What Is Too Much or Too Little? The Curvilinear Effects of Job Tension on Turnover Intent, Value Attainment, and Job Satisfaction

What Is Too Much or Too Little? The Curvilinear Effects of Job Tension on Turnover Intent, Value

Attainment, and Job Satisfaction

SUZANNE ZIVNUSKA CHRISTIAN KIEWITZ De,uurtmen/ ifMunugement Munagement/Markrtmg Drpurtment

Floridu Siute Un ivr rdy Univertity of Dayton

WAYNE A. HOCHWARTER~ AND PAMELA L. PERREWE Department CfManugement

Floridu Stute Univrrsi!v

KELLY L. ZELLARS Department of Munugemmt

Univerwy ofNorth Carolina, Charlotte

This study addressed 2 rclativcly unexplored issues in the current job stress literature. First, a significant body of research has substantiated a curvilinear relationship between job ten- sion and performance without examining the potential impact of work stress on other important outcomes. Second past research has failed to control for the influence of role stressors and other important variables that might alter the job-tension/work-outcomes rela- tionship. We hypothesized that experienced job tension would predict turnover intent, value attainment, and job satisfaction, and that this relationship would he nonlinear. Data gath- ered from 270 hotel managers confirrncd our hypotheses. Notably, tension predicted turn- over intent (U shape), value attainment (inverted-U shape), and job satisfaction (inverted-U shape) in the expcctcd manner, while explaining additional variance beyond that contrib- uted by demographic factors, role stressors, and linear tension terms. Implications of these findings for theory and practice, as well as avenues for future research are provided.

For most individuals, the words job tension conjure a variety of reactions, most of which can be categorized as unpleasant. Accordingly, job tension has been linked empirically with a number of negative outcomes including dissatis- faction, depression, somatic disorders, and turnover intent (Frone, 1990; 0' Driscoll & Beehr, 1994). From a practical perspective, recent accounts indicate that stress costs American organizations between $200 billion and $300 billion annually (DeFrank & Ivancevich, 1998), partially because of its contribution to

'Correspondence concerning this article should he addressed to Wayne A. Hochwarter, Depart- ment of Management, Florida State University, Tallahassee, FI. 32306-1 1 10. E-mail: whochwartj cob.fsu.edu

1344

Journal of Applied Social Psychology, 2002, 32, 7, pp. 1344-1 360. Copyright 0 2002 by V. H. Winston & Son, Inc. All rights reserved.

TENSION AND WORK OUTCOMES 1345

increased absenteeism and turnover, and to decreased performance (Dwyer & Ganster, 1991).

Given its theoretical and practical importance, academicians have spent a considerable amount of effort examining both precursors and outcomes of orga- nizationally induced tension (Motowidlo, Packard, & Manning, 1986; O’Driscoll & Beehr, 1994). Academicians have generally embraced a “less is best” view, which suggests that low levels of tension result in fewer dysfunctional outcomes than do high levels of tension (Gupta & Beehr, 1979; House & Rizzo, 1972). Significant correlations between job tension and a number of important work outcomes support this view (Cropanzano, Howes, Grandey, & Toth, 1997). Addi- tionally, Keller (1984) found that job stress was correlated with job satisfaction. Finally, Hemingway and Smith (1999) reported that role stress was related to intent-to-leave scores in a study utilizing a nursing sample.

Although relationships between job tension and various work outcomes are customarily viewed as linear, alternative conceptualizations exist in the literature (Jamal, 1984, 1985). For example, research has indicated that jobs characterized by low levels of tension may not be as satisfying as those that provide greater lev- els of anxiety-provoking stimuli (Singh, 1998). The central argument employed in this line of research, and one that is elaborated upon in subsequent sections of this paper, is that a job that provides no tension (or close to no tension) might fail to offer much in the way of excitement, resulting in understimulation and frustra- tion. Similarly, jobs that offer substantial levels of tension might overwhelm incumbents, leading to overstimulation and dysfunctional outcomes (Kahn & Byosiere, 1992).

Although very low and very high concentrations of tension are thought to have negative impacts on employees, moderate quantities of tension are actually thought to be beneficial. Incumbents perceive moderate tension as stimulating and challenging, without being burdened by an overabundance of anxiety, thus contributing to positive work outcomes. Based on this theoretical development, proponents of this perspective advocate adopting nonlinear conceptualizations of the job-tensiodwork-outcomes relationship (Anderson, 1 976).

The current study was developed to further examine the validity of the nonlin- ear conceptualizations of job tension outlined earlier. In doing so, we seek to ful- fill a number of objectives. Specifically, researchers who have adopted the understimulation-overstimulation view generally have used job characteristics as the independent variable (Champoux, 1978; Xie & Johns, 1995). Although exam- ples of stress or job-related stressors as the independent variable can be found in the literature (e.g., Jamal, 1984, 1985; Singh, 1998), these studies generally are underrepresented in the literature. Therefore, the first objective of the present study is to address this lack of development. Moreover, job performance has served as the dependent variable in nearly all cases (Levi, 1972; Xie & Johns, 1995). Incorporating other worthy outcome variables appears to be warranted and

1346 ZIVNUSKA ET AL.

represents the second objective of the current research. Specifically, we include two traditional outcome variables (i.e., job satisfaction and turnover intent), and one that is considered less traditional (e.g., value attainment), in an attempt to expand the current level of understanding in this area and to generalize the curvi- linear tension effects to other important outcomes.

