The Role of Resources

Stressors do not necessarily have a negative effect on the in­dividual. The degree to which a stressful work situation af­fects the individual might be contingent on the availability of resources. Hobfoll (1998) defines resources as "objects, con-"ions, personal characteristics, and energies that are either hemselves valued for survival, directly or indirectly, or that ^ser^e as a means of achieving these ends" (p. 54). With re-•P^ect to organizational stress, resources refer to conditions ^lt1m the work situation and to individual characteristics "M can be used to attain goals. Both with respect to the ad-cement of stress theory and practical implications, it is

Empirical Evidence 467

highly relevant to establish whether these resources buffer (i.e., moderate) the effects of stressors on strains.

Resources at work most often studied were control at work and social support. Individual resources are coping styles, locus of control, self-efficacy, and competence. Addi­tionally, we shall briefly refer to other factors such as Type A behavior pattern, hardiness, and sense of coherence.

Control at Work

Control at work refers to an individual's opportunity to influ­ence one's activities in relation to a higher-order goal (Frese, 1989). P. R. Jackson, Wall, Martin, and Davids (1993) differ­entiated between control over timing and methods to do the work. Many studies addressed the question of whether high control at work buffers the negative effects of a stressful work situation on an individual's health and well-being. Most of these studies have been conducted within the framework of Karasek's (1979) job demand-job control model.

Epidemiological studies on cardiovascular diseases an as outcome variable tended to confirm the major assumptions of Karasek's model (for reviews, cf. Kristensen, 1995; Schnall et al., 1994; Theorell & Karasek, 1996). Individuals in high-strain jobs often suffered from cardiovascular illnesses. Moreover, in about half of the studies, high-strain jobs were associated with cardiovascular risk factors such as high blood pressure and smoking (Schnall et al, 1994).

With respect to other outcomes including psychological well-being and mental health, the findings are less conclu­sive. Several reasons for these inconsistent findings can be mentioned. First, there are many studies that did not explic­itly test the interaction effect but that compared high demands-low control subgroups (i.e., high-strain jobs) with high demands-high control subgroups (i.e., active jobs). This comparison often revealed significant differences in health and well-being between high-strain jobs and active jobs (e.g., Eriksen & Ursin, 1999; Landsbergis, 1988). Theorell and Karasek (1996) have recently suggested that this proce­dure be used in general (for a critique, cf. Kasl, 1996).

In a qualitative review of empirical studies on the job demand-job control model published between 1979 and 1997, Van der Doef and Maes (1999) examined whether individuals in high-strain jobs experience poorer psychological well-being than do individuals in other jobs. Their review revealed that in 28 of the 41 studies with general psychological well-being as dependent variable, individuals in high-strain jobs indeed showed the lowest well-being scores. For job-related well-being such as job satisfaction, burnout, and job-related mood as dependent variables, a similar picture emerged. Strictly speaking, such a comparison between high-strain jobs

468 Stress in Organizations

and other jobs examines the main effects of job demands and job control—not the hypothesized interaction effect. When testing the interaction effect with the more appropriate mod­erated regression analysis, the job demand-job control model was supported less frequently. Some researchers reported support for the model (Fox, Dwyer, & Ganster, 1993; Sargent & Terry, 1998), whereas others did not (Landsbergis, 1988; Schaubroeck& Fink, 1998).

In the aforementioned review by Van der Doef and Maes (1999), 8 of 31 studies showed (partial) evidence for the inter­action effect. An additional seven studies confirmed the inter­action effect for subgroups of individuals, dependent on their personality, type of organization, and hierarchical position. A more recent study found support for the postulated interaction effect when using a multilevel analysis approach (VanYperen & Snijders, 2000). It is noteworthy that significant interaction effects were also found in longitudinal studies (Parkes et al., 1994; Sargent & Terry, 1998).

