13

Click here to load reader

Stressful Work, Psychological Job Strain, and Turnover: A 2-Year Prospective Cohort Study of Truck Drivers

Embed Size (px)

DESCRIPTION

A 2-Year Prospective Cohort Study of Truck Drivers

Citation preview

Page 1: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

Stressful Work, Psychological Job Strain, and Turnover: A 2-YearProspective Cohort Study of Truck Drivers

Einar M. de Croon and Judith K. SluiterCoronel Institute for Occupational and Environmental Health,

University of Amsterdam/Academic Medical Center,Research Institute Amsterdam Center for Health and

Health Care Research (AmCOGG), Amsterdam

Roland W. B. BlonkTNO Work and Employment, Hoofddorp

Jake P. J. BroersenSKB Center of Expertise on Health and Work, Amsterdam

Monique H. W. Frings-DresenCoronel Institute for Occupational and Environmental Health,

Research Institute Amsterdam Center for Health andHealth Care Research (AmCOGG), Amsterdam

Based on a model that combines existing organizational stress theory and job transition theory, this 2-yearlongitudinal study examined antecedents and consequences of turnover among Dutch truck drivers. Forthis purpose, self-reported data on stressful work (job demands and control), psychological strain (needfor recovery after work and fatigue), and turnover were obtained from 820 drivers in 1998 and 2000. Inagreement with the model, the results showed that strain mediates the influence of stressful work onvoluntary turnover. Also in conformity with the model, job movement to any job outside the truckingindustry (i.e., interoccupational turnover) resulted in a larger strain reduction as compared to jobmovement within the trucking industry (intraoccupational turnover). Finally, strain was found tostimulate interoccupational turnover more strongly than it stimulated intraoccupational turnover. Thesefindings provide a thorough validation of existing turnover theory and give new insights into the turnover(decision) process.

Numerous studies have investigated voluntary withdrawal fromthe organization (i.e., voluntary employee turnover). Two catego-ries of theories can be distinguished that serve as a frame for asubstantial part of these studies. The first category of theories triesto explain how stressful work results in voluntary turnover (i.e.,organizational stress theory). Theories in this category—althoughdiffering in, among other things, terminology, abstraction level,and extent to which cognitive processes are integrated—all seemto perceive turnover as the result of a two-step sequence (Kahn &Byosiere, 1992, p. 593). In the first step of this sequence, stressfulwork is believed to produce a condition of psychological strain. Inthe second step, this psychological strain causes employees todisplay several behavioral reactions among which is voluntaryturnover.

The second category of theories tries to explain how turnoveraffects the departing employee (i.e., job transition theory). Ini-tially, theories in this category regarded turnover from a negative“stress” perspective (Adams, 1976; Hayes & Hough, 1976; Wer-bel, 1983). In particular, turnover itself was seen as a potentialdisruptive and stressful event that produces strain. More recently,however, transition theory regards turnover from a positive pro-active growth perspective. According to this perspective, workerswho turnover strive to find work that fits their personality (e.g.,Clarke, 1980) and that allows active learning and growth (e.g.,Nicholson & West, 1988). From this perspective, then, turnoverreduces rather than produces strain. Indeed, the empirical research,so far, seems to support this last “positive” perspective.

To obtain a more complete picture of the turnover phenomenon,the present study examines a turnover model that combines thepreviously described turnover theories1 (see Figure 1). Accordingto this model, stressful work instigates a condition of strain. Thisstrain, in turn, invokes the worker to turnover. Next, turnoverresults in decreased strain, which, consequently, reduces the like-lihood of successive turnover (negative feedback loop). Below, theconcepts of the model are described in detail. Thereafter, a reviewof the literature that underpins the relationships between the vari-ables of the model is presented.

1 We acknowledge the helpful suggestion of an anonymous reviewer inthis respect.

Einar M. de Croon, Judith K. Sluiter, and Monique H. W. Frings-Dresen, Coronel Institute for Occupational and Environmental Health,University of Amsterdam/Academic Medical Center, Research InstituteAmsterdam Center for Health and Health Care Research (AmCOGG),Amsterdam, The Netherlands; Roland W. B. Blonk, TNO Work andEmployment, Hoofddorp, The Netherlands; Jake P. J. Broersen, SKBCenter of Expertise on Health and Work, Amsterdam, The Netherlands.

The Dutch Central Bureau of Occupational Health Care in Road Trans-port (BGZ Wegvervoer) provided partial funding for this study.

Correspondence concerning this article should be addressed to Einar M.de Croon, Coronel Institute for Occupational and Environmental Health,Academic Medical Center, Meibergdreef 9, 1105 AZ Amsterdam, TheNetherlands. E-mail: [email protected]

Journal of Applied Psychology Copyright 2004 by the American Psychological Association2004, Vol. 89, No. 3, 442–454 0021-9010/04/$12.00 DOI: 10.1037/0021-9010.89.3.442

442

Page 2: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

STRESSFUL WORK

Stressful work characteristics are defined as work conditionsthat are assumed, hypothesized, and (cumulatively) demonstratedto have certain undesirable effects—adverse physiologicalchanges, decrements in role performance, emotional tensions, on-set of physical symptoms, and so forth—in the population aboutwhich we propose to generalize (Kahn & Byosiere, 1992, p. 575).This study examines two categories of stressful work characteris-tics, namely job demands and job control.