Research Background and Hypotheses

Despite the fact that research assessing nonlinear associations between ten- sion and job outcomes is a relatively new phenomenon, its foundation can be traced back nearly a century. Yerkes and Dodson (1908) proposed an inverted-U- shaped association between tension and performance in clinical studies assessing the relationship between the level of arousal/stress and memory performance. Research that has built on this foundational work has generally supported the notion that extremely low and extremely high levels of tension are dysfunctional. Low tension levels fail to encourage employees to actively engage themselves in work, whereas high levels of stress or tension result in an unmanageable level of anxiety that often renders individuals incapacitated (Levi, 1972; McGrath, 1970). Between low and high levels of tension, moderate levels have a beneficial impact on employee attitudes and behaviors. When tension levels increase from low to moderate, attentiveness is improved and challenge is furnished, until an optimal level is reached. However, once this optimal level is exceeded, further increments in tension are associated with decreased levels of positive work outcomes (Christianson, 1992).

Similarly, activation theory (Malmo, 1959; Scott, 1966) suggests that stimula- tion can serve as a powerful motivator only until an optimal level is achieved. In support of this theory, Champoux’s (1 978) work revealed that job characteristics led to detrimental outcomes when they were perceived as disproportionately high. Champoux argued that inordinately high levels of job characteristics (i.e., exceedingly high levels of skill variety and autonomy) would result in heightened levels of job tension by representing “too much of a good thing.”

Further supporting the negative impact of low and high levels of tension, Gardner and Cummings (1988) stated that “job stress occurs whenever job- related stimuli cause a job holder’s experienced activation level to deviate sub- stantially from one’s characteristic level of activity” (p. 106). Thus, unless one maintains either a chronically high or a chronically low level of anxiety through- out all life activities, a job that provides a modest level of stress should not neces- sarily provoke negative reactions. Likewise, Schaubroeck and Ganster ( 1993) noted that passive coping occurs in situations characterized by both high and low tension. As a consequence, individuals experience low motivation coupled with a lack of resource activation or a lack of resource availability caused by excessive job demands. The use of passive coping, either by choice or necessity, inherently

TENSION AND WORK OUTCOMES 1347

diminishes one’s ability to adapt to the environment, resulting in hrther deterio- ration of performance. Although most research in this domain has focused almost exclusively on performance-related outcomes, Singh (1 998) argued that compa- rable theorizing is applicable to other important outcomes such as job satisfaction and turnover intent.

Expansion of the Job-TensionlWork-Outcomes Relationship

The current study represents an attempt to expand on our understanding of the job-tensionlwork-outcomes relationship. As noted in the previous section, the theoretical tenets of activation theory and other frameworks (e.g., Yerkes & Dodson, 1908) used to explain the tension-performance association can be applied easily to other theoretically relevant outcomes. Given that past research has corroborated positive relationships between tension and job dissatisfaction (Keller, 19841, as well as turnover intent (Cropanzano et al., 1997), including ten- sion and dissatisfaction in research assessing the nonlinear prediction of job ten- sion appears plausible.

We contend that job tension will exhibit significant linear associations with job satisfaction and turnover intent. However, we maintain that including the non- linear term will explain a significant level of variance beyond the contribution of the linear term alone. Consistent with arguments summarized earlier, exception- ally low levels of job tension decrease satisfaction and increase turnover intent by providing employees with only minimal levels of stimuli that encourage engage- ment in work. Although beyond the scope of the current study, it is plausible that at low levels of activation, employees are not motivated to involve themselves fully in work at the level required for satisfaction and desire to remain. At the opposite end of the continuum, overstimulation (i.e., excessive levels of tension) burdens the employee (Schwab & Cummings, 1976). Decreases in satisfaction and increases in turnover intent are therefore logical consequences for those who have reached this level of distress. Given the conceptualization of these two out- come variables, we expect that the relationship between tension and job satisfac- tion will be depicted best by an inverted-U shape, and that a U shape will best depict the relationship between tension and turnover intent. Therefore, the fol- lowing hypotheses are presented:

Hypothesis 1. An inverted-U shape will characterize the relation- ship between job tension and job satisfaction. Specifically, job sat- isfaction will be low when tension is low, will increase when tension is moderate, and will decline as tension increases.

Hypothesis 2. A U shape will characterize the relationship between job tension and turnover intent. Specifically, turnover intent will be

1348 ZIVNUSKA ET AL.

high when tension is low, will decrease when tension is moderate, and will increase as tension increases.

