIW. The result shows that family role expectation (β=−.14, p<.05) is a negative predictor of WIF; therefore, Hypothesis 6a is not supported. Parental demands are positive predictors of both WIF (β=.17, p<.05) and FIW (β=.18, p<.05) which supports Hypotheses 7a and 7b. No significant effects were found for family time commitment on WIF (Hypothesis 5a) or FIW (Hypothesis 5b) and family role expectation on FIW (Hypothesis 6b).

Within-domain moderating effects of support

From Hypotheses 9 to 11, we predicted that support from the same domain as the stressors will buffer the relationship between stressors and work family conflict. That is, family-friendly policies and supervisor support will moderate relationships between the work domain stressors and WIF, and domestic helper support will interact with the family domain stressors to influence FIW. Among all these interactions, only family-friendly policies moderates the relationship between work time commitment (β=−.38, p<.05) and WIF (Hypothesis 9a). As predicted, family-friendly policies buffer or weaken the relationship between work time commitment and WIF (See Fig. 2a). No support was found for other within-domain moderating effects.

Figure 2.

Plot of significant moderating effects. (a) Plot of the moderating effects of family-friendly policies on the relationship between work time commitment and WIF. (b) Plot of the moderating effects of family-friendly policies on the relationship between family role expectation and FIW. (c) Plot of the moderating effects of supervisor support on the relationship between family role expectation and FIW. (d) Plot of the moderating effects of domestic support on the relationship between work role expectation and WIF.

Cross-domain moderating effects of support

According to Hypotheses 12–14, support from cross-domain variables will exaggerate relationships between the work domain stressors and WIF and between family domain stressors and FIW. The results show that family-friendly policies enhance the relationship between family role expectations and FIW (β=.24, p<.10), which supports Hypothesis 12a (see Fig. 2b). As predicted by Hypothesis 13b (see Fig. 2c), supervisor support also exacerbates the relationship between family role expectation and FIW (β=.24, p<.05). In contrast with Hypothesis 14b (see Fig. 2d), domestic support reduces the effect of work role expectations on WIF (β=.27, p<.01).

Discussion

The results of this study provide evidence for the influence of both domain-specific and domain-spanning stressors and support resources on both forms of work–family conflict (WIF and FIW). Relative to work and family stressors, resources specific to both domains did little to attenuate their corresponding forms of work–family conflict. Targeting a research context (i.e. Hong Kong) in which the boundaries between work and family are less demarcated, we were able to demonstrate that cross-domain variables played an important role in the experience of work–family conflict. These relationships were more complex, with some stressors having a negative effect on work–family conflict. As predicted, however, cross-domain support resources generally exhibited an exacerbating effect on stressor–strain relationships within the opposite domain.

Consistent with most of the work–family conflict research (e.g. Greenhaus & Beutell, 1985; Greenhaus, Collins, Singh, & Parasuraman, 1997), time commitment towards work and work role expectations had a great impact on WIF. The significant negative effect of work role expectations on FIW, however, was contrary to our predictions. One reason may have to do with the work values of Hong Kong Chinese. In Chinese societies, work is viewed as a means of enhancing family well-being (Yang et al., 2000). If Hong Kong Chinese experience stress from work, because of psychological involvement in work or the expectations of their work, then they still perceive that their hardship is necessary for family welfare. Redding (1993) explains how the Chinese strive to bring honour and prosperity to their families through their work. Work is not something that interferes with family life; instead, it is a means of providing for the family.

Also opposite to our hypothesis, family role expectations had a negative effect on WIF. According to Westwood (1992), individuals in a Chinese society will put the interests of the family above their own, and above any wider interests. For Chinese, family relationships are strong and persistent. They tend to have low importance assigned to sufficient time for personal and family life (Bond & Hwang, 1986; Chia et al., 1994; Redding, 1993). Most importantly, they view work as being clearly more important than leisure, and as contributing to family welfare instead of competing with it (Metzger, 1997; Redding, 1993; Redding & Wong, 1986). Perhaps because of this strong link between work and family values, when pressures to fulfil family responsibilities are high, Chinese are less likely to perceive work as something that interferes with family.

Parental demand was a positive predictor of both WIF and FIW. This finding is consistent with the literature (e.g. Aryee et al., 1999b; Bedeian et al., 1988; Greenhaus et al., 1997) and our argument about Hong Kong parents' expectations of their children. In contrast with the literature, however, none of the other family domain variables had a significant effect on FIW. A possible explanation is that the overwhelming majority of respondents (83.3%) in this sample were male. Results from past research indicate that men balance work and family identity without trading off one for the other (Dibenedetto & Tittle, 1990). As mentioned earlier, Aryee et al. (1999b) proposed that Hong Kong Chinese assign greater importance to family roles. As Hong Kong Chinese, our respondents perceived that family is the main part of their life, so time committed to their families was absolutely necessary. At the same time, Hong Kong Chinese maintain very strong gender role stereotypes: they believe males have the responsibility to work and to improve the family welfare (Kao, 1987). Therefore, family time commitment and family role expectation had no impact on FIW.

Our hypotheses for the direct effects of work domain support, such as family-friendly policies and supervisor support, and family domain support, such as domestic helper support, on WIF and FIW were not substantiated. Even though family-friendly policies had a significant influence on FIW, it was in an opposite direction to what we proposed. This might be due to the lack of these types of policies provided by organizations in Hong Kong, and to the larger power distance in Hong Kong. It is also possible that these policies signalled to the employees that they were allowed to have family matters intrude on their jobs. Further research is needed to understand the possible negative ramifications of these instrumental forms of work support.

