Job Stress, Depression and Work-to-Family Conflict A Test of the Strain and Buffer Hypotheses Stress au travail, dépression et conflit travail-famille Un test des hypothèses de tension et de tampon Estrés de trabajo, depresión y conflicto trabajo-familia Un test de tensión e hipótesis de amortiguadores

Abstract

In this paper, the Job Demand-Control (JDC) model is used to predict depression and work-to-family conflict for married lawyers working full-time. The objectives of this paper are: (1) to determine whether the JDC model applies to work-to-family conflict; (2) to incorporate domain-specific job demand and job control variables; and (3) to examine a wider array of different forms of social support. First, the JDC model also helps explain work-to-family conflict. Second, domain-specificity does not appear key to documenting the buffering effects for job control. Third, spouse’s support of one’s career has the strongest main effect on both depression and work-to-family conflict, whereas coworker support functions as a moderator of lawyers’ job demands and has both buffering and amplifying effects. This paper closes by discussing the possible conditions under which members of support systems may transfer or exacerbate stress effects rather than alleviate them.

Résumé

Le modèle du contrôle des exigences d’une tâche (CET) est utilisé ici en vue de prédire la dépression et le conflit travail-famille chez des juristes mariés et travaillant à plein temps. Cet essai cherche à évaluer dans quelle mesure le contrôle de la tâche (en termes de la latitude dans l’emploi de son temps au travail et le choix du nombre d’heures) et le support social servent de tampons aux effets négatifs des exigences excessives de la tâche chez les avocats en exercice. Le but premier de cette recherche consiste aussi à vérifier si le modèle CET, dont on se sert pour comprendre la dépression dans les travaux sur le stress, s’applique ou non au conflit travail-famille. Un second objectif consiste à retenir des variables de contrôle de la tâche rattachée à un domaine particulier comme autant de ressources éventuelles aidant à affronter des situations stressantes. Un troisième objectif consiste à analyser un éventail plus large de types de soutien social (appui des collègues de travail, de l’entreprise, du conjoint ou de la conjointe, etc.) pour vérifier si, au cours d’une période de stress, une forme de soutien est plus efficace ou non qu’une autre à titre de tampon face aux exigences de la tâche.

Dans ce travail, les observations sont fondées sur des réponses à une enquête auprès de 1201 juristes mariés et travaillant à plein temps en droit, en Alberta, en juin 2000. L’échantillon comprend 365 femmes (30 %) et 836 hommes (70 %) qui ont rempli les questionnaires expédiés par la poste. Les avocats travaillaient dans une diversité de milieux de travail dont des sociétés d’avocats, des bureaux de grandes entreprises et des services publics.

On a vérifié, dans un premier temps, les hypothèses de l’effet de stress en analysant les effets respectifs des exigences de la tâche (charge de travail, horaires, activités extra-professionnelles, travail à la maison), du contrôle de la tâche et du soutien social sur la dépression et sur le conflit travail-famille. Ensuite, on a abordé les hypothèses de l’interface tampon-tâche en cherchant à savoir si les effets croisés des termes de l’interaction entre les exigences de la tâche et le contrôle de la tâche et entre les exigences de la tâche et le soutien social exerçaient une influence statistiquement significative sur la dépression et le conflit travail-famille.

En revenant au premier objectif de cette recherche, nous observons que les modèles qui s’appliquent à la dépression dans les travaux sur le stress valent aussi pour l’étude du conflit travail-famille. Trois des quatre variables liées aux exigences de la tâche contribuent de façon significative au conflit travail-famille, alors que le contrôle du temps de travail réduit ce même conflit; le soutien social agit de manière à tempérer les effets des variables reliées aux exigences de la tâche. De plus, le modèle qu’on a vérifié dans cette étude rendait compte de plus de la moitié de la variance dans l’explication du conflit travail-famille. Ces observations nous incitent à croire que les chercheurs, qui désirent accéder à une meilleure compréhension du conflit travail-famille, peuvent retenir le modèle du fonctionnement du stress, qu’on retrouve dans les travaux empiriques et théoriques largement connus, plus particulièrement celui de Karasek.

Dans un deuxième temps, on a introduit dans le modèle CET des variables liées au contrôle de la tâche et aux exigences de la tâche inhérentes au domaine. Puisque les avocats constituent l’échantillon de cette étude, les exigences fondées sur le temps de même que sur le contrôle du travail occupèrent le centre de notre analyse. On soutenait que si les variables liées aux exigences de la tâche et celles reliées au contrôle se rencontraient au sein d’un même domaine particulier, il était plus probable que la maîtrise de la dimension temporelle de son travail serve de tampon aux effets des exigences de temps liées à la pratique du droit. Les données n’ont apporté aucun appui à cette dimension du modèle CET. Il semble donc que la spécificité du domaine ne fournit pas nécessairement la clef à l’explication des effets tampon que le modèle propose. Il ne faut pas pour autant penser que le contrôle de la tâche n’est pas une variable importante dans la compréhension du phénomène de stress au travail. Les observations ici démontrent que la maîtrise de l’horaire de travail contribue directement au bien-être du juriste, indépendamment des exigences de la tâche, plutôt que de servir de tampon dans des conditions d’exigences élevées.

Le troisième but de ce travail consistait à examiner un éventail plus vaste de types diversifiés de support social en vue de vérifier si un type en particulier constituait ou non un tampon plus efficace qu’un autre dans le développement du stress au travail. Encore là, nos observations font état de modèles intéressants, qui font ressortir l’importance des types discriminants de soutien social. Une remarque particulière est à l’effet que l’appui des collègues de travail apparaît inefficace, parce que cette variable n’a pas d’effets importants tant sur la dépression que sur le conflit travail-famille. C’est seulement en effectuant des tests d’interaction qu’il est possible d’identifier l’effet tampon important de l’appui des collègues qui consiste en une réduction de l’effet négatif d’une surcharge de travail sur l’état de dépression et sur le conflit travail-famille. Ceci démontre bien l’importance de modéliser et de vérifier le type opérationnel de relations, tel que les écrits sur le sujet le présentent.