Value attainment, relative to job satisfaction and turnover intent, represents a consequence of work that has received much less research attention despite the fact that it has been shown to be an important outcome variable (Rokeach, 1973). The attainment of values reflects a comparison of one’s life activities relative to a predetermined standard, such as success or virtue (Tatarkiewicz, 1976). Over 50 years ago, Lewin ( 1 95 1) argued that activities that occurred at work had the potential to either enhance or frustrate the development of values. According to George and Jones (1 996), value attainment “describes the degree to which people view their jobs as helping them to achieve important overarching life goals and standards, [and] the extent to which work enables people to progress toward desirable states of existence” (p. 320).

Finally, Locke (1976) noted that the work environment might be perceived more favorably if it allows employees to attain their values. Research has sup- ported the belief that the attainment of terminal values (Rokeach, 1973) can rep- resent an important consequence of work. For example, George and Jones (1996) found that the attainment of values was significantly associated with turnover intent and job satisfaction.

We maintain that job tension will exhibit significant effects on value attain- ment, given that values can be predicted by the work experiences that individuals encounter on the job (James & James, 1989). It might be that low-tension tasks require too little of an employee. If an employee is nut provided a challenging task or is unable to utilize goal-directed behavior, value attainment becomes less likely. In related job-stress research, Schabracq, Cooper, and Winnubst (1 999) argued that work underload (i.e., low tension) often leads to a host of dysfunc- tional consequences, such as concentration problems, irritability, apathy, and boredom. Alternatively, high-tension environments place excessive demands on employees, which results in success becoming less likely. At both extremes, attaining one’s goals, and thus achieving what one values, becomes problematic and less feasible. Given this discussion, the following hypothesis is provided:

Hypothesis 3. An inverted-U shape will characterize the relation- ship between job tension and value attainment. Specifically, value attainment will be low when tension is low, will increase when ten- sion is moderate, and will decline as tension increases.

Research Design of the Current Study

Critical to the design and execution of the present study was the acknowl- edged importance of control variables that might alter the relationship between

TENSION AND WORK OUTCOMES 1349

tension and our dependent variables. To account for the impact of these variables, we controlled for age, gender, and role stressors (i.e., role conflict, role ambigu- ity, and work overload). Each of these variables has been shown to explain signif- icant levels of criterion variance in past research. For example, age and gender have been shown to predict turnover in that younger employees and females have been shown to leave more often than older employees and males (for a discus- sion, see Griffeth & Hom, 1995; Kacmar & Ferris, 1989).

Role stressors such as role conflict, role ambiguity, and work overload have been shown to contribute to employee strain, tension, satisfaction, and turn- over in numerous studies (e.g., Beehr, 1976; Frone, 1990; Jackson & Schuler, 1985; Lysonski & Johnson, 1983; O’Driscoll & Beehr, 1994; Rhoads, Singh, & Goodell, 1994; Schaubroeck, Cotton, & Jennings, 1989). Because of the impact that role stressors have been shown to have on tension as well as turnover and job satisfaction, they were considered important control variables.

Method

Participants

The sample consisted of managerial employees in the hotel industry through- out the United States. Roughly 15 to 20 individuals employed at participating hotels agreed to complete surveys following regularly scheduled staff meetings during each session. Respondents represented a variety of hospitality establish- ments, although national hotel chains employed most individuals who participated in this study. None of the variables included in the present study were discussed at the time that surveys were completed. A total of 275 surveys were distributed and voluntarily completed. Because of missing data, the usable sample size was reduced to 270. Respondents’ mean age was 33.5 years (SD = 8.17; range = 23 to 70), and 54.5% ( N = 147) of the sample was female. Tenure ranged from 1 to 48 years, with a mean of 1 1.17 years (SD = 8.73). Respondents reported working an average of 52 hr per week (SD = 1 1.3 1 ; range = 10 to 90) and were responsible for directing the activities of roughly 22 employees (SD = 44.78; range = 0 to 400).

Measures

Role stressors. Role conflict (8 items; a = .81), role ambiguity (6 items; a =

.84), and overload (1 item) were measured using a scale developed by Rizzo, House, and Lirtzman (1970). Each scale utilizes a 7-point Likert format ranging from 1 (strongly agree) to 7 (strongly disagree). Examples of conflict items include “I have to work under vague directions or orders,” and “I receive incom- patible requests from two or more people.” Ambiguity items include “I know exactly what is expected of me” (reverse scored), and “I feel certain about how

1350 ZIVNUSKA ET AL.

much authority I have” (reverse scored). Finally, the overload item was “It often seems like 1 have too much work for one person to do.”

Job-induced tension. We measured job-induced tension with a seven-item measure developed by House and Rizzo (1972). Sample items include “I work under a great deal of tension,” and “If 1 had a different job, my health would prob- ably improve.” Participants were asked to provide information using a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly agree). The scale provided an adequate level of internal consistency (a = .84).