Past research has demonstrated contradictory results for the moderating role of within-domain support resources (Frone et al., 1995; Parasuraman et al., 1992). Our results are also contradictory. Only one interaction involving domain-specific support resources was significant, and it was in the opposite direction hypothesized. For those reporting greater levels of family-friendly policies, WIF was higher across all levels of work time commitment. In contrast with Western research, family-friendly policies seem to act as an additional stressor for our Chinese respondents. Perhaps the inherent reciprocity that is so characteristic of Chinese culture requires employees to respond to family-friendly policies by working harder. This may be traced to Confucian values, which play a predominant role in moulding Chinese characteristics and behaviour. Chinese perceive the workplace as not only a place to work but also a place to reside (Lai, 1995; Shaw, 1996). Employees have an obligation to maintain a good relationship (harmony) with the organization, which is responsible for the well-being of its workers. As a result, Hong Kong employees have even less time to spend with family. To fully understand the dynamics of this, however, more research is needed.

Challenging past research, we proposed and tested for cross-domain effects: work domain support will moderate the relationships between family domain variables and FIW, and family domain support will moderate the relationships between work domain variables and WIF. As predicted, cross-domain work support resources were transformed into a kind of demand that worsened the stress of FIW. As depicted in Fig. 2b, the overall level of FIW is higher across all levels of family role expectation for those with greater family-friendly policies. Also, as presented in Fig. 2c, supervisor support exacerbates the relationship between family role expectation and FIW. That is, those with supportive supervisors reported higher levels of FIW across all levels of family role expectations. This suggests that those who are more engaged (psychologically and in practice) in their family role find it easier to focus on family if they have non-supportive organizations and supervisors. From a Chinese cultural perspective, these phenomena could be explained by the Confucian's five most important relationships that people should maintain (ruler/subject, father/son, older brother/younger brother, husband/wife, and older/younger friend), which emphasize the importance of maintaining a hierarchy of relationships. If employees have supportive supervisors or supportive organizations, they have stronger obligations to return high-quality work in order not to disappoint their supervisors and organizations. In addition, Confucian philosophy values harmonious relationships, so workers are required to build good relationships with their supervisors and organizations to secure favourable treatment (Shenkar, Ronen, & Chow, 1988). Insofar as work supportive policies and supervisors become a demand that requires a reciprocation of commitment to their jobs, greater demands from the family result in more FIW.

In contrast with the work support moderators, our results suggest that domestic helpers weaken the effect of work role expectations on WIF (see Fig. 2d). That is, with more domestic helper support, all levels of WIF are lower across all levels of work role expectation. This result supports the traditional view that having a domestic helper at home is generally a resource that helps to reduce workload and involvement in family matters. For employees who have a domestic helper at home, even if they have high work role expectations, the level of WIF is lower. Although this buffering effect is contrary to our predictions, an explanation for it could also stem from Confucian philosophy. Domestic helpers do not fit into the hierarchy of important relationships – they are hired help, so no reciprocation is required as a result of them doing their assigned duties. Because these findings contradict past studies (e.g. Fu & Shaffer, 2000), further research is needed to look at the effect of domestic helpers on work family conflict.

There are several limitations in this study. Firstly, this study was a cross-sectional field study; only correlations among the variables could be examined, and no causal inferences could be made. For example, we argued that supportive supervisors require greater commitment to work and higher levels of FIW. However, it is possible that as supervisors become more aware of employees' FIW, they provide additional support. Although difficult to conduct, longitudinal studies should be undertaken to allow for stronger causal inferences and a better understanding of the timing associated with such conflicts. Secondly, in this study, we only collected data from employees who had children in school; therefore we may have a truncated sample. Thirdly, we do not have any way to verify that each spouse completed the appropriate survey. For those in traditional family structures, however, it is unlikely that spouses would have been able to answer some of the background questions about the job and the organization. For those in dual-income structures, we tried to avoid gender bias by encouraging couples to randomly select a survey (i.e. employee or spouse survey) based on the result of a coin toss. Insofar as 33% of the dual-earner employees who completed the employee survey were female (compared with 17% females in the traditional family structure), we think this strategy was successful. However, future researchers may want to be more specific in assigning surveys to particular individuals. Finally, generalizability is also limited, as this study consisted only of married couples with children. As family structures continue to evolve and care of parents becomes a greater concern, the determinants of work–family conflict may differ from those proposed here.

The findings of this study contribute to both the work–family conflict literature and human resource management. The theoretical contributions of this study are (1) a systematic, comprehensive assessment of work–family conflict that integrates Western models and the Hong Kong context; (2) a consideration of variables that may be especially relevant to the Hong Kong context (e.g. instrumental support such as domestic helper support); and (3) an assessment of direct and indirect influences of both within- and cross-domain stressors and support resources on WIF and FIW. From a practical perspective, our study acts as a guideline for firms that value employees' integration of work and family responsibilities. Although family-friendly policies did not seem to reduce WIF, it is possible that our measure was too general or it may be due to the lack of this kind of policy in Hong Kong. A consideration of more specific practices, such as flexible work schedules, child care assistance and parental leave, might find more evidence for the role of these practices in assisting employees who strive to balance their work and family responsibilities. This is becoming increasingly important for those with elderly dependents and children at home.

In conclusion, this research has rigorously examined the work and family domain stressors and support of work–family conflict experienced by Hong Kong Chinese. We examined and tested the complexity of work and family relationships by expanding Western models to take into consideration the more fluid boundaries between work and family within a Chinese context.

Acknowledgements

An earlier version of this paper was submitted to the 2002 Asia Academy of Management Meetings, Bangkok.

Ancillary