Job Demands

Researchers use different formulations to describe the stimulithat are responsible for the stress response. Following Selye(1956), many researchers use the term stressor. Following Karasek(1979) and Frankenhaeuser (1991), the formulation job demands isused in this study. This formulation is used because demands canbe conceptualized in negative (i.e., stressful) and positive (i.e.,challenging) terms. In analogy with Demerouti, Bakker, Nachre-iner, and Schaufeli (2001), job demands are defined in this studyas physical, social, or organizational aspects of the job that requiresustained physical or mental effort. This definition implies that theexposure to demands is associated with certain psychological andpsychophysiological costs (e.g., fatigue, increased catecholaminesecretion). Psychological job demands, such as role overload(Kahn, Wolfe, Quinn, Snoek, & Rosenthal, 1964), are a centralconstruct of the demand-control theory of organizational stress(Karasek, 1979; Karasek & Theorell, 1990). Numerous demand-control studies have demonstrated that psychological job demandspredict psychological strain (see review in Van der Doef & Maes,1999). Demands have also been found to predict physiologicalstress reactions (e.g., Aronsson & Rissler, 1998; Evans & Carrere,1991) and behavioral stress reactions such as sickness absence(Kristensen, 1991; Vahtera, Kivimaki, Pentti, & Theorell, 2000).

Job Control

Job control or autonomy plays a central role in many organiza-tional stress theories (see review in Frese, 1989; Ganster, 1989).Hackman and Oldham (1975) defined the construct as “the degreeto which the job provides substantial freedom, independence, anddiscretion to the individual in scheduling the work and in deter-mining the procedures to be used in carrying it out” (p. 162). Theinfluence of job control on strain may be threefold (e.g., Frese,1989). First, job control may influence the extent to which objec-tive job demands are perceived as stressful (i.e., control resultingin the perception of demands as less stressful; Spector, 1998).

Second, control may attenuate the effect of demands on strain(e.g., Tetrick & Larocco, 1987). In particular, high control maylead to a less threatening secondary appraisal of job demands anda healthier coping style. Third, the lack of job control may frustratethe intrinsic need to be competent and feel competent in interactingwith the working environment (White, 1959), thereby bringingabout strain in a direct manner.

Theoretically, one may argue that the interaction between jobcontrol and job demands is most important for the prediction ofstrain and subsequent turnover. Research, however, has providedmost support for the main effect of job control on strain in apsychological (see review in Van der Doef & Maes, 1999), phys-iological (e.g., Steptoe, 2001), as well as behavioral respects(Kivimaki et al., 1997; North, Syme, Feeney, Shipley, & Marmot,1996). Therefore, the present study tests the main effect of jobcontrol on strain and turnover only.

PSYCHOLOGICAL JOB STRAIN

Psychological job strain is defined as aversive and potentiallyharmful psychological reactions of the individual to stressful work.This study examined two important aspects of psychologicalstrain: need for recovery after work and fatigue. In the followingdiscussion, these concepts and their relation with other strains andstressful work are described.

Need for Recovery After Work

Building on biopsychosocial organizational stress research(Frankenhaeuser, 1991; Levi, 1981), Meijman, Mulder, Van Dor-molen, and Cremer (1992) introduced the concept of need forrecovery after work (i.e., the need to recuperate from work-induced fatigue after the working day). The need-for-recoveryconcept is related to emotional exhaustion (Schaufeli & VanDierendonck, 1999; De Vries, Michielsen, & Van Heck, 2003),psychological distress (Jansen, Kant, & Van den Brandt, 2002),and physiological stress reactions (Sluiter, Van der Beek, &Frings-Dresen, 2001) and resembles the concept of postwork irri-tability (Glass & Singer, 1972). Recently, Sluiter, De Croon,Meijman, and Frings-Dresen (2003) showed that elevated need forrecovery after work is a consequence of stressful work in terms ofdemands and control.

Fatigue

Employees who display increased occupationally induced needfor recovery after work must exert additional effort (i.e., increased

Figure 1. Stress-to-turnover-to-strain model.

443STRESSFUL WORK, JOB STRAIN, AND TURNOVER

Page 3: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

psychophysiological activity) the next working day to cope withthe demands of the job. When job demands are chronically high,this process starts a vicious circle in which high need for recoveryafter work requires additional effort the next day, which, in turn,results in elevated need for recovery after the working day. Thiscycle may result in prolonged fatigue, which resembles Maslach’s(1982) emotional exhaustion (being overextended and exhaustedby one’s work). In contrast to short-term work-related fatigue (i.e.,need for recovery after work), prolonged fatigue is not easilyreversible in the short term and is task-nonspecific. It manifestsitself in inefficient action patterns; declining interest, involvement,and commitment; reduced concentration and motivation; and neg-ative emotions (Meijman & Schaufeli, 1996). Epidemiologicalresearch has demonstrated strong relations between fatigue on theone hand and psychological distress and emotional exhaustion onthe other hand (Beurskens et al., 2000). Moreover, longitudinalresearch has found stressful work in terms of demands and controlto predict the onset of fatigue (Bultmann, Kant, Van den Brandt, &Kasl, 2002).