Job satisfaction. Job satisfaction was measured using a three-item scale of the Michigan Organizational Assessment Questionnaire (Cammann, Fichman, Jenkins, & Klesch, 1979). A sample item is “All in all, I am satisfied with my job.” The 7-point scale had a response format ranging from 1 (strongly disagree) to 7 (strongly agree). The coefficient alpha for this scale was .67.

Turnover intent. We used six items to measure turnover intention (Mobley, Horner, & Hollingsworth, 1978; Mowday, Koberg, & MacArthur, 1984). A sam- ple item in the 7-point scale ranging from 1 (strongly disagree) to 7 (strongly agree) is “I intend to quit my job.” The reliability estimate for this scale was .92.

Value attainment. Consistent with past research (George & Jones, 1996), we used a modified version of the Rokeach Value Survey (Rokeach, 1973) to assess attainment of terminal values. The 18-item scale asked respondents to indicate the extent to which one’s present job encourages value attainment. Sample items include “A sense of accomplishment” and “Self-respect.” The 5-point scale rang- ing from 1 (to no extent) to 5 (to a great extent) had an internal consistency esti- mate of .93.

Results

Table 1 reports the means, standard deviations, and intercorrelations for all of the variables in the present study. All reliability estimates, with the exception of the three-item measure of job satisfaction, were greater than .70, providing evi- dence of scale acceptability (Nunnally, 1978). In general, most correlations were consistent with those reported in past research with respect to direction and mag- nitude. Given the somewhat high correlations between job satisfaction, turnover intent, and value attainment (correlations ranging from .44 to -.56), we conducted additional analyses to assess the uniqueness of these outcomes.

Differentiating Job Satisfaction, Turnover Intent, and Value Attainment

Confirmatory factor analysis (CFA), using maximum likelihood estimation, was employed to assess the extent to which turnover intent, job satisfaction, and value attainment represented discrete constructs. Two models were estimated utilizing LISREL 8 (Joreskog & Sorbom, 1993). In Model 1, all items loaded on a single, global factor. Model 2 consisted of three factors, with items loading on

Tabl

e 1

Mea

ns, S

tand

ard

Dev

iatio

ns, a

nd In

terc

orre

latio

ns A

mon

g St

udy

Var

iabl

es

Var

iabl

e M

SD

1

2 3

4 5

6 7

8

1. A

ge (i

n ye

ars)

2. G

ende

ra

3. R

ole

conf

lict

4. R

ole

ambi

guity

5. O

verlo

ad

6. J

ob te

nsio

n

7. T

urno

ver i

nten

t 8.

Val

ue a

ttain

men

t

9. J

ob sa

tisfa

ctio

n

33.5

0 8.

18

1.54

0.

51

3.21

1.

19

2.83

1.

26

4.34

1.

17

3.83

1.

29

2.77

1.

62

3.11

0.

76

5.41

1.

27

-

.01

-.09

-.16*

* -.0

1

-.02

-.15*

*

.08

.17*

*

__

-.03

-.01

.37*

* -

-

.08

.21*

* .3

3**

-

.o 1

.40*

* .2

3**

.33*

** -

.03

.31*

* .1

6**

-.04

.20*

* -

.o 1

-.25*

* -.2

0**

.07

-.26*

* -.4

6**

-

.03

-.37*

* -.3

2**

.03

-.33*

* -.5

6**

.44*

*

qend

er: 0

= fe

mal

e, 1

= m

ale.

**

p <

.05.

***p

< .0

1.

1352 ZIVNUSKA ET AL.

their corresponding construct. The distinctiveness of the three factors was assessed by means of the parsimony normed fit index (PNFI) and a chi-square difference test (Bollen, 1989).

Model 1 did not fit the data well (GFI = .55 , AGFI = .47, CFI = .63, IF1 =

.64). Although still far from optimal, Model 2 provided a much better model fit (GFl = .78, AGFI = .72, CFI = 24 , IF1 = .84). The difference between the PNFI for the two models was found to be .I7 ( .55 for Model 1; .72 for Model 2), which exceeds the value deemed indicative of substantial model improvement (i,e., .06 to .09; Hair, Anderson, Tatharn, & Black, 1992, p. 687). Moreover, the chi-square difference between the models was significant, ~ ~ ( 3 , 265) = 1,943.25, p < . O l . From these data, we conclude that Model 2 represents a significantly different and better fitting model, hence supporting the distinctiveness of the three out- come measures.

Hypothesis Testing

We conducted hierarchical polynomial regression analyses to assess the form and magnitude of the relationship between job-induced tension and our three dependent variables. In the first step, the control variables of age and gender were entered, followed by the role stressors (i.e., conflict, ambiguity, and overload) and the linear job-tension term in the second step. In Step 3 , the curvilinear tension term was entered, followed by the cubed tension term in Step 4. If linearity is the best representation of the form of the tension-turnover relationship, tension alone should explain a significant amount of criterion variance. However, if the inclu- sion of the tension2 term explains variance beyond that attributed to the linear term, a curvilinear form (U shape or inverted-U shape) best represents the pro- posed relationship. Finally, if the tension3 term explains variance beyond that accounted for by the linear and curvilinear terms, the relationship would be char- acterized by two bends in the curve (i.e., an S shape).