Une deuxième observation importante impliquant le support social est à l’effet que pour certaines combinaisons de support et d’exigences, le support vient tempérer les effets des exigences du travail, alors que, pour d’autres combinaisons, ce même support vient amplifier les effets des exigences du travail. En d’autres termes, la même stratégie qui sert à gérer le stress (par exemple, l’appui des collègues de travail) peut présenter des effets bénéfiques en termes de tampon à certaines exigences du travail (par exemple, une surcharge); mais, elle peut avoir des effets négatifs qui vont en s’accentuant dans le cas de d’autres exigences (par exemple, les heures au bureau). Cette observation nous invite à penser qu’il est important de retenir la particularité à la fois des exigences du travail et du support social afin d’être en mesure d’identifier les relations spécifiques entre diverses combinaisons de variables. Cet essai se termine par une discussion de la manière dont les partenaires des systèmes de support social peuvent déplacer ou amplifier les effets du stress plutôt que de les adoucir.

Resumen

En este articulo, el modelo Demanda - Control de trabajo (DCT) es utilizado para predecir la depresión y el conflicto trabajo - familia de los abogados casados que trabajan a tiempo completo. Los objetivos de este articulo son : (1) determinar si el modelo DCT se aplica al conflicto trabajo – familia, (2) incorporar las variables especificas de este ámbito respecto a las demandas de trabajo y el control en el trabajo; y (3) examinar un espectro mas amplio de diferentes formas de apoyo social. Primero, el modelo DCT ayuda también a explicar el conflicto trabajo - familia. Segundo, la especificidad de ámbito no aparece como clave para documentarlos efectos amortiguadores de control en el trabajo. Tercero, el apoyo del cónyuge tiene el efecto mas fuerte sobre la depresión y el conflicto trabajo - familia, mientras que el apoyo de colegas funciona como un moderador de las demandas de trabajo de los abogados que tiene efectos amortiguadores y amplificadores. Este artículo termina con una discusión sobre las condiciones posibles bajo las cuales los miembros del sistema de apoyo pueden transferir o exacerbar los efectos del estrés en lugar de aliviarlos.

Research on the legal profession suggests that excessive work demands and long hours are stressful conditions of work that contribute to significant numbers of lawyers feeling dissatisfied with their jobs and leaving the profession (Brockman, 1992; Hagan and Kay, 1995). The practice of law generally demands total commitment and expects lawyers to put work first so that they are endlessly available to work long hours around the clock at the office and at home (Wallace, 1997, 1999; Epstein et al., 1999; Hagan and Kay, 1995; Seron and Ferris, 1995). Because the literature on the legal profession clearly illustrates that law is a “greedy institution” (Coser, 1974), this paper focuses on work-related time pressures and demands that may negatively affect lawyers’ well-being. Lawyers’ well-being is examined in this study in terms of depression, which has been recognized as the most common form of psychological distress that is experienced by everyone at some time to some degree (Pearlin and Johnson, 1977; Ross and Mirowsky, 1989).

In examining potentially stressful working conditions, it has been suggested that these conditions may have implications beyond an individual worker’s well-being and that their effects can spill over into one’s life at home (Westman, 2001) by contributing to work-to-family conflict. Work-to-family conflict occurs when work pressures are incompatible with family responsibilities. It is argued that excessive time pressures and demands may not only negatively affect lawyers’ general well-being, but may also affect their ability to balance work and family (Wallace, 1999, 2001; Brockman, 1992; Hagan and Kay, 1995). Despite recognition of the dynamic interaction between various life domains, most researchers studying job stress limit their attention to what takes place in the work domain and neglect how work experiences can affect other areas, such as the family domain (Cooper, Dewe and O’Driscoll, 2001; Thoits, 1995). Throughout the literature on work-to-family conflict, however, it has been well established that long work hours, working evenings and weekends and inflexible schedules exacerbate work-family tensions (MacDermid et al., 1994). The first objective of this paper then is to determine whether the models applied to depression in the stress literature also apply to work-to-family conflict.

In examining married lawyers’ depression and work-to-family conflict, hypotheses were derived from Karasek’s (1979) model of job demands and job control, as well as from the social support literature. Job demands are a commonly studied form of stress and a key determinant of depression and work-to-family conflict (Cooper, Dewe and O’Driscoll, 2001). Job control and social support are the most frequently studied coping resources typically examined by those researching the stress process (Thoits, 1995). Coping resources include social and personal characteristics that individuals may draw upon in response to stressful conditions (Pearlin and Schooler, 1978). In applying Karasek’s model to depression and work-to-family conflict, this paper also examines the extent to which job control and social support buffer the negative effects of excessive job demands experienced by practicing lawyers.

In doing so, the second objective of this paper is to incorporate domain-specific job control variables as potential coping resources. The literature, described in greater detail below, suggests that if both the job demand and the job control variables are located within the same specific domain, such as work time, it is more likely that having control over the time aspects of one’s job will buffer the effects of the time demands. Given that lawyers comprise the sample of this study, time-based demands and time-based control are the focus of this analysis.

The third objective of this study is to examine a wider array of various forms of social support in order to determine whether one type is more effective in buffering job demands than is another. Social support is generally viewed as a coping strategy that may reduce or alleviate the effects of stressors on workers’ well being either directly as a main effect or as a moderating “buffer” effect. Emotional support from coworkers and one’s spouse is examined, and these sources of support have been found to be important coping resources in the stress literature. In addition, this paper proposes two forms of social support that are not often examined in the literature, namely organizational support and spouse’s support of one’s career.

Karasek’s Job Demand-Control Model

The Job Demand-Control (JDC) model (Karasek, 1979) has been one of the most influential models of occupational stress for the last 25 years (see Van der Doef and Maes (1999) for a review). Also known as the strain model, the JDC model argues there are two key elements of the work environment that are important for understanding the job stress process: (1) the job demands placed on the worker; and (2) job control or how much discretion the worker has in attempting to meet these demands (Karasek, 1979). Job demands refer to workload stressors and have been typically examined in terms of time pressures, role conflict or role overload. Job control, also referred to as decision latitude, denotes a worker’s ability to control his or her work activities.