EMPLOYEE TURNOVER

Research predominantly reflects on employee turnover as anindistinct behavior, defined as voluntary withdrawal from theorganization. However, as noted by Jackofsky (1984), severaltypes of voluntary turnover can be distinguished. One relevantturnover distinction relates to the destination of the new job. In thisrespect, Wright and Bonett (1992) distinguished between intraoc-cupational and interoccupational turnover. Intraoccupational turn-over involves job movement within the relevant occupationalgrouping (i.e., truck driving). Interoccupational turnover concernsmovement to any job external to the targeted occupational group-ing (i.e., outside truck driving). The distinctive nature of intra-versus interoccupational turnover was demonstrated by Wright andBonnet who found in their study of social welfare workers that,compared to intraoccupational turnover, interoccupational turn-over results in more drastic changes in work characteristics andcoinciding psychological strain.

LITERATURE REVIEW

Stressful Work as an Antecedent of Voluntary Turnover

Although some studies (e.g., Spector & Michaels, 1986) failedto demonstrate statistically significant effects of stressful work onturnover, research supports the hypothesis according to whichstressful work is an antecedent of voluntary turnover (Bloom,Alexander, & Nichols, 1992; McFadden & Demetriou, 1993; Bal-four & Neff, 1993; Todd & Deery-Schmitt, 1996; Arnold &Mackenzie Davey, 1999). Balfour and Neff (1993), for example,showed that high psychological job demands predicted subsequentturnover in nurses. Likewise, Todd and Deery-Schmitt (1996) andArnold and Mackenzie Davey (1999) found high psychological jobdemands to predict voluntary turnover among hospital workers.

Psychological Job Strain as an Antecedent of VoluntaryTurnover

With the exception of some research (e.g., Manlove & Guzell,1997), the majority of studies investigating the influence of strain

on turnover have demonstrated that employees with elevated strainare more likely to voluntarily resign from their organization thanemployees with low strain (Drake & Yadama, 1996; Todd &Deery-Schmitt, 1996; Wright & Cropanzano, 1998). Drake andYadama (1996), for instance, demonstrated that social welfareworkers who reported high levels of emotional exhaustion had anincreased chance to quit their job. Similarly, Todd and Deery-Schmitt evidenced that child-care workers experiencing emotionalexhaustion were more likely to turnover than their nonexhaustedco-workers. Finally, Wright and Cropanzano (1998) found emo-tional exhaustion to predict subsequent turnover in health careworkers.

Psychological Job Strain as a Mediator Between StressfulWork and Voluntary Turnover

To our knowledge, until now, no single study has examined thepresumed mediating role of psychological job strain betweenstressful work and actual turnover. Numerous studies, though,investigated the presumed mediating role of psychological strainbetween stressful work and the intent to turnover (e.g., Bedeian &Armenakis, 1981; Kemery, Mossholder, & Bedeian, 1987;Klenke-Hamel & Mathieu, 1990; Lang, Wittig-Berman, & Riz-kalla, 1992; Moore, 2000; Taris, Schreurs, & Van Iersel-VanSilfhout, 2001; Taylor, Daniel, Leith, & Burke, 1990). Thesestudies largely provide support for the two-step sequence modelaccording to which stressful work affects turnover via strain. Forinstance, Moore (2000) showed that exhaustion partially mediatedthe “effect” of working conditions on the intent to leave. In likemanner, Taris et al. (2001) found stressful work to predict de-creased turnover intention through elevated emotional exhaustion.

Voluntary Turnover as an Antecedent of PsychologicalJob Strain

Research examining the separate effect of intraoccupational andinteroccupational voluntary turnover on psychological strain isrestricted. In fact, the single study investigating these effects is thepreviously described study by Wright and Bonett (1992). Severalstudies (Newton & Keenan, 1990; Van der Velde & Feij, 1995;Swaen, Kant, Van Amelsfoort, & Beurskens, 2002), however,have examined the effect of general turnover on strain. Thesestudies consistently showed that voluntary turnover is associatedwith positive changes for the departing employee. Newton andKeenan (1990), for instance, demonstrated that young engineersreported decreases in strain after their job change. Similarly,Swaen, Kant, Van Amelsfoort, and Beurskens (2002), in a pro-spective cohort study among 12,140 Dutch employees, found thatthe job changers reported decreased psychological strain after thejob change.

CONCLUSION

Taken together, the existing literature seems to support theproposed turnover model. Two central aspects of the model, how-ever, have not been examined sufficiently. The first aspect relatesto the presumed mediating role of strain between stressful workand turnover. Past research, investigating this widely acceptedmediation paradigm, used a cross-sectional design and an attitu-

444 DE CROON ET AL.

Page 4: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

dinal criterion (i.e., turnover intention). Consequently, to somedegree, findings of these cross-sectional studies could be explainedby common method variance.