A second reason for failing to find the postulated interac­tion effect between demands and control may lie in the oper­ational! zation of the core variables. For example, Wall et al. (1996) argued that Karasek's (1979) measure of decision latitude (used in many studies) is a conglomerate of many as­pects of control such as decision over working methods, decision over scheduling of one's tasks, aspects of skill use, and task variety. Probably only proper job control attenuates the negative effects of high demands, whereas skill use and task variety do not. Wall et al. (1996) tested this assumption explicitly and found the hypothesized interaction effect for a relatively narrow job control measure but not for the broader decision latitude measure (for similar findings, cf. De Croon, Van der Beek, Blonk, & Frings-Dresen, 2000; Sargent & Terry, 1998).

A third reason for the inconsistent findings on the job demand-job control model lies in the effects of additional vari­ables such as social support or self-efficacy. For example, Johnson and Hall (1988) incorporated social support into the model. This extended demand-control-support model showed social support to buffer the negative effects of the combination of high demands and low control. Stated differently, the detri­mental effects of a high-strain job unfolded only when social support was low but not when social support was high. Thus, a three-way interaction was found.

Van der Doef and Maes (1999) suggested that field studies that tested the hypothesized three-way interaction—and that controlled for main effects and two-way interactions— resulted in inconclusive findings. For example, Parkes et al. (1994) reported support for the demand-control-support model. Most studies found no evidence for a three-way inter­action between demands, control, and support (Dollard et al, 2000;Furdaetal., 1994;Melamed,Kushnir,&Meir, 1991; tor

a summary, cf. Van der Doef & Maes, 1999). Some author* even reported findings that cast doubt on the predictions of rh demands-control-support model (Landsbergis, Schnall, Deit? Friedman, & Pckering, 1992; Schaubroeck & Fink, 199g\ Recent research suggests even more complex interactions and stresses the importance of coping (Daniels, 1999).

Fourth, Warr (1987) and Frese (1989) have argued that at work it should be very difficult to find interaction effects of stressors and control: Control implies that people can do some­thing about the stressors. If people are bothered by stressors they reduce the stressors; but they can only reduce stressors if they have control. If stressors continue to exist, it may be be­cause they are noncontrollable by definition. Because non-controllability and stressors are intertwined, it is difficult to show an interaction effect. It should be much easier to find an interaction effect if people are confronted with a new situation, such as in an experiment.

Fifth, experimental research tends to support the job demand-job control model. In such experiments, interaction effects of perceived demands and perceived control on dependent measures such as anxiety, task satisfaction, and subjective task performance were found (Jimmieson & Terry, 1997; Perrewe & Ganster, 1989), although there is also dis-confirming evidence (Perrewe & Ganster, 1989; Searle et al., 1999). There is a large body of literature on the learned help­lessness paradigm (Seligman, 1975), which also posits an interaction effect of stressors and control. Experimental re­search in this tradition has repeatedly replicated the interac­tion effects of bad events and noncontrol on reduction in well-being (Peterson, Maier, & Seligman, 1993).

In summary, there is strong empirical evidence for the ad­ditive main effect of job demands and job control. Individu­als in high-strain jobs show the lowest well-being scores and suffer most from illnesses. However, the interaction effect has received far less support. Adequate operationalization of job control may be crucial for finding significant interaction effects. Experimental findings tended to support the helpless­ness concept with its interaction effects of stressors and non-control. In all, Karasek's (1979) model has contributed to a fair amount of empirical controversy that has been fruitful. Given the previous arguments and the experimental findings, the fact that noncontrol and stressors produce at least additive effects and that a number of field studies find an interaction effect after all, we tend to think that Karasek's model has not done that badly.

Social Support and Work Group Factors

Social support is important for protecting an individual's health and well-being. It can be characterized as resources provided by others (Cohen & Syme, 1985) and comprises emotional.

informational, and instrumental (i.e., tangible) support (House, 1981). In general, the literature assumes that the beneficial effect of social support works both via main and interaction effects. Arecent meta-analysis based on a total of 68 effect sizes addressed the main effect and has shown that social support is negatively associated with strains (Viswesvaran, Sanchez, & Fisher, 1999). We find it interesting that social support was also negatively related to stressors at work.