Karasek (1979) suggests that high job demands are not problematic in and of themselves, but when these demands are accompanied by low decision-making power, this negatively affects workers’ well being. The demands of the job may lead to negative psychological outcomes, such as depression, when the worker is constrained in how she or he may take action to reduce or cope with such potential stressors. A sense of control benefits individuals’ well-being because it encourages active problem solving and allows them to change their environment in order to cope with the demands of the job (Ross and Mirowsky, 1989). Thus control is argued to be one possible coping strategy that workers may use to improve their well-being.

Two basic hypotheses may be drawn from the JDC model. First, the strain (main effect) hypothesis predicts that workers in what Karasek (1979) labels “high strain” jobs will experience less well-being as compared to workers in “low strain” jobs. “High strain” jobs are characterized by excessive work demands and workload and little job control. This hypothesis suggests that depression will result from the additive, independent effects of demands and control. Second, the buffer (interaction effect) hypothesis predicts that control has an interactive effect, such that control can moderate the effects of job demands on well-being. That is, having control over one’s work can buffer the potentially negative effects of excessive job demands on worker health and well-being. Increased control reduces the effects of potential stressors because it provides workers the opportunity to adjust their work demands according to their needs, abilities and circumstances (Wall et al., 1996). The crucial issue for the second hypothesis then, is not whether having high demands and low control has a negative effect on worker well-being, but whether there is an interaction between the two.

There has been considerable empirical support for the strain hypothesis where working in a demanding job and having little control over one’s job are associated with lower job satisfaction, lower psychological well-being, more job-related distress and greater work-to-family conflict. The results in the literature are less consistent in regards to the buffering, or moderating, effects of job control on the relationship between demands and psychological well-being (Van der Doef and Maes, 1999). Some suggest these inconsistent findings are because the arguments have not been tested with the proper statistical techniques (i.e., they failed to test for statistical interactions) or that the measures are not valid indicators of the model’s constructs (Beehr et al., 2001; Fox, Dwyer and Ganster, 1993). Wall et al. (1996) argue convincingly and demonstrate empirically that the commonly documented failure to find empirical support for the buffer hypotheses likely results from the inadequate operationalization of control where the measures are poor reflections of the intended theoretical construct. It is also proposed that buffering effects may be expected when the specific type of control corresponds more closely to the specific demands of the job (Kushnir and Melamed, 1991). That is, control over one’s hours may be more effective in buffering the effects of working long hours than merely having general discretion in one’s job or life. As indicated above, one objective of this paper is to include job demand and job control variables that are domain-specific, which may provide a more valid test of the JDC hypotheses. That is, they are both work time-based because they relate to time pressures (demands) and control over one’s work time.

As mentioned above, Karasek’s original model is typically limited to the prediction of stress in the workplace and worker well-being. It is proposed here that his model should be applicable to work-to-family conflict because the central arguments and predictors are consistent with those in the literature examining how the work domain interferes with the family domain. More specifically, it seems reasonable to expect that the extent to which work conflicts with family will be positively related to one’s work demands and negatively related to the amount of control workers have over their work demands.

Several studies support the application of the JDC model to work-to-family conflict. For example, Voyandoff (1988) found that perceived control over one’s work schedule and job demands buffered some of the relationships between work role characteristics (e.g., workload pressure, working a non-day or weekend shift) and work-family conflict. Duxbury, Higgins and Lee (1994) made arguments similar to the ones proposed here and found empirical support for their application of Karasek’s model to work-to-family conflict. For example, they found that those with low perceived control over their life had significantly higher levels of interference from work to family. Neither study, however, examined the role that social support played in their test of Karasek’s model, which is discussed in greater detail next.

Social Support and the Buffering Hypothesis

Karasek’s (1979) Job Demand-Control model is often extended to include social support (Karasek, Triantis and Chaudhry, 1982; Searle, Bright and Bochner, 2001). Social support is an interpersonal coping resource where one person helps the other and enhances their well-being. It often involves having someone to talk to who is supportive and understanding of the individual (Ross and Mirowsky, 1989). Social support is generally viewed as an effective coping strategy that significantly reduces or alleviates the effects of stressors in models of worker well-being (House et al., 1985; Thoits, 1995) and the work-to-family interface (Carlson and Perrewé, 1999; Ducharme and Martin, 2000; Thomas and Ganster, 1995).

Two basic hypotheses are set out in the literature regarding the role of social support in the stress process. The strain (main effect) hypothesis suggests that support exerts a direct effect on outcomes regardless of the amount of job demands a person is experiencing. That is, social support simply improves one’s health and well-being (Haines, Hurlbert and Zimmer, 1991). The buffer (interaction effect) hypothesis suggests that social support moderates the effects of job demands on well-being (House, 1981). That is, individuals experiencing highly demanding work conditions in combination with high levels of social support will report higher levels of well-being, whereas individuals with low levels of support while working under highly demanding work conditions will report lower levels of well-being.

There are a number of different definitions and measures of social support found in the literature where the source and type of support are differentiated (House et al., 1985). The sources of support in studies of job stress tend to focus on supports available in the workplace, such as coworker and supervisor support (Searle, Bright and Bochner, 2001). The types of support that are most often examined in the literature include emotional support (also referred to as affective or esteem support) and instrumental support. Emotional support refers to providing individuals with feelings of being cared for and supported. It may involve the provision of affection, sympathy, and understanding. Instrumental support refers to providing material or concrete assistance in response to the individual’s specific needs, such as information, financial aid and help with one’s work. These two types of support are usually highly correlated with one another (House, 1981) and most studies tend to favour the inclusion of emotional support (LaRocco, House and French, 1980; Menaghan and Merves, 1984; Thoits, 1995) or collapse the items into a single scale measuring overall support (e.g., Frone, Russell and Cooper, 1991). As indicated earlier, the third objective of this study is to examine a wider array of different forms of social support in order to determine whether one type is more effective in buffering the negative effects of job demands than is another in the stress process. In doing so, four different social supports are examined: two are work-based and two are spouse-based. The work-based supports include emotional coworker support and organizational support. The support from one’s spouse includes emotional support and support for one’s career. The support variables are described in greater detail below.