The second aspect of the model that has not been examinedsufficiently involves the distinction between interoccupational andintraoccupational voluntary turnover. As noted already, researchhas tended to treat voluntary turnover as an indistinct behavior. Forsome occupations, however, it may be necessary to distinguishbetween intra- and interoccupational turnover for a more completepicture of the consequences of voluntary turnover. Truck driving isexemplary in this respect as it is characterized by high psycholog-ical demands (Hedberg, Jacobsson, Janlert, & Langendoen, 1993;e.g., long working hours and just-in-time deliveries; Hamelin,2001), high physical demands (e.g., manually [un]loading thetruck; Van der Beek, Frings-Dresen, Kemper, & Van Dijk, 1993),and high supervisor demands (i.e., authoritative style of supervis-ing; Orris et al., 1997) the stressful nature of truck driving isreflected in elevated psychological strain (e.g., Kuiper, Van derBeek, & Meijman, 1998). The high level of work-induced psycho-logical strain among truck drivers suggests that only interoccupa-tional turnover (i.e., leaving the trade) may reduce the drivers’strain.

PRESENT INVESTIGATION

This study was designed to overcome shortcomings of thepreviously reviewed research and, by these means, to provide acomplete test of the turnover model outlined heretofore. In partic-ular, using a behavioral criterion (i.e., actual voluntary turnover)and a longitudinal design, this study examined (a) the presumedmediating role of strain between stressful work and turnover, and(b) the separate effect of interoccupational and intraoccupationalturnover on strain. In accordance with existing organizationalstress theory, it was expected that strain mediates the effect ofstressful work (i.e., demands and control) on turnover. Further-more, given the stressful nature of truck driving, a significantnegative feedback loop from interoccupational voluntary turnoverto strain only was expected.

Method

Participants

In September 1998, a random sample of truck drivers was taken from thedirectory of the Dutch Central Bureau of Occupational Health Care in RoadTransport (N � 2,000). Of the 2,000 mailed self-administered question-naires, 1,225 were returned. Two years later, in September 2000, 102 ofthese 1,225 drivers were not traceable. Therefore, 1,123 (1,225 - 102)drivers received a second questionnaire, which 820 participants returned(72%). Nonresponse analyses revealed two significant differences in 1998between nonrespondents in 2000 (n � 303) and all respondents in 2000(n � 820). Compared with the respondents, nonrespondents were youngerin 1998 (mean age 38 years vs. 40 years, t(1121) � �3.41, p �.01) and hada lower tenure in 1998 (mean tenure 8 years vs. 10 years, t(1121) � �4.00,p �.01). Because of retirement, dismissal, and incomplete questionnairesfor this study, 137 respondents at follow-up were excluded from the presentanalyses.

Measures

Stressful Work

We measured stressful work (demands and control) in 1998 and 2000using scales from the Dutch Questionnaire on the Experience and Assess-ment of Work (VBBA)2 (Van Veldhoven & Meijman, 1994; Van Veld-hoven & Broersen, 2003). Items of the Job Demand scales were derivedlargely from the Job Content Questionnaire (Karasek, 1985). Items of theJob Control scale were derived from the focused measure of timing andmethod control developed by Jackson, Wall, Martin, and Davids (1993;e.g., Wall, Jackson, & Mullarkey, 1995; Wall, Jackson, Mullarkey, &Parker, 1996). Dutch researchers in the field of organizational and occu-pational psychology frequently use these VBBA scales (e.g., Geurts, Rutte,& Peeters, 1999; Jansen, Van Amelsfoort, Kristensen, Van den Brandt, &Kant, 2003; Janssen, 2000; Meijman & Kompier, 1998;Van Yperen &Snijders, 2000). Van Veldhoven (1996) evidenced positive psychometricqualities of the Job Demands and Job Control scales. During constructionof the scales, rho, comparable to Cronbach’s alpha, varied between .89(psychological demands) and .90 (job control). Loevinger’s H variedbetween .42 (job control) and .58 (supervisor demands). These values areconsistent with underlying unidimensional constructs.

Job demands. Three job demands were assessed in 1998 and 2000 bymeans of VBBA questionnaire scales, namely, psychological job demands,physical job demands, and supervisor job demands. An 11-item scaleassessed psychological job demands. A 7-item scale assessed physical jobdemands. Supervisor job demands were measured by means of the Rela-tionship With Your Immediate Boss scale of the VBBA comprising 9items. All items of the three job demand scales were scored on a 4-pointscale (1 � never, 2 � sometimes, 3 � often, 4 � always). Job Demandsscale scores were calculated by adding the individual’s scores on therespective items. All scale scores were transformed into scales rangingfrom 0 to 100. High scores on the Job Demands scale reflect a negativeassessment.

Job control. An 11-item scale quantified job control in 1998 and 2000.Items were scored on a 4-point scale (1 � never, 2 � sometimes, 3 � often,4 � always). A Job Control scale score was calculated by adding theindividual’s scores on the items. Subsequently, this scale score was trans-formed into a scale ranging from 0 to 100. Low scores on the Job Controlscale are indicative of a negative evaluation.