Results from this analysis are shown in Table 2. Age predicted turnover intent and job satisfaction, while gender failed to predict any work outcome. In general, the control variables collectively predicted our dependent variables, with role conflict providing the most consistent effect. As suggested, the linear tension term predicted each outcome in the expected direction. Finally, the tension2 term predicted job satisfaction, turnover intent, and value attainment, explaining an additional 2% of criterion variance in each instance beyond the variance explained by the linear tension term. The positive beta in Step 3 for turnover intent indicates that the curvilinear relationship is best depicted by a U shape. Conversely, the negative beta estimates found in Step 3 for value attainment and job satisfaction indicate that an inverted-U shape best represents the form of the relationship. The confirmed curvilinear relationships between job tension and all work outcomes are illustrated in Figure 1.

TENSION AND WORK OUTCOMES 1353

Table 2

Hierarchical Regression Analysis of Linear and Nonlinear Tension Terms Predicting Turnover Intent, Value Attainment, and Job Satisfaction

Turnover intent Value attainment Job satisfaction

Predictor P AR2 P AR2 p A R ~

Step 1

Age Gender

Role conflict Role ambiguity Overload Tension

Tension2

Tension3

Step 2

Step 3

Step 4

-.03** .13 .03

.35** -.01 -. 14 .14* .lo**

.11** .02**

.o 1 .oo

.01 .03***

.01 .01 .07 .03

-.08* -.24*** -.04 -.16*** .02 -.13**

-.lo*** .12*** -.22*** .21***

-.05** .02** -.07** .02**

-.01 .oo -.01 .oo *p < . lo. **p < .05. ***p < .01.

Discussion

Results from the current investigation provide strong support for each of our hypotheses. Notably, when tension was perceived as exceedingly low or high, dysfunctional outcomes surfaced, as evidenced by lower satisfaction and value attainment, and higher turnover intent scores. By substantiating nonlinear rela- tionships between tension and job satisfaction, turnover intent, and value attain- ment, we have made an initial attempt to extend the current body of literature by examining important outcomes other than performance (Xie & Johns, 1995). In a general sense, our results, coupled with those provided earlier (Jamal, 1984, 1985), suggest that both very high and very low levels of tension have the ability to predict a host of negative consequences.

Strengths and Limitations of the Present Study

Three strengths of the present study emerge. First, we expanded the criterion domain to include variables not often examined in this line of research. Second, to ascertain the most accurate depiction of tension-outcomes relationships, we included a variety of control variables that have the potential to alter bivariate

1354 ZIVNUSKA ET AL.

2.4 i

2 2

E 1 8

5 1.6

1.4

In 20

8

1 0 I-, I 1 I , I , 1 2 3 4 5 6 7

Tension scores

+Turnover intent *Value attainment 4 Job satisfaction

Figure 1 . Curvilinear effects of job tension on job satisfaction, value attainment, and turnover intent.

relationships between independent and dependent variables. For the most part, the chosen control variables were related to our outcome variables, hence sup- porting their inclusion. Additionally, most studies that have substantiated U- shaped relationships between tension and various work outcomes have relied on student samples or were conducted in laboratory settings (Cohen, 1980). Accord- ingly, past investigations have noted this overreliance as a limitation of previously conducted research (Jamal, 1984). The current research therefore represents one of the few empirical investigations that has corroborated nonlinear relationships between tension and work outcomes (other than job performance) utilizing par- ticipants employed in field environments. Finally, the sample was large enough to ensure adequate statistical power.

Despite these strengths, the results of the current study need to be interpreted in light of its limitations. Of primary concern is our reliance on self-report data, which raises the possibility that common method variance might artificially inflate the relationships noted here. However, James, Gent, Hater, and Corey (1979) noted that claims of common method variance materialize only if there appears to be a general and pervasive influence that is operating systematically to inflate observed relationships. Observation of the correlation matrix identifies a number of significant correlations among study variables. However, as shown in Table 1 , there does not appear to be an overwhelming number of correlations that one would consider very high, given the theoretically proposed relationships out- lined earlier. Moreover, the relationships that appear to be the most problematic represent our three outcome variables (i.e., job satisfaction, turnover intent, and value attainment) with correlations ranging from -44 to -.56. To determine the extent to which problems related to common method variance existed, we under- took the additional step needed to determine the distinctiveness of these three outcome variables. These results were provided earlier and indicated that the three-factor solution was superior to the one-factor model. Finally, Crampton and

TENSION AND WORK OUTCOMES 1355

Wagner (1 994) noted that percept-percept inflation, although an issue to take into account when designing studies, typically does not have the far-reaching, adverse effects envisioned by critics. In addition, they argued that much of the potentially harmful impact of opting for this research strategy is abated when demographic variables are used as covariates (as was the case in this study). Taken together, these results greatly reduce the possibility that our findings were adversely impacted by common method variance.