Applying the Model to Lawyers

In this section, a job stress model is proposed that is specifically tailored to lawyers and their work experiences. In doing so, the hypotheses presented below refer to both dependent variables – depression and work-to-family conflict. As well, the job demands and the job control coping strategies centre specifically on the time demands associated with practicing law and are tested in terms of both main and buffer effects. Finally, in testing the JDC model, four different forms of social support are examined in terms of their main and buffer effects.

Job Demands

The demands of one’s job are usually recognized as a key source of stress for workers (Cooper, Dewe and O’Driscoll, 2001). Some authors distinguish between different types of job demands, such as time (pressure demands), problem-solving demands, and monitoring demands (e.g., Jackson et al., 1993; Beehr et al., 2001). In this study, time pressure demands are examined because they are most often identified as major sources of stress and dissatisfaction in the legal profession (Brockman, 1992). Four different variables are included that tap the time pressure job demands that lawyers often face: work overload, work hours at the office, work hours at home and participation in extra professional activities.

Work overload refers to the extent to which the demands of the job are felt to be excessive (Wallace, 1999). Feeling overwhelmed by the time demands and pressures of one’s job, in combination with working long hours, are often cited as illustrative of the all-encompassing nature of practicing law (Wallace, 1997; Kessler, 1997). Lawyers are renowned for the long hours that they work where throughout the literature on the legal profession there are reports of lawyers working from dawn to midnight, around the clock, for days. While most of their work hours are at the office, many lawyers regularly spend time working at home in the evenings and on the weekends. In addition, lawyers are often expected to work or attend professional functions after hours in the evenings, over lunch or on weekends. Kanter (1977) refers to such tasks as “diplomatic tasks” that involve after-hours dinners and weekend golf games with business partners and potential clients.

Hypothesis 1: Job demands (overload, hours at the office, hours at home, extra activities) will be positively related to depression and work-to-family conflict.

Job Control

As indicated above, the literature suggests that domain-specific coping strategies should be an effective coping strategy (Cooper, Dewe and O’Driscoll, 2001; Van der Doef and Maes, 1999). Accordingly, two different ways in which lawyers may exert control over the time demands of their jobs are examined in this study: control over the flexibility in their work hours and control over the number of hours that they work. Greater flexibility and discretion in one’s work hours may help lawyers structure their workdays so that they can better cope with competing role demands and reduce or buffer the impact of work demands that may contribute to depression and work-to-family conflict (Parasuraman et al., 1996; Voyandoff, 1988; Thomas and Ganster, 1995).

Hypothesis 2a: Job control (over flexibility and number of hours) will be negatively related to depression and work-to-family conflict.

Hypothesis 2b: Job control (over flexibility and number of hours) will moderate the relationships of job demands with depression and work-to-family conflict. More specifically, the positive relationship between job demands and depression/work-to-family conflict will be stronger for workers with less job control.

Social Support

As mentioned above, social support may be received from different sources (e.g., coworkers, family, friends). Four different forms of support are examined in this paper: two are work-based and two are spouse-based.

In terms of work-related social support, emotional support provided by one’s coworkers is examined in this study. This type of support is characterized by having coworkers who listen to and empathize with the demands of one’s job and who show concern and offer support and encouragement to the individual (Thomas and Ganster, 1995).[1] In addition, organizational support is a second type of work-related support which is proposed in this paper. Few researchers have examined the role of organizational factors in alleviating the negative effects of employee stress (Stamper and Johlke, 2003). Organizational support refers to the extent to which the work-family culture of the employing organization supports employees who take advantage of the family benefits they offer. A supportive organization values the integration of employees’ work and family lives (Thompson, Beauvais and Lyness, 1999) and does not penalize employees who use work-family benefits or devote time to family (Bailyn, 1997; Clark, 2001). This source of support is expected to be relevant to lawyers because it relates to the time norms and pressures associated with practicing law.

Research on work-family dynamics has also demonstrated the importance of support from one’s spouse and how this contributes to an individual’s well-being (Frone, Russell and Cooper, 1992; Parasuraman, Greenhaus and Granrose, 1992). An emotionally supportive spouse listens to and empathizes with the stresses of their partner’s job and offers support, encouragement and concern to their spouse (Thomas and Ganster, 1995). A second type of spouse support is spousal career support, which focuses on the extent to which the spouse respects and encourages the respondent’s career. It is argued to be important for individuals in highly demanding professional jobs to have a spouse who supports their career, both in terms of the stress it incurs and the rewards it offers. Career support has been found to be important in reducing depression and work-to-family conflict for samples of working women (e.g., Suchet and Barling, 1986; Beatty, 1996). This type of support has not been examined widely in the stress literature and will be examined for both men and women practicing law.

Hypothesis 3a: Social support (coworker, organizational, spouse (emotional), spouse (career)) will be negatively related to depression and work-to-family conflict.

Hypothesis 3b: Social support (coworker, organizational, spouse (emotional), spouse (career)) will moderate the relationships of job demands with depression and work-to-family conflict. More specifically, the negative relationship between job demands and depression/work-to-family conflict will be stronger for workers with less social support.

Data and Methods

Sample

All lawyers practicing in the Province of Alberta were mailed a survey to their place of work in June 2000. The mailing list was obtained from the Law Society of Alberta and contains the names of all active members. Questionnaires were sent to 5,921 lawyers and 1,827 were returned, yielding a response rate of 31%. Data provided by the Law Society of Alberta in terms of all active members’ gender and employment situation by city allows for comparisons between the provincial data and that of the sample (available from author). A comparison of the sample data with the provincial statistics obtained from the Law Society of Alberta indicates that similar proportions of lawyers participated when compared by gender, practice setting (e.g., law firm, government office, etc.) and city (e.g., Calgary, Edmonton, etc.).