With respect to the interaction effect. Cohen and Wills (1985) pointed out that social support functions only as a buffer in the stressor-strain relationship if the available sup­port matches "the specific need elicited by a stressful event" (p. 314). A number of cross-sectional studies suggest that social support buffers the negative effects of stressors (for a review, cf. Kahn & Byosiere, 1992).

Longitudinal studies are needed to arrive at a conclusion ^ about causality. Dormann and Zapf (1999) reviewed 10 lon-Pgitudinal studies published between 1985 and 1999 that examined the interaction effect of social support. Three of these studies found no moderator effects. In some of the other studies, moderator effects missed the conventional signifi­cance level or were only significant for a small part of all the effects tested. Thus, the evidence for an across-the-board moderator effect of social support is not very strong. A closer look at some of the recently published studies suggests that there might be specific mechanisms underlying the stress-buffering potential of social support. For example, in corre­spondence to the stress matching hypothesis (Cohen & Wills, 1985), Frese (1999) found the strongest effects for social stressors and socially related aspects of psychological dys-fimctioning. Dormann and Zapf (1999) found a lagged mod­erator effect of social support only with an 8-month time lag, but neither for shorter nor for longer time lags. More research is needed that examines in more detail how the effects of so­cial support unfold over time.

Moreover, there is increasing evidence that social support does not have unequivocal positive effects. A number of au­thors reported that a high degree of social support or related variables increased the relationship between stressors and strain symptoms (Schaubroeck & Fink, 1998). Peeters, Buunk, ^d Schaufeli (1995) showed that a high level of instrumental ^s°cial support may induce feelings of inferiority that are detri­mental to an individual's well-being.

In addition to social support, group work factors such as

e^roup cohesion or team climate play a role when it comes to

■ ^ress in organizations. First, research suggests that individu-

^s who work in teams experience better well-being than do

'viduals working in no team or a pseudoteam (Carter &

st, 1999) Second, group cohesion and favorable team

^ates were found to be associated with team members'

*eU-being (Carter & West, 1998; Sonnentag, Brodbeck,

Empirical Evidence 469

Heinbokel, & Stolte, 1994; for an overview, cf. Sonnentag, 1996). Third, work group factors such as psychologi­cal safety (Edmondson, 1999) or collective efficacy (Schaubroeck, Lam, & Xie, 2000) might buffer the negative effects of stressors. However, empirical studies are still rare (for a related recent study, cf. Bliese & Britt, 2001). Forth, there is increasing evidence that emotional contagion occurs in work groups (Bakker & Schaufeli, 2000; Totterdell, Kellett, Techmann, & Briner, 1998). Emotional contagion refers to processes by which an individual's mood is trans­mitted to other persons—for example, other team members. On the one hand, this phenomenon implies that a stressful events can influence more persons than simply those directly faced with the stressor. On the other hand, other team mem­bers' positive moods can serve as a resource when another member is confronted with a stressful situation. Linking group work factors to stress issues seems to be a fruitful avenue for future research.

Coping Styles

A favorable coping style can be a core resource for bolstering an individual's health and well-being. Lazarus and Folkman (3994) defined coping as "constantly changing cognitive and behavioral efforts to manage specific external and/or internal demands that are appraised as taxing or exceeding the re­sources of the person" (p. 141). They differentiated between problem-focused and emotion-focused forms of coping. Problem-focused coping includes problem-solving behaviors that aim directly to change the stressor, other aspects of the en­vironment, or one's own behavior. Emotion-focused coping refers to attempts to manage cognitions or emotions directly (for a critique and extension, cf. Semmer, 1996).

Problem-focused coping has been found to be positively related to mental health and well-being, whereas emotion-focused coping and an additional style of avoidance coping were often found to be associated with poorer well-being (Guppy & Weatherston, 1997; Hart, Wearing, & Headey, 1995; Leiter, 1991; Sears, Urizar, & Evans, 2000).