Finally, we need to note that the reliability estimate for one of our outcome measures, job satisfaction (a = .67), can be classified as less than acceptable by conventional standards. Nunnally (1 978) argued that internal consistency esti- mates greater than .70 are sufficient and that efforts to increase reliability above .80 normally do not yield an adequate return on one’s investment, given that cor- relations are affected negligibly by measurement error at this level. At this point, we are confident that the reliability estimate for this measure is close enough to warrant an adequate level of acceptance. We undertook additional analyses to determine which of the three job satisfaction items were potentially problematic. Exploratory factor analyses indicated that the item “In general, I do not like my job” was somewhat questionable. We deleted this item and reexamined the rela- tionships hypothesized in this study. Hierarchical regression results failed to indi- cate any tangible differences when employing the two-item versus the three-item measure of job satisfaction. In summary, we contend that the lower than expected level of agreement for our measure of job satisfaction is likely a function of scale length, rather than some underlying empirical or conceptual concern.

Directions for Future Research

Results of this investigation provide fertile ground for additional research. At a minimum, replications are needed to further substantiate the results reported here, given that respondents in the current study represented a very narrow job level (e.g., managers) in only one specific industry (e.g., hospitality). Moreover, we were selective with respect to the control variables examined in this study. Despite our best intentions, we employed a relatively small subset of variables that we thought would impact the relationships investigated here. Specifically, we included both individual-difference variables (i.e., age and gender) and environ- mental cues (i.e., role stressors). The inclusion of dispositional factors might also be fruitful in this regard. For example, including negative affectivity (NA) and positive affectivity (PA) would be helpful, given that past research has shown that dispositions such as these have the ability to affect the way that individuals interpret stressful events (Brief, Burke, George, Robinson, & Webster, 1988; Burke, Brief, & George, 1993). Specifically, because both NA and PA have been shown to be associated with perceptions ofjob tension (Cropanzano et al., 1997), i t is plausible that the high NAs will view all levels of tension negatively the

1356 ZIVNUSKA ET AL.

magnitude of objective tension notwithstanding. On the other hand, high PAS might report positive work outcomes, regardless of the level of tension (Jex & Bliese, 1999).

Additionally, we found that low and high levels of tension led to dysfunc- tional consequence and that positive outcomes occurred somewhere between these two extremes. In essence, this line of research suggests that tension scores reach an optimal point that provides evidence of positive work (e.g., satisfaction, turnover), and that these outcomes become less positive beyond this point. One question worthy of investigation is Where is this optimal point? Is it relatively stationary for all individuals, or does it move about as a function of the strength and magnitude of the perceived tension? At this point of departure, it might be advantageous to adopt qualitative methods of data collection to provide the best illustration of where this point resides, as well as its stability.

Finally, the coping literature might provide some insight as to when extremely low levels and extremely high levels of stress are the most harmful (Anderson, 1976). For example, it is conceivable that high levels of tension are harmful only when accompanied with an inability to cope, or when low levels of social support are present. For example, past research has shown social support to reduce the likelihood that stressful situations will be perceived as such (Carlson & Perrewe, 1999). In this regard, we are suggesting that the ability to cope and the presence of social support might serve to moderate the relationships examined in this study. Without empirical substantiation, however, these ideas are only speculative.

Managerial Implications

In addition to the theoretical contribution noted earlier, this study provides several prescriptions for managerial practice. For example, our results suggest that some tension might be appropriate, as well as advantageous, to organiza- tional functioning. As a consequence, development programs aimed at eliminat- ing all workplace tension, in fact, might be doing more harm than good. Instead of looking for ways to abolish tension at work, managers might find it fruitful to manage its predictors and outcomes to ensure that individuals are receiving enough activation and challenge. By doing so, managers have a better chance of securing an adequate level of performance, and employees are more likely to maintain a sense of satisfaction and commitment. Finally, managers need to be cognizant of the fact that the relationship between tension and important work outcomes is not linear. Acknowledging that some tension is appropriate does not suggest that “more is better and most is best.” As noted in this study, once tension reaches an optimal level, its effects become an impediment to organizational and individual functioning. As noted earlier, determining where this optimal level resides is an important job for managers and one that will undoubtedly challenge even the most adept decision maker.

TENSION AND WORK OUTCOMES 1357

In summary, both specific and general contributions come into focus. With respect to specific contributions, the findings reported here further corroborate the influence of job tension on a host of work outcomes. As a consequence, it is apparent that investigations assessing the predictive ability of job tension on a plethora of work outcomes will remain a meaningful area of inquiry. With regard to general contributions, evidence for expanding our conceptualizations to include theoretically derived nonlinear terms into research appears warranted. Despite the fact that substantiating a linear relationship does not preclude deeper, more complex relationships (Kacmar & Ferris, 1989; MacCallum, 1998; MacCal- lum & Mar, 1995), only a limited number of studies have opted for this approach. Few would maintain that the relationships that we encounter in life are best viewed as a straight line. Once researchers expel linear attitudes in favor of more realistic ways of viewing organizational phenomena, research should improve and intensify. We hope that this study represents a logical step in that direction.