The analyses in this paper are based on the survey responses of 1201 lawyers who were married and working full-time in the legal profession in Alberta in June 2000. The sample consists of 365 (30%) women and 836 (70%) men who completed the mail-out questionnaires. The lawyers worked in a variety of different work settings including law firms, corporate offices and government.[2] On average, the respondents were about 42 years of age, had practiced law for approximately 13½ years and earned about $128,298 (in Canadian dollars) in the previous tax year. On average, they worked about 48½ hours a week at the office. Most of the lawyers in this study (78%) had a spouse who works for pay. More than half (63%) had children living at home and, for 40% of the parents, at least one child was preschool-aged.

Measures

Many of the measures used in this study are multiple-item Likert scales. The values obtained for these variables represent the mean scores that were calculated by summing the items and dividing the sum by the number of items for the particular scale. Reliability coefficients (Cronbach’s alpha) are reported for the multiple item measures.

Depression and Work-to Family Conflict

Depression is measured by seven items from Ross and Mirowsky’s (1989) modified form of the Centre for Epidemiological Studies of Depression (CES-D) scale (Radloff, 1977). This scale measures symptoms of depression in the general population and does not indicate a diagnosis of clinical depression. Respondents were asked to report how often they had experienced the following in the previous week: they had trouble getting or staying asleep, they felt they just couldn’t get going, they felt they couldn’t shake the blues, they had trouble keeping their mind on what they were doing, they felt that everything was an effort, they felt sad and they felt lonely. The response categories range from never (coded 1) to most of the time (coded 4) (α = .85). Work-to-Family Conflict is measured by Netemeyer, Boles and McMurrian’s (1996) five-item Work-Family Conflict Scale that reflects the extent to which work demands interfere with home and family life, family activities have to be changed to accommodate work, it is difficult to fulfill family responsibilities because of work demands, strain from work makes it difficult to fulfill family duties and the demands of the job make it difficult to do things they want to do at home. The response categories range from strongly disagree (coded 1) to strongly agree (coded 5) (α = .90).

Job Demand Variables

Four variables were used to assess the time demands associated with practicing law. Work Overload is measured by four Likert items selected from Caplan, Cobb and French’s (1975) five-item Workload scale that assesses whether respondents have enough time to get everything done in their job, their workload is too heavy, they have to work very quickly to get things done in their job and they often feel rushed in their job.[3] The response categories range from strongly disagree (coded 1) to strongly agree (coded 5) (α = .78). Work Hours at Office is the average number of hours that respondents reported working at the office per week, including evenings and weekends. Work Hours at Home is the average number of hours respondents worked per week at home, including evenings and weekends. Extra Professional Activities is the number of times a month respondents attend social, work-related professional activities (e.g., related to business or client development, conferences, meetings or receptions) outside of regular working hours (e.g., before 8:00 am, over lunch, after 6:00 pm during the week, or on weekends (day or night)). It is coded 1 if less than once a week; 2 is 1 to 2 times a week; and 3 is more than twice a week.

Job Control Variables

Two measures of control over one’s time demands are included in this study. Control over Flexibility is measured by a single Likert item from Holtzman and Glass (1999) that refers to the ease involved in taking time off to take care of personal or family matters. Control over Work Hours is measured by a single Likert item that indicates the amount of control respondents have over the number of hours they work. Both items have response categories that range from strongly disagree (coded 1) to strongly agree (coded 5).

Social Support Variables

Social support is measured by four variables: coworker support, organizational support, spouse (emotional) support and spouse (career) support.[4] Coworker support is measured by four Likert items that Thomas and Ganster (1995) used to assess emotional supervisor support where reference to one’s “supervisor” was changed to refer to one’s “coworkers”. These items measure the extent to which other lawyers listen to their work-related problems, empathize with the stress of the job, offer support and encouragement and show concern. The response categories range from never (coded 1) to most of the time (coded 4) (α = .89). Organizational support is measured by three Likert items selected from Thompson, Beauvais and Lyness’ (1999) five-item Career Consequences subscale of their Work-Family Culture Supportiveness scale.[5] These items measure the extent to which turning down work for family-related reasons will not seriously hurt one’s career, other lawyers in the employing are not resentful when people take extended leaves to care for new or adopted children and lawyers who participate in work-to-family programs are not viewed as less serious about their careers. The response categories range from strongly disagree (coded 1) to strongly agree (coded 5) (α = .75). Spouse (emotional) support is measured by four Likert items that Thomas and Ganster (1995) used to assess emotional supervisor support. The reference to “supervisor” was changed to “partner”. The items assess the extent to which one’s spouse listens to their work-related problems, empathizes with the stress of the job, offers support and encouragement and shows concern. The response categories range from never (coded 1) to most of the time (coded 4) (α = .94). Spouse (career) support is measured by five Likert items from Beatty’s (1996) eight-item scale that taps the degree to which one’s spouse considers their partner’s career as important as their own, is supportive of their partner’s career, is comfortable with the money they earn, is not resentful of their career and does not complain about the amount of time devoted to work.[6] The response categories range from strongly disagree (coded 1) to strongly agree (coded 5) (α = .75).

Family Context Variables

Four variables take into account the respondent’s family situation. These are the amount of time the respondent spends doing housework during the week, the number of children and presence of preschool-age children currently living with the respondent and the employment status of the respondent’s partner. Although these variables do not reflect the actual responsibilities and duties associated with, or performed by, those in various family roles, inclusion of these objective status variables allows for controlling for different domestic situations that married lawyers may be involved in. These variables were also included as control variables in order to demonstrate that the relationships found between the work-related determinants and the dependent variables were not spurious results obtained through the omission of key variables from the analysis.