Psychological Job Strain

Need for recovery after work. Need for recovery after work wasmeasured by means of the similarly named scale of the Dutch VBBAquestionnaire (Van Veldhoven & Meijman, 1994; Van Veldhoven &Broersen, 2003). The Need for Recovery After Work scale comprises 11dichotomous items. Typical items of this scale are: “At the end of aworking day I am really feeling worn-out” and “I find it hard to relax at theend of a working day.” The Need for Recovery After Work scale containsone reversed item that is recoded. A Need for Recovery After Work scalescore is calculated by adding the individual’s scores on the 11 (recoded)items. This scale score is transformed into a scale ranging from 0 to 100.Higher scores indicate a higher degree of need for recovery after work.Need for recovery after work is strongly related to emotional exhaustion(Schaufeli & Van Dierendonck, 1999). During construction of the scale(Van Veldhoven, 1996), rho and Loevinger’s H were .88 and .50,respectively.

Fatigue. Fatigue was measured by means of the Checklist IndividualStrength (CIS) (Vercoulen et al., 1994). The CIS consists of 20 statements

2 The Dutch questionnaire and the English translation are available fromthe SKB Center of Expertise on Health and Work, Amsterdam.([email protected]).

445STRESSFUL WORK, JOB STRAIN, AND TURNOVER

Page 5: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

for which the person has to indicate on a 7-point scale to what extent theparticular statement applies to him or her (1 � yes, that is true; to 7 � no,that is not true). The statements refer to four fatigue aspects: (a) subjectivefatigue (e.g., I feel tired), (b) reduced motivation (e.g., I feel no desire todo anything), (c) reduced activity (e.g., I don’t do much during the day),and (d) reduced concentration (e.g., My thoughts easily wander). The CISis well validated within the clinical setting (e.g., Vercoulen et al., 1996) andthe working situation (e.g., Beurskens et al., 2000; Bultmann et al., 2000).In this study, a composite CIS total score (ranging from 20 to 140) wascalculated by adding the individual’s scores on the four factors. Higherscores indicate a higher degree of fatigue, more concentration problems,reduced motivation, or low levels of activity.

Employee Turnover

Past researchers have defined voluntary turnover in several ways (e.g.,Wright & Bonett, 1992). This study defined turnover only as voluntaryactual withdrawal from the organization. Voluntary employee turnoveramong the participating truck drivers was measured in 2000 by means ofthe following question, “During the last 2 years (since you completed thefirst questionnaire), did you voluntarily quit your job?” Thereafter, therespondents were also asked, “How many times did you voluntarily quityour job during the last 2 years (since you completed the first question-naire)?” Subsequently, as a check on the reliability of the answers, therespondents were asked to specify the exact date of the (first) quit. Finally,to distinguish between interoccupational turnover and intraoccupationalturnover, the drivers were asked to specify their current job title. Theintraoccupational turnover group included job changers who were stillworking as a truck driver at the second measurement. The interoccupa-tional turnover group included job changers who were no longer workingas a truck driver at the second measurement.

Results

Preliminary Analyses

Of the respondents, 23% (n � 156) changed jobs voluntarilyduring the 2-year follow-up period. Of these respondents, 124(79%) changed jobs once, 23 respondents (15%) changed jobstwice, 8 respondents (5%) changed jobs three times, and 1 respon-dent changed jobs four times (1%). Furthermore, 78% of the jobchangers (n � 123) still worked as truck drivers at the second

measurement (intraoccupational turnover group). In contrast, 22%of the job changers (n � 33) had switched from the drivingprofession to a different profession (interoccupational turnovergroup). These employees found employment in the road transportindustry (e.g., transportation manager, warehouse manager) aswell as outside the road transport industry (e.g., security worker,sales worker).

Descriptive statistics, internal consistency, and intercorrelationsfor the variables under study are presented in Table 1. Inspectionof Table 1 reveals that—on a bivariate level—older drivers (r ��.16, p �.01), higher tenure drivers (r � �.25, p �.01), anddrivers with more job control (r � �08, p � .05) were less likelyto change jobs. In addition, drivers with higher physical jobdemands (r � .07, p �.05), higher supervisor job demands (r �.07, p �.05), higher need for recovery after work (r � .12, p�.01), and higher fatigue (r � .08, p �.05) were more likely tochange jobs. Furthermore, all questionnaire scales evidenced fa-vorable internal consistency (Cronbach’s alpha ranging between.88 and .92).

Mediation Tests

Plan of Analysis

The following criteria were used in testing the mediation prop-osition embedded in the turnover model (e.g., Baron & Kenny,1986, p. 1177). First, each separate stressful work characteristicmust significantly affect psychological strain (Equation 1). Sec-ond, each work characteristic and strain indicator must be shownto significantly affect employee turnover (Equation 2). Third,when these conditions hold in the predicted direction, the effect ofeach work characteristic on turnover must be less when each strainindicator (i.e., presumed mediator) is introduced into the model(Equation 3).