References

Anderson, C. ( 1976). Coping behaviors as intervening mechanisms in the inverted-U stress-performance relationship. Journal of Applied Psychology,

Beehr, T. (1976). Perceived situational moderators of the relationship between sub- jective role ambiguity and role strain. Journal ofApplied Psycholoa, 61,35-40.

Bollen, K. (1989). Structural equations with latent variables. New York, NY: John WiIey & Sons.

Brief, A., Burke, M., George, J., Robinson, B., & Webster, J. (1988). Should neg- ative affectivity remain an unmeasured variable in the study of job stress? Journal of Applied Psychology, 73, 193- 198.

Burke, M. J., Brief, A. P., & George, J. M. (1993). The role of negative affectivity in understanding relations between self-reports of stressors and strains: A comment on the applied psychology literature. Journal of Applied Psychol-

Cammann, C., Fichman, M., Jenkins, D., & Klesch, J. (1979). The Michigan Organizational Assessment Questionnaire. Unpublished manuscript, Univer- sity of Michigan, Ann Arbor,

Carlson, D., & Perrewe, P. (1 999). The role of social support in the stressor-strain relationship: An examination of work-family conflict. Journal of Manage- ment, 25,5 13-540.

Champoux, J. (1 978). A preliminary examination of some complex job scope- growth needs strength interactions. Academy of Management Proceedings,

Christianson, S. (1992). Emotional stress and eyewitness memory: A critical

61,30-34.

ogy, 78,402-4 12.

59-63.

review. Psychological Bulletin, 112,284-309.

1358 ZIVNUSKA ET AL.

Cohen, S. (1980). After-effects of stress on human performance and social behav- ior: A review of research and theory. Psychological Bulletin, 88, 82-108.

Crampton, S., & Wagner, J. (1 994). Percept-percept inflation in microorganiza- tional research: An investigation of prevalence and effect. Journal of Applied

Cropanzano, R., Howes, J., Grandey, A., & Toth, P. (1997). The relationship of organizational politics and support to work behaviors, attitudes, and stress. Journal of Organizational Behavior, 18, 159- 180.

DeFrank, R. S., & Ivancevich, J. M. (1998). Stress on the job: An executive update. Academy of Management Executive, 12,55-66.

Dwyer, D. J., & Ganster, D. C. (199 1). The effects of job demands and control on employee attendance and satisfaction. Journal of Organizational Behavior,

Frone, M. (1 990). Intolerance of ambiguity as a moderator of the occupational role stress-strain relationship: A meta-analysis. Journal of Organizutional Behavior, I I , 309-320.

Gardner, D., & Cummings, L. (1 988). Activation theory and job design: Review and reconceptualization. In B. Staw & L. Cummings (Eds.), Research in organizational behavior (pp. 8 1-122). Greenwich, CT: JAI.

George, J., & Jones, G. (1996). The experience of work and turnover intention: Interactive effects of value attainment, job satisfaction, and positive mood. Journal ofApplied Psychology, 81,318-325.

Griffeth, R., & Horn, P. (1995). The employee turnover process. In G. R. Ferris (Ed.), Research in personnel and human resources management (pp. 245- 293). Greenwich, CT: JAI.

Gupta, N., & Beehr, T. (1 979). Job stress and employee behavior. Organizational Behavior and Human Pevformance, 23,313-381.

Hair, J., Anderson, R., Tatham, R., & Black, W. (1992). Multivariate data aaaly- sis. New York, NY: Sage.

Hemingway, M., & Smith, C. (1999). Organizational climate and occupational stressors as predictors of withdrawal behaviors and injuries in nurses. Journal of Occupational and Organizational Psychology, 3,285-299.

House, R., & Rizzo, J. (1972). Toward the measurement of organizational prac- tices: Scale development and validation. Journal of Applied Psychology, 56,

Jackson, S. E., & Schuler, R. S. (1985). A meta-analysis and conceptual critique of research on role ambiguity and role conflict in work settings. Organiza- tional Behavior and Human Decision Processes, 36, 16-78.

Jamal, M. (1 984). Job stress and job performance controversy: An empirical assessment. Organizational Behavior and Human Performance, 33, 1-24.

Jamal, M. (1985). Relationship ofjob stress to job performance: A study of man- agers and blue-collar workers. Human Relations, 38,409-424.

PSychology, 79,67-76.

12,595-608.

388-396.

TENSION AND WORK OUTCOMES 1359

James, L., Gent, M., Hater, J., & Corey, K. (1979). Correlations of psychological influence: An illustration of the psychological climate approach to work envi- ronment perceptions. Personnel Psychology, 32, 563-588.

James, L., & James, L. (1989). Integrating work environment perceptions: Explo- rations into the measurement of meaning. Journal of Applied Psychology, 74, 739-75 1.