Housework Hours is measured by self-reports of how many hours a day on average, in a typical week, respondents spend on home chores, such as cooking, cleaning, repairs, shopping, yard work or banking on days that they work and on days that they do not work. Since all lawyers included in this analysis are working full-time, the hours spent doing housework on days that they work was multiplied by five and the hours they spend on days that they do not work was multiplied by two and both were added together to compute a weekly average of time spent on housework. Studies have reported that time spent doing housework is associated with greater depression and greater work-to-family conflict (Bird, 1999; Coltrane, 2000; Glass and Fujimoto, 1994). It is argued that greater involvement in housework contributes to greater role conflict and increased stress and anxiety due to the sheer time and energy demands associated with multiple role demands (Rosenfeld, 1989). Number of Children is the number of children respondents report are currently living at home with them. Preschool Children indicates whether (coded 1) or not (coded 0) respondents have any children under the age of six currently living at home with them. A greater number of children living at home and the presence of children under the age of six generally require more time and energy from working parents (Aryee, 1992; Bedeian, Burke and Moffett, 1988; Voyandoff, 1988). The effects of children were taken into account because it was expected that they are positively related to depression and work-to-family conflict. Partner Works indicates whether (coded 1) or not (coded 0) the respondent’s spouse works for pay. Individuals who are part of a dual-career couple are simultaneously trying to balance the demands of two careers and often place a premium on time, which usually becomes their scarcest and most valued resource (Harriman, 1982; Parasuraman, Greenhaus and Granrose, 1992; Voyandoff, 1988). Consequently, this variable is expected to be positively associated with depression and work-to-family conflict.

Control Variables

Three control variables are included in the analysis. These variables were included as control variables in order to demonstrate that the relationships found between job demands, job control and social support and depression and work-to-family conflict were not spurious results obtained through the omission of key variables from the analysis. Adequate Income is measured by Kelly and Voyandoff’s (1985) single item measure of Income Inadequacy, which was reworded from a negative statement to a positive one: I feel I have enough money for all of my needs. The response categories are strongly disagree (coded 1) to strongly agree (coded 5). Economic hardship, or inadequate income has been found to be associated with greater depression (Bird, 1997; Mirowsky and Ross, 2001; Ross and Mirowsky, 1989) and work-family conflict (Voydanoff, 1990). Law Experience was computed by subtracting the year the respondent was called to the Bar to practice law from the date of the survey. The early years of a professional career, such as law, can be quite demanding and stressful resulting in greater depression and work-to-family conflict (Wallace, 1997). It is expected that law experience will be negatively related to depression and work-to-family conflict. Sex (Male) was coded 1 for men and 0 for women. Research shows that women tend to report greater levels of depression (Bird, 1997; Ross and Mirowsky, 1989) and work-to-family conflict (Tausig and Fenwick, 2001) than do men.

Analytic Strategy

First, the strain (main) effect Hypotheses 1, 2a and 3a are tested by examining the independent effects of job demands, job control and support on depression and work-to-family conflict. Next, the two buffer (interaction) hypotheses (Hypotheses 2b and 3b) are tested by determining whether the cross-product interaction terms between job demands and job control and the cross-product interaction terms between job demands and social support have statistically significant effects on depression and work-to-family conflict. To do this, a series of intermediate models were estimated to derive the final model for each dependent variable that is presented in Table 2 (available from author). Six separate models were estimated for each possible set of cross-product interactions between the job demand variables and the two potential control buffers and the four potential support buffers. All statistically significant interaction terms were then included in the final models presented in Models 3 of Tables 2 and 3. Empirical support for the buffer hypotheses is indicated by statistically significant negative interaction coefficients. A statistically significant positive interaction coefficient suggests a “reverse buffering” or amplifying effect.

Table 1 contains the means, standard deviations and zero-order correlations for all variables included in the analysis. Examination of the zero-order correlations shows that none suggest collinearity problems. In addition, following Fox (1991), variance-inflation factors (VIF) were estimated for all of the variables included in the analysis. These results (available from author) also suggest that multicollinearity among the predictors is not evident for any of the variables.

Table 1

Means, Standard Deviations and Zero-Order Correlations

Note: Zero-order correlations that are greater that +/– .06 are statistically significant at the .05 level.

Results

Table 2 shows the regression results for depression and Table 3 shows the regression results for work-to-family conflict. In both tables: Model 1 contains the regression results for the control and family variables; Model 2 contains the main effect regression results for testing Hypotheses 1, 2a and 3a; and Model 3 contains the significant interactions for testing Hypotheses 2b and 3b.

Job Demands

Model 2 shows that there is support for Hypothesis 1: greater job demands result in greater depression (Table 2) and greater work-to-family conflict (Table 3). Overload in particular appears to be an important determinant of both dependent variables. As shown in Model 2 and discussed below, however, the effects of all four of the job demand variables interact with some form of social support.

Job Control

Model 2 shows that the both forms of job control are significant deterrents of depression (Table 2) and work-to-family conflict (Table 3) as predicted by the strain hypothesis (Hypothesis 2a). The buffer hypothesis (Hypothesis 2b) was not supported. None of the eight interactions of the job demand-by-job control cross-products was statistically significant at the .05 level (one-tailed test). This means that having control over work-time flexibility and control over the number of hours worked does not buffer the effects of job demands on either depression or work-to-family conflict. Instead, control functions to reduce depression and work-to-family conflict independently of the amount of job pressures lawyers’ experience.

Social Support

One objective of the study was to examine whether various forms of social support are differentially related to depression and work-to-family conflict. The results suggest that the four different forms of social support clearly have different relationships with depression and work-to-family conflict. First, the findings in Model 2 show that the extent to which one’s spouse supports the individual’s legal career is the most important form of support in directly reducing depression (Table 2) and work-to-family conflict (Table 3). As indicated earlier, this form of support has not been widely studied in the stress literature and appears to be an important source of support in this study. Second, the findings for the emotional support from one’s spouse are mixed. In Model 2, emotional spousal support reduces depression (Table 2) whereas the results show that emotional spousal support contributes to work-to-family conflict (Table 3). It appears that receiving support and understanding from one’s spouse exacerbates the extent to which an individual experiences conflict between their work and family domains. Third, organizational support has significant main effects as predicted in that it reduces depression and work-to-family conflict. Fourth, Model 2 shows that initially, contrary to Hypothesis 3a, it appears that coworker support is not an effective coping response for alleviating depression (Table 2) or work-to-family conflict (Table 3). It fails to significantly reduce either dependent variable. It is important, however, to turn next to the interaction results presented in Model 3 for depression and work-to-family conflict.