Need for Recovery

Table 2 presents the results of the regression analyses usingneed for recovery after work as strain indicator. In the first set of

Table 1Means, Standard Deviations, Internal Consistency, and Zero-Order Pearson Correlation Coefficients of the StudyVariables in the Cohort

Variable M SD � 1 2 3 4 5 6 7 8

1. Age (y)a 39.1 9.4 — —2. Organizational tenure (y)a 9.8 8.4 — .54** —3. Job controlb 46.4 20.4 .89 �.05 �.03 —4. Psychological job demandsb 45.0 16.0 .88 .15** .13* �.38** —5. Physical job demandsb 34.9 22.0 .88 .07* .08* �.10** .44* —6. Supervisor job demandsb 27.2 19.4 .89 .07* .10** �.37** .45** .33** —7. Need for recovery after workb 35.7 32.6 .90 .09* .04 �.32** .56** .41** .40** —8. Fatigueb 49.0 21.9 .92 .07* .03 �.31** .44** .31** .37** .71** —9. Employee turnoverc,d — — — �.16** �.25** �.08* .04 .07* .07* .12** .08*

Note. ns � 664–683 because of missing values.a y � years. b For job demands, need for recovery after work and fatigue, high scale scores reflect a negative evaluation. High scores on the job controlscale are indicative of a positive evaluation. c 0 � remained with the job during the two-year follow-up; 1 � changed jobs during the two-yearfollow-up. d Spearman rank correlation coefficients (rho).* p � .05, one-tailed. ** p � .01, one-tailed.

446 DE CROON ET AL.

Page 6: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

analyses (Equation 1), need for recovery was regressed on eachstressful work characteristic. As can be seen from Table 2, low jobcontrol (� � -.31, p � .01), high psychological demands (� � .53,p � .01), high physical demands (� � .39, p � .01), and highsupervisor demands (� � .41, p � .01) significantly predictedneed for recovery in the expected direction.

The next set of regression analyses concerned the prediction ofturnover. Because turnover is a dichotomous outcome variable,this study used logistic regression analyses to examine the rela-tions of the predictor variables to turnover. For each predictorvariable, the standardized regression coefficient and odds ratio arereported. In the present context, the odds ratio represents the ratioof the turnover probability by high job demands (above the medianvalue) or low job control (below the median value) to the turnoverprobability by low job demands (below the median value) or highjob control (above the median value). Compared with the stan-dardized regression coefficients, the odds ratio provides a means toquantify the practical importance of a relation between a predictorvariable and a dichotomous outcome. As can be seen in Table 2, in

Equation 2, when need for recovery was not included, job controlsignificantly predicted turnover (B � -.011, p � .05) (Model 1).Similarly, psychological job demands (B � .013, p � .05), phys-ical job demands (B � .010, p � .05), and supervisor job demands(B � .013, p �.05) were positively related to employee turnover.In terms of odds ratios, drivers who reported low job control were1.39 times more likely to turnover as compared to drivers whoexperienced high job control. Compared to the drivers who expe-rienced low levels of psychological, physical, and supervisor jobdemands, those who experienced high levels were 1.36, 1.67, and1.47 times more likely to change jobs, respectively.

In Equation 3, need for recovery after work was introduced intothe models. After introduction of this strain (in addition to age andorganizational tenure), all relations between the stressful workcharacteristics and turnover became nonsignificant, with the ex-ception of supervisor job demands (B � .010, p � .05). When needfor recovery was added to Equation 3, the drop in the contributionof job control and psychological, physical, and supervisor jobdemands to the prediction of turnover (in terms of standardized

Table 2Linear Regression of Need for Recovery After Work on Each Separate Stressful WorkCharacteristic (Job Control and Job Demands; Equation 1), Logistic Regression of EmployeeTurnover on Each Separate Stressful Work Characteristic (Equation 2), and Both Stressful WorkCharacteristic and Need for Recovery After Work (Equation 3) in the Cohorta

Predictor variables

Need for recovery after work Employee turnover

� B Odds Ratio B Odds Ratio

Equation 1Job control �.31**

Equation 2Job control �.011** 1.39

Equation 3Job control �.006 1.18Need for recovery .088** 1.86

Equation 1Psychological demands .53**

Equation 2Psychological demands .013** 1.36

Equation 3Psychological demands .001 1.01Need for recovery .094** 1.96

Equation 1Physical demands .39**

Equation 2Physical demands .010** 1.67

Equation 3Physical demands .004 1.46Need for recovery .089** 1.79

Equation 1Supervisor demands .41**

Equation 2Supervisor demands .013** 1.47

Equation 3Supervisor demands .010* 1.29Need for recovery .085** 1.85

Note. All equations are adjusted for age and organizational tenure.a ns � 665–674 because of missing values.* p � .05, one-tailed. ** p � .01, one-tailed.

447STRESSFUL WORK, JOB STRAIN, AND TURNOVER

Page 7: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

regression coefficients) was 54%, 92%, 60%, and 23%, respec-tively. Based on these figures, then, need for recovery seems tofully mediate the relation from psychological job demands toturnover and partially mediate the relation of the other stressfulwork characteristics to turnover.