Jex, S., & Bliese, P. (1999). Efficacy beliefs as a moderator of the impact of work-related stressors: A multi-level study. Journal of Applied Psychology, 84, 349-361.

Joreskog, K., & Sorbom, D. (1993). LISREL 8 user5 guide. Uppsala, Sweden: University of Uppsala.

Kacmar, K., & Ferris, G. (1 989). Theoretical and methodological considerations in the age-job satisfaction relationship. Journal of Applied Psychology, 74,

Kahn, R., & Byosiere, P. (1992). Stress in organizations. In M. Dunnette & L. Hough (Ed.), Handbook of industrial and organizational psychology (pp. 571 -650). Palo Alto, CA: Consulting Psychologists Press.

Keller, R. (1984). The role of performance and absenteeism in the prediction of turnover. Academy of Management Journal, 27, 176-1 83.

Levi, L. (1972). Stress and distress in response to psychological stimuli. Elms- ford, NY Pergamon.

Lewin, K. (1951). Field theory in social sciences. New York, NY Harper. Locke, E. (1976). The nature and causes of job satisfaction. In M. Dunnette

(Ed.), Handbook of industrial and organizational psychology (pp. 1297- 1349). Chicago, IL: Rand-McNally.

Lysonski, S., & Johnson, J. (1983). The sales manager as boundary spanner. Journal of Personal Selling and Sales Management, 3,8-21.

MacCallum, R. (1998). Commentary on quantitative methods in 1-0 research. The Industrial-Organizational Psychologist, 35, 18-3 1.

MacCallum, R., & Mar, C. (1995). Distinguishing between moderator and quad- ratic effects in multiple regression. Psychological Bulletin, 118, 405-42 1.

Malmo, R. (1 959). Activation: A neuropsychological dimension. Psychological Review, 66, 367-386.

McGrath, J. (1970). A conceptual formulation for research on stress. In J. McGrath (Ed.), Social and psychological factors in stress (pp. 10-21). New York, NY Holt, Rinehart, and Winston.

Mobley, W., Horner, S., & Hollingsworth, A. (1978). An evaluation of the pre- cursors of hospital employee turnover. Journal of Applied Psychology, 63,

Motowidlo, S., Packard, J., & Manning, J. (1986). Occupational stress: Its causes and consequences for job performance. Journal of Applied Psychology, 71,

20 1-207.

408-414.

6 18-629.

1360 ZIVNUSKA ET AL.

Mowday, R., Koberg, C., & MacArthur, A. (1 984). The psychology of the with- drawal process: A cross validation test of Mobley’s intermediate linkages model of turnover in two samples. Academy of Management Journal, 27, 79-94.

Nunnally, J. (1978). Psychometric theory. New York, NY McGraw-Hill. O’Driscoll, M. I?, & Beehr, T. A. (1994). Supervisor behaviors, role stressors,

and uncertainty as predictors of personal outcomes for subordinates. Journal ojOrganizationa1 Behavior, 15, 141-155.

Rhoads, G. K., Singh, J., & Goodell, I? W. ( 1 994). The multiple dimensions of role ambiguity and their impact on psychological and behavioral outcomes of industrial salespeople. Journal of Personal Selling and Sales Management,

Rizzo, J., House, R., & Lirtzman, S. (1970). Role conflict and ambiguity in com-

Rokeach, M. (1973). The nature of human values. New York, N Y Free Press. Schabracq, M., Cooper, C., & Winnubst, J. (1996). Work and health psychology:

Toward a theoretical framework. In M. Schabracq, D. Cooper, & J. Winnubst (Eds.), Handbook of work and health psychology (pp. 3-29). Chichester, UK: John Wiley & Sons,

Schaubroeck, J., Cotton, J., & Jennings, K. (1989). Antecedents and conse- quences of role stress: A covariance structure analysis. Journal of Organiza- tional Behavior, 10, 35-58.

Schaubroeck, J., & Ganster, D. (1993). Chronic demands and responsivity to challenge. Journal of Applied Psychology, 78, 73-85.

Schwab, D., & Cummings, L. (1976). A theoretical analysis of the impact of task scope on employee performance. Academy of Management Review, I , 23-35.

Scott, W. ( 1966). Activation theory and task design. Organizational Behavior and Human Performance, I , 32-50.

Singh, J. (1998). Striking a balance in boundary-spanning positions: An investi- gation of some unconventional influences of role stressors and job character- istics on job outcomes of salespeople. Journal of Marketing, 62, 69-86.

Tatarkiewicz, W. ( 1976). Analysis of happiness. The Hague, The Netherlands: Martin Nijhoff.

Xie, J., & Johns, G. (1995). Job scope and stress: Can job scope be too high? Academy of Management Journal, 38, 1288-1 309.

Yerkes, R., & Dodson, J. (1908). The relation of strength of stimulus to rapidity of habit formation. Journal o j ~ Comparative Neurology and Psychology, 18,

14, 1-23.

plex organizations. Administrative Science Quarterly, 15, 150-1 63.

459-482.