Table 2

Regression Results for Depression (N = 1109)

* p < .05; ** p < .01; *** p < .001 (one–tailed test)

Table 3

Regression Results for Work-Family Conflict (N = 1109)

* p < .05; ** p < .01; *** p < .001 (one–tailed test)

From the results presented in Model 3, it appears that it is critical to examine the moderating effects of coworker support, since five of the nine significant interactions in the two tables involve this form of support. More specifically, the results show that coworker support effectively buffers the negative effects of overload on both depression and work-to-family conflict as well as the negative effects of extra professional work activities on depression. These results clearly support buffer hypothesis (Hypothesis 3b). Contrary to the buffer hypothesis, however, coworker support also appears to accentuate the negative effects of the hours lawyers’ work at the office on both dependent variables. That is, lawyers who work long hours and have higher levels of coworker support also report higher levels of depression and work-to-family conflict than do those with low levels of coworker support.

Turning next to the other significant interactions, we see that the emotional support provided by spouses is also important. Lawyers with spouses who listen, empathize and show support and concern are less likely to find that long hours at the office significantly increase depression (Table 2), consistent with Hypothesis 3b. Also, lawyers with spouses who support their legal careers are also buffered from the negative effects of overload on depression (Table 2).

The last two interactions shown in Model 3 of Table 3 are contrary to what was predicted by the buffer hypothesis because they indicate a mediating amplifying effect. Lawyers with spouses who provide high levels of emotional support find that the more hours they spend working at home, the more they feel their work is conflicting with their family life. As well, lawyers who are involved in many extra professional activities outside of regular work hours and who report that their organization is supportive of work-to-family balance, experience greater work-to-family conflict rather than less. These two findings are consistent with the interaction effects discussed above showing that emotionally supportive coworkers and working long hours at the office increases, rather than reduces, depression and work-to-family conflict.

Family Context and Controls

The results in Model 2 of Tables 2 and 3 show that the number of hours of housework performed by respondents per week does not significantly affect depression or work-to-family conflict. And while preschool children alleviate depression, the more children lawyers have at home, the more work-to-family conflict they report. An employed partner has no effect on lawyers’ depression but significantly reduces work-to-family conflict.

Adequate income significantly reduces depression for lawyers, whereas greater law experience is associated with less work-to-family conflict. Lastly, the results in Model 2 show that women report greater depression and work-to-family conflict than do men, consistent with the literature.

Discussion

This paper set out to examine whether hypotheses derived from Karasek’s (1979) model of job demands and job control and from the social support literature hold regarding the effects of these variables on married lawyers’ depression and work-family conflict. In doing so, one objective of this paper was to determine whether the models applied to depression in the stress literature also apply to work-family conflict. The results of this study suggest that they do. Three of the four job demand variables contribute significantly to work-family conflict, control over work time reduces work-family conflict, and social support functions to moderate the effects of the job demand variables. In addition, the model tested in this paper explained more than half of the variance in work-family conflict. These results suggest that researchers attempting to better understand work-family conflict may consider the long-established theoretical and empirical literature on the stress process, such as Karasek’s JDC model.

The second objective of this paper was to incorporate domain-specific job demand and job control variables. Given that lawyers comprise the sample of this study, time-based demands and time-based control were the focus of this analysis. It was argued that if the job demands and the job control variables were located within the same specific domain, that is being time-based, it would be more likely that having control over the time aspects of one’s job would buffer the effects of the time demands associated with practicing law. Empirical support was not found for this aspect of Karasek’s (1979) JDC model. It appears that domain-specificity is not necessarily key to documenting the buffering effects that JDC model proposes, contrary to Van der Doef and Maes (1999) conclusions. This does not mean that job control is not important to understanding the stress process, however. The results of this study show that control over one’s hours contributes directly to lawyers’ well-being rather than acting as a buffer.

The third objective of this study was to examine a wider array of different forms of social support in order to determine whether one type is a more effective buffer than another in the stress process. The results suggest several interesting patterns that highlight the importance of differentiating between various forms of support. One notable finding is that the extent to which one’s spouse is supportive of the individual’s career is the most important form of support directly reducing depression and work-to-family conflict. This form of support has not been examined widely in the literature and should be considered in future studies. As well, the emotional support and understanding one’s spouse provides reduces depression, but increases work-to-family conflict, which was unexpected. The amplifying effects of support are discussed in greater detail below. An additional important finding is that coworker support initially appears ineffective because it fails to have significant effects on either depression or work-to-family conflict. Only by conducting interaction tests is it possible to determine the significant buffering effect coworker support offers in reducing the negative impact of overload on lawyers’ depression and work-to-family conflict. This shows that it is imperative to model and test the functional form of the relationships as theorized in the literature. Several authors (e.g., Cooper, Dewe and O’Driscoll, 2001; Fox, Dwyer and Ganster, 1993; Van der Doef and Maes, 1999) have criticized researchers for failing to model the hypotheses properly by not incorporating, for example, interaction tests in their models when the theoretical framework suggests moderating relationships. Obviously, inaccurate conclusions would be drawn about the effectiveness of supportive coworkers if the potential buffering effects of this coping strategy were not tested.

A particularly important finding of this study involves the moderating effects of social support. The results indicate that for certain demand-support combinations, support buffers the effects of work demands, whereas for other combinations, it amplifies the effects of work demands. That is, the same coping strategy (e.g., coworker support) can have beneficial buffering effects for some job demands (e.g., overload) but negative amplifying effects for other jobs demands (e.g., hours at the office). This finding suggests that it is important to retain the specificity of both job demands and social support in order to be able to observe the specific relationships among different combinations of these variables.

In this study, four different measures of the time demands associated with practicing law were used to essentially measure one type of job stressor, namely job demands associated with time pressures. In contrast, some studies include a global measure of stressors, for example, by combining a series of role stressors and/or a global measure of support by combining support from different sources and/or different types (Ducharme and Martin, 2000; Van der Doef and Maes, 1999). It is possible that the use of global measures may mask the impact of more specific stressors and supports, and yield no evidence of interaction effects if positive and negative interactions are essentially cancelling one another out. Clearly, future studies should take a more fine-grained approach to operationalizing the key concepts of the stress process.