Fatigue

Table 3 presents the results of regression analyses using fatigueas strain indicator. In Equation 1, after adjustment for age andorganizational tenure, fatigue was regressed on each stressful workcharacteristic. In Equation 1, control (� � -.30, p � .01), psycho-logical demands (� � .44, p � .01), physical demands (� � .31,p � .01), and supervisor demands (� � .38, p � .01) significantlypredicted fatigue in the expected direction. In Equation 2, turnoverwas regressed on each work characteristic. The results of theseanalyses show that job control (B � -.011, p � .05), psychologicaldemands (B � .013, p � .05), physical demands (B � .010, p �.05), and supervisor demands (B � .013, p �.05) significantlypredicted turnover. In Equation 3, after introduction of fatigue, the

standardized regression coefficients of job control, psychologicaljob demands, physical job demands, and supervisor job demandswere reduced by 36%, 54%, 40%, and 23%, respectively. Thesefigures indicate that fatigue partially mediates the relation fromthese stressful work characteristics to turnover.

Effect of Turnover on Psychological Job Strain

Table 4 displays information on need for recovery and fatigue ofthe job stayers, the intraoccupational job changers, and the interoc-cupational job changers. To examine differences between the jobchangers and the job stayers prior to the job change, we conductedone-way analyses of variance (ANOVAs). These analyses revealedsignificant differences in 1998 between the three groups on needfor recovery, F(2, 675) � 6.35, and fatigue, F(2, 668) � 5.79. Posthoc comparisons by means of the least significant difference test(LSD test) showed that, compared to the job changers, the intraoc-cupational job changers, t(675) � 2.53, p �.01, and the interoc-cupational job changers, t(675) � 2.82, p �.01, experiencedhigher levels of need for recovery after work before the job

Table 3Linear Regression of Fatigue After Work on Each Separate Stressful Work Characteristic (JobControl and Job Demands; Equation 1), Logistic Regression of Employee Turnover on EachSeparate Stressful Work Characteristic (Equation 2), and Both Stressful Work Characteristic andFatigue (Equation 3) in the Cohorta

Predictor variables

Fatigue Employee turnover

� B Odds ratio B Odds ratio

Equation 1Job control �.30**

Equation 2Job control �.011** 1.39

Equation 3Job control �.007 1.22Fatigue .011** 1.51

Equation 1Psychological demands .44**

Equation 2Psychological demands .013** 1.36

Equation 3Psychological demands .006 1.17Fatigue .011** 1.52

Equation 1Physical demands .31**

Equation 2Physical demands .010** 1.67

Equation 3Physical demands .006 1.54Fatigue .011** 1.44

Equation 1Supervisor demands .38**

Equation 2Supervisor demands .013** 1.47

Equation 3Supervisor demands .010* 1.37Fatigue .010* 1.50

Note. All equations are adjusted for age and organizational tenure.a ns � 665–674 because of missing values.* p � .05, one-tailed. ** p � .01, one-tailed.

448 DE CROON ET AL.

Page 8: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

change. The post hoc analyses further revealed that, compared tothe job stayers, only the interoccupational job changers experi-enced an elevated level of fatigue before the job change, t(668) �3.24, p � .01.

From Table 4, it can be further seen that an elevation of needfor recovery, t(515) � 3.23, p � .01, and fatigue, t(502) � 5.38,p � .01, occurred between 1998 and 2000 among the jobstayers. Among the interoccupational changers, oppositechanges over time in need for recovery, t(32) � �2.95, p � .01,and fatigue, t(31) � �2.89, p � .01, were discerned. Finally,the autocorrelations of need for recovery and fatigue in thestayers (r � .66 and r � .57, respectively) and in the intraoc-cupational changers (r � .43 and r � .62, respectively) weremoderately high. Contrarily, the autocorrelations in the interoc-cupational changers for need for recovery (r � -.04) and fatigue(r � .11) are small.

ANOVAs were used to test whether changes in psychological

strain over time had been different across the stayers, the intraoc-cupational changers, and the interoccupational changers. Differ-ence scores of need for recovery and fatigue between baseline andfollow-up were used in these analyses as an indicator of changeover time. In all the ANOVAs, post hoc pairwise comparisonsbetween the groups were performed by means of the LSD test.These analyses revealed significant differences across the threegroups with respect to changes over time in need for recovery, F(2,703) � 13.20, p �.001, and fatigue, F(2, 689) � 12.75, p �.001.Post hoc comparisons demonstrated a significant difference be-tween the interoccupational changers and the job stayers on needfor recovery after work, t(580) � 5.30, p �.001, and fatigue,t(567) � 4.94, p �.001, in the expected direction (see Figure 2 andFigure 3). The interoccupational changers also significantly dif-fered from the intraoccupational changers on need for recoveryafter work, t(153) � 2.95, p �.05, and fatigue, t(151) � 3.66, p�.001. At last, a significant difference was observed between the