With respect to moderator effects, empirical findings are less conclusive. Many studies did not find the hypothesized moderator effects of coping on the relationship between stressors and strains (e.g., Ingledew, Hardy, & Cooper, 1997). Most studies that found a moderator effect of coping identi­fied problem-solving coping as a favorable coping style, whereas emotion-focused coping turned out to be an unfa­vorable coping style (Parkes, 1990). This implies that indi­viduals who approach the stressors directly or engage in other problem-solving behaviors are better off than individuals who concentrate on the management of their emotions and cognitions.

470Stress in Organizations

Authors like Perrez and Reicherts (1992) have argued that coping behavior should match the situation in order to be ef­fective. A recent study in a hospital setting supports this as­sumption (de Rijk, Le Blanc, Schaufeli, & de Jonge, 1998). Problem-focused coping was found to be only superior in sit­uations in which nurses could exert control over their work situations. In low-control situations, attempts of problem-focused coping were negatively associated with individuals' well-being.

Locus of Control

Locus of control (Rotter, 1966)—an individual difference concept—refers to whether individuals see themselves as pri­marily able to control their lives and their major experiences (internal locus of control) or whether individuals think that other people or forces beyond themselves (e.g., luck) deter­mine what happens to them (external locus of control). At the most general level, it is assumed that individuals with an in­ternal locus of control exert more direct action against the stressor than do those with an external locus of control. Therefore, it is expected that they will suffer less from work-related stressors (Cohen & Edwards, 1989). Indeed, individ­uals with an internal locus of control experience better mental health than do individuals with an external locus of control (for reviews, cf. Glass & McKnight, 1996: Kahn & Byosiere, 1992). Such a positive effect of an internal locus of control was also confirmed in longitudinal studies (Daniels & Guppy, 1994; Newton & Keenan, 1990).

Additionally, it was tested whether a high internal locus of control buffers the negative effects of a stressful work situa­tion. Findings from cross-sectional studies seem to support such a moderator effect (for a review, cf. Kahn & Byosiere, 1992). However, results from longitudinal studies are less conclusive. For example, in the study by Newton and Keenan (1990), only a small portion of the tested moderator effects reached their significance level. Longitudinal studies by Parkes (1991) and Daniels and Guppy (1994) reported more complex three-way interactions between stressors in the work situation, job control, and locus of control.

Taken together, research suggests that locus of control has a main effect on well-being. However, longitudinal studies did not provide evidence for a simple moderator effect of locus of control on the relationship between stressors and strains.

Self-Esteem, Self-Efficacy, andCompetence

Self esteem and self-efficacy are important for an individual's health and well-being. There is consistent empirical evidence for a main effect of self esteem and self-efficacy (for reviews,

cf. Kahn & Byosiere, 1992; Sonnentag, 2002). Evidence f_ora moderator effect of self-esteem is weak (Jex & Elacqua, 1 99q\

With respect to self-efficacy, there is more evidence_____

although not unequivocal—for a moderator effect. Some stud­ies show that the relationship between stressful work situations and poor well-being is stronger for individuals low on self-efficacy than for individuals high on self-efficacy (Jex & Bliese, 1999; VanYperen, 1998). There are additional studies that reported this moderator effect for some but not all of the studied stressor or strain measures (Bhagat & Allie, 1989; Jex & Elacqua, 1999). Jex and Gudanowski (1992) and Saks and Ashforth (2000) did not find an interaction effect for self-efficacy. Parker and Sprigg (1999) provide evidence that proactive personality—a concept closely related to self-efficacy—attenuates the stressor-strain relationship, particu­larly when job control is high. Also recent work by Schaubroeck and his coworkers suggests a more complex pic­ture with three-way interactions between stressors, job con­trol, and self-efficacy (Schaubroeck, Lam, & Xie, 2000; Schaubroeck &Merritt, 1997).

Because self-efficacy is an individual's belief that he or she is competent, the issue of subjective competence can be discussed within the self-efficacy framework. Surprisingly, we know of no studies on objective competence and skills as resources in the stress process. This is all the more surprising because skills needed at work should be the prime candidates for dealing with stressors.