The “reverse buffering” or amplifying effects of coworker support, organizational support and spouse (emotion) support presented in Tables 2 and 3 also deserve further discussion. The results show that not all support strategies are beneficial and instead, some may be detrimental to worker well-being. That is, support from others may reaffirm the aversive nature of the work environment and the individual’s negative responses may be heightened rather than reduced (Kaufman and Beehr, 1986). Talking about one’s problems may involve dwelling on the problem, revisiting and rehearsing failures and injustices or other depressing events where the supportive others may help to see that conditions are as bad as, or even worse than, the individual thought (Carlson and Perrewé, 1999; LaRocco, House and French, 1980; Ross and Mirowsky, 1989). Karasek, Triantis and Chaudhry (1982) refer to this as the stress-transfer effect when social support has a significant interaction that exacerbates rather than reduces the effects of stressors. They suggest that the effectiveness of buffering depends on the amount of stress being experienced by those surrounding the individual. They also suggest that stress-transfer is more prevalent when individuals have emotionally supportive relationships from others. If members of one’s support system are also feeling high levels of stress, and especially if it is due to the same factors, individuals seeking support may absorb more feelings of stress from those around them rather than be protected from stress. As Karasek, Triantis and Chaudhry state: “In times of overall group crisis, little “solace” will be found from others who are exposed to the same problem” (1982: 195). Applying their argument to the present study, the interpretation might be as follows. A lawyer shares his or her experiences about the excessive time demands of the job with coworkers who are also working long hours that negatively affect their well-being. The similarly highly stressed coworkers provide high levels of understanding, concern and empathy to each other and may legitimize negative feelings about the workplace. As a result, they are likely to transfer feelings of strain to each other, thereby amplifying the negative effects of their job demands on their well-being rather than reducing them. Similarly, in regards to spousal support, if both spouses share the belief that the time demands of the lawyer’s job interferes with their family life, the spouse is more likely to reinforce and exacerbate the effects of stressors because they are both feeling the negative consequences of the situation. An important direction for future research is to examine the specific conditions under which stress transfer is likely to occur. One factor to consider seems to be the extent to which members of an individual’s support system are also exposed to high levels of stressors.

Along related lines, in their study of stress transmission between spouses, Rook, Dooley and Catalano (1991: 165) suggest, particularly amongst close associates or spouses, that the pain felt by one individual is also felt by the other as their own. They propose, for example, that in closer marriages, spouses are more likely to suffer the “greatest reverberations” from events that affect their partner, whereas relationships characterized by marital discord may be less dramatically affected by their partner’s hardships. A similar argument might be made in regards to the closeness of coworker relationships as well, which is consistent with Karasak, Triantis and Chaudhry’s (1982) comments about stress-transfer occurring in emotionally close relationships. Future research may also explore the closeness of the relationships individuals have with members of their support system in order to better understand the stress-transfer effect. By examining the levels of stressors amongst members of an individual’s support system, and the closeness of the individual to members of their support system, we might gain a clearer understanding of the conditions under which social support has a positive or negative moderating effect on an individual’s well-being.

Lastly, it is important to note that the results of this study suggest that the two different coping strategies, job control and social support, function differently in the job stress process. Control over the hours that lawyers work is important in reducing depression and work-to-family conflict, but job control does not reduce the negative effects of work-related time demands and pressures as predicted by the buffer hypothesis. In contrast, social support, specifically coworker support as an example, acts as a significant buffer that reduces the negative effects of certain job demands on lawyers’ depression and work-to-family conflict. Whereas the job control variables only have an additive effect, several of the social support variables buffer the effects of job demands on depression and work-to-family conflict. These findings again reiterate the idea that specificity is key to understanding the relationships among various demands and coping strategies. Coping strategies clearly need to be differentiated between those where an individual exerts control over potential stressors in their environment, versus those strategies involving seeking support from others.

In closing, several limitations of this study must be noted. The cross-sectional data presented in this study are based on subjective self-reports from the perspective of individual lawyers. Such data are susceptible to distortions results from respondents answering so as to maintain a consistent series of answers or to present themselves in a favourable light or from other effects of common method variance (Podaskoff and Organ, 1986). Future research may consider using more objective measures of the job demand variables assessed, as well as coworker and spouse reports about the support provided, in attempting to reduce such threats. Along related lines, both job control variables were measured by single indicators. More valid measures might be used that capture a wider range of content of these important variables. In addition, future research should involve longitudinal analyses that can more clearly establish the causal ordering among job demands, control and social support and the subsequent outcome variables. It is also important to note that the analysis was limited to a single high-status, professional occupation, namely lawyers. Some of the findings here may be limited to this particular occupation under study or professionals in general. It is important to determine whether the results are generalizable to more occupationally diverse groups of workers.

Notes

• [1] Supervisor support is the other popular form of work-based support. It is not examined in this study because it is not relevant to this particular sample. Lawyers, much like academics, do not usually have someone who they can identify as their supervisor.
• [2] Note that women and men working in solo practice were excluded from this analysis because the variables tapping coworker and organizational support are not applicable.
• [3] The fifth item that taps whether respondents feel there has been a large increase in their workload was not included in the survey instrument.
• [4] Confirmatory factor analyses were conducted for the social support measures using principal components analysis (results available from author). The 16 items loaded highly on four factors with low cross-factor loadings (e.g., .001 to .252). Spouse support (emotion) loaded on the first factor with loadings ranging from .834 to .908. Coworker support (emotion) loaded on the second factor with loadings ranging from .843 to .883. Spouse support (career) loaded on the third factor with loadings ranging from .474 to .740 and organization support loaded on the fourth factor with loadings ranging from .756 to .836.
• [5] One item was not included in the survey instrument because it had a relatively low factor loading of .49 in Thompson, Beauvais and Lyness’ (1999) original principal components analysis. In addition, they included two items that referred to employees being resentful to “men” and another referring to “women” for taking extended leaves. These two items were combined into a single item that referred to employees being resentful to “people” and thereby eliminated the reference to their coworker’s sex.
• [6] Three items from Beatty’s (1996) original scale were excluded from the measurement instrument used in this study because of their similarity and conceptual overlap with other measures included in the survey (e.g., emotional spousal support and family-to-work conflict).