Other Person Factors

In the past, researchers paid attention to the Type A behavior pattern as one important individual difference variable in explaining negative effects of stressful work situations, par­ticularly with respect to cardiovascular diseases. Type A indi­viduals are competitive, hostile, impatient, and hard driving. Ganster and Schaubroeck (L991) and Kahn and Byosiere (1992) summarized the findings of studies on Type A behav­ior pattern. There is some support for a main effect of Type A behavior on strain. More specifically, the hostility component was found to be closely related to physiological reactivity (Ganster, Schaubroeck, Sime, & Mayes, 1991). In contrast, the evidence for a moderator effect of Type A behavior pat­tern is weak (Kahn & Byosiere, 1992). More recent longitu­dinal studies are inconclusive. Type A behavior enhanced the relationship between stressors and strains in one study (Moyle & Parkes, 1999), whereas it attenuated this relation­ship in another study (Newton & Keenan, 1990).

Hardiness is another individual difference variable assumed to moderate the stressor-strain relationship. Hardiness com­prises the dimensions commitment, control, and challenges

(Kobasa, Maddi, & Kahn, 1982). There is some evidence for a main effect of hardiness on individual health, but support for a moderator effect was found only in some studies (e.g., Howard et al., 1986) but not in others (e.g., Tang & Hammontree, 1992). Sense of coherence (Antonovsky, 1991) is a concept closely related to hardiness. Its central aspects are perceived comprehensibility, manageability, and meaningfulness of the environment. Recently, researchers included sense of coher­ence as a potential moderator in studies on work-related stress. Cross-sectional research suggests that sense of coher­ence can attenuate the negative impact of high-strain jobs (Soderfeldt, Soderfeldt, Ohlson, Theorell, & Jones. 2000). Longitudinal tests are needed to substantiate this effect.

Conclusions About Moderator Effects

Methodological reasons make it difficult to detect moderator f effects, particularly in nonexperimental studies. Moderated regression analysis is a conservative procedure that makes it hard to establish moderator effects. Thus, the field of moder­ators in stress research may very well have to deal with a large Type II error (i.e., not finding in research what exists in reality). First, main effects are entered first into the regression equation, and therefore not much variance remains to be ex­plained by the interaction term. This problem is enhanced in longitudinal studies in which the initial level of the strain measure (i.e., the dependent variable) is also entered into the regression equation as a control variable. Because individual strain measures are fairly stable over time, a large proportion of the variance of the dependent variable is already ex­plained. Thus, there is little variance left to be explained by the interaction effect. Second, most stress studies rely on rel­atively small sample sizes; this implies that the studies do not have enough power for detect the moderator effects even if they exist (Aiken & West, 1991).

Consequently, empirical findings on moderator effects are mixed. There are some studies—including those using longi-tadinal designs—that speak for a moderator effect of control, social support, and coping styles. Cross-sectional findings on a moderator effect of self-efficacy are encouraging. However, ^suPport for a moderator effect of locus of control, Type A khavior, or hardiness are weak.

" we analyze these findings in the light of methodological P^ems associated with the test of moderator effects, it

^ms warranted to continue research in this area. However, think that the following recommendations may make it

^re likely to find moderator effects: First, more attention

W be paid to a match between specific stressors and

^Specific moderators (cf. Cohen & Wills, 1985). For example, it

P ^ausible to assume that social support, which provides

Empirical Evidence 471

additional information on role requirements, will attenuate the negative impact of role ambiguity but not the negative impact of high time pressure. Second, large sample sizes are needed for ensuring sufficient power for detecting effects. Third, de­sign issues are important as well. Given the power issues in­volved, one can select workplaces with the extremes of stressors (high vs. low stressors) and resources (e.g., very high vs. very low control) and test for interactions within such a de­sign (Aiken & West, 1991). Fourth, it is necessary to under­stand better whether the resources have an impact on stressors (and vice versa). One reason may be that, for example, control at work leads to a reduction of certain stressors (particularly those that match the control). If this is the case, then we would know why resources are sometimes negatively related to stres­sors. One way to deal with the problem of confounding be­tween resources and stressors is to study people who are new in their jobs. Finally, we suggest combining experimental and field studies to a larger extent, attempting to simulate in the ex­periment the same types of stressors and resources that are studied in the field.

In summary, research on resources has revealed main ef­fects of resources on health an well-being; this implies that the availability of resources is helpful and beneficial in itself and across a wide range of situations. Additionally, there is some—although not unequivocal—evidence that certain re­sources can attenuate the negative effects of stressors on health and well-being. Particularly important are control at work, social support, coping styles, and self-efficacy.

Stress andPerformance

Stress in organizations may influence not only individual health and well-being but may also influence performance. Performance refers to individuals' actions that are relevant for organizational goals (Campbell, McCloy, Oppler, & Sager, 1993). Borman and Motowidlo (1993) differentiated between task and contextual performance. Task performance refers to in-role behaviors that contribute to the organiza­tion's technical core. Contextual performance refers to extra-role, discretionary behaviors that do not directly contribute to an organization's technical core but that are assumed to sup­port its broader organizational, social, and psychological environment.

There are several contradictory assumptions about how stressors in organizations affect performance. It is plausible to assume that stressors have a negative linear effect on per­formance. Such a negative effect can be explained by direct and indirect effects. The direct effect implies that stressors— particularly situational constraints—make task accomplish­ment more difficult, if not impossible. For example, if a task

different task strategies (Sperandio, 1971). Hockey (2000) of­fers an additional explanation for the inconsistency between laboratory and field study results: Many laboratory tasks are relatively simple, trivial, and undernamed. If stressors occur in such a situation, study participants have few possibilities to switch to different strategies, be it because of a lack of skills in the specific task, or because of the restrictions of the laboratory setting. Real-life work tasks, however, are usually well-learned and complex. If stressors occur in these real-life situa­tions, individuals often possess the necessary skills to pursue different strategies. Moreover, in organizational settings, goal attainment has high priority; this implies that task performance must be protected, if necessary, at the expense of increased ef­fort or neglect of subsidiary activities. Klein (1996) addition­ally argues that some of the cognitive strategies affected by stressors in laboratory settings play a minor role in real-life ^settings. For example, analytical decision strategies suffer "from time pressure, but such strategies are rarely used in nat­ural decision making; therefore, the negative impact of perfor­mance is limited.

There are a few studies that examined the relationship be­tween stressors and contextual performance. For example, Motowidlo, Packard, and Manning (1986) reported negative relationships between the intensity and frequency of stress-fill events on the one hand and interpersonal aspects of job performance of nurses on the other hand. Kruse {1995, cited in Jex, 1998) tested whether situational constraints were related to organizational citizenship behavior (OCB) and reported negative relationships between situational con­straints and three aspects of OCB. These findings suggest that in stress situations, individuals assign priority to maintain task performance at the expense of discretionary behaviors such as contextual performance. However, a longitudinal study by Fay and Sonnentag (in press) suggests that the ex­perience of stressors at work can even have an enhancing effect on extrarole performance and personal initiative. Sim-ilarly, Bunce and West (1994) reported that health care pro­fessionals responded with innovations to the experience of Pressors at work.

Taken together, laboratory studies showed that stressors impair basic cognitive processes. However, as field studies indicate, this impairment does not necessarily result in a de­cease in overall job performance. In particular, workload ^Has found to be associated with higher job performance. "ese findings suggest that individuals spend more effort, Prioritize the most relevant tasks, and use compensatory strategies for upholding their performance under stressful ^Sl'Uations. It remains unclear whether and how such a per-"Tfiance management strategy is associated with health or ^eU-being effects. It might be that such an approach ex-^Usts an individual's resources in the long run and there-

Empirical Evidence 473

fore affects an individual's health and well-being in a nega­tive way.

Date: 2016-03-03; view: 927