Table 4Means, Standard Deviations, Paired t Values, Autocorrelations (Stability) of Need for RecoveryAfter Work and Fatigue in 1998 and 2000 for the Job Stayers, the Intraoccupational JobChangers, and the Job Changers

1998 2000

Paired ta Autocorrelation nM SD M SD

Job stayersNeed for recovery 33.4 31.7 37.4 34.7 3.23** .66 516Fatigue 47.9 21.1 52.8 22.9 5.38** .57 503

Intraoccupational job changersNeed for recovery 41.57 34.06 40.32 35.78 �0.37 .43 122Fatigue 50.97 23.68 52.43 23.97 0.77 .62 121

Interoccupational job changersNeed for recovery 48.04 35.36 23.88 29.70 �2.95** �.04 33Fatigue 58.80 25.93 43.56 17.90 �2.89** .11 32

a Paired t value is an indicator of the difference between the mean value in 1998 and the mean value in 2000 ofthe same variable of the same respondents.** p � .01, one-tailed.

Figure 2. Need for recovery after work at the first measurement (initial job) and at the second measurement(new job) in the job stayers, the intraoccupational job changers, and the interoccupational job changers.

449STRESSFUL WORK, JOB STRAIN, AND TURNOVER

Page 9: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

intraoccupational changers and the stayers on need for recoveryafter work, t(636) � 1.73, p � .10.

Discussion

This study makes three contributions to turnover research. Thefirst contribution is an examination of a voluntary turnover modelthat integrates existing organizational stress theory (addressingturnover antecedents) and job transition theory (addressing turn-over consequences). Past turnover research has tended to examineantecedents and consequences of voluntary turnover as isolatedaspects. The second contribution of this study is that it used abehavioral criterion (i.e., actual turnover) and a longitudinal designto examine the proposed turnover model. Previous research, re-viewed earlier in this article, has used primarily an attitudinalcriterion (i.e., turnover intention) and a cross-sectional design. Thethird contribution of this study concerns the distinction it madebetween interoccupational and intraoccupational voluntary turn-over. Prior research has investigated turnover primarily as anindistinct behavior. However, theoretically, for some occupations,intra- and interoccupational turnover may have different conse-quences that are worthy of examination.

In accordance with the model’s predictions, logistic regressionanalyses found stressful work and strain to be significant predic-tors of voluntary turnover. Furthermore, in agreement with themodel’s propositions, the regression analyses demonstrated thatthe effect of stressful work on turnover decreased substantiallyafter adjustment for strain. Hence, in agreement with previouscross-sectional research, this longitudinal study supports thewidely accepted idea according to which stressful work influencesturnover via strain. However, compared to the cross-sectionalrelations observed in past research, the present longitudinal rela-tions were weak. This discrepancy suggests that, to an essentialextent, method bias is responsible for cross-sectional “effects” ofself-reported stressful work characteristics and strain on turnoverintention (e.g., Spector, 1994). The present small longitudinaleffects of stressful work and strain on turnover may be attributed

to the multifactorial nature of turnover. Turnover reviews (e.g.,Griffeth, Hom, & Gaertner, 2000) showed that various factors inthe work domain (e.g., distributive justice) and family domain(e.g., family constraints) influence turnover. Furthermore, person-ality factors may affect turnover as well. Barrick and Mount(1996), for instance, showed that conscientious and emotionallystable truck drivers, compared to their colleagues, were less likelyto change jobs.

On a more specific level, the present results can be interpretedin the light of the effort-recovery model of Meijman, Mulder, VanDormolen, and Cremer (1992) and Leiter’s process model ofburnout (Leiter, 1991, 1993). The effort-recovery model states thatrepeated occupationally induced fatigue starts a vicious circle inwhich extra effort is required every new working period to copewith the demands of the job. Consequently, more fatigue is expe-rienced after the working period. This vicious circle causes cumu-lated fatigue, which, over the longer term, may result in conditionssuch as prolonged fatigue and emotional exhaustion. In accordancewith process theory on burnout (Leiter, 1991, 1993), these condi-tions may compel employees to withdraw themselves from theworking situation. According to Leiter, this withdrawal responsemay first come about in an attitudinal adjustment (decreasedcommitment) and, thereafter, in a behavioral adjustment (i.e.,performance deterioration, absenteeism, turnover).

This study extended the stress-to-strain-to-turnover pathway byincorporating a negative feedback loop running from turnoverback to strain. In particular, based on prior research (e.g., Newton& Keenan, 1990; Wright & Bonett, 1992; Van der Velde & Feij,1995; Swaen et al., 2002), turnover was hypothesized to providethe opportunity for job improvement and, thereby, a reduction instrain. However, in view of the stressful nature of truck driving,this strain-reducing effect of turnover was expected to occur onlywhen turnover involved movement to another occupation (i.e.,outside the trucking industry). In agreement with this supposition,the job-changing truck drivers who found employment in a differ-ent branch displayed a substantial reduction in psychological

Figure 3. Fatigue at the first measurement (initial job) and at the second measurement (new job) in the jobstayers, the intraoccupational job changers, and the interoccupational job changers.

450 DE CROON ET AL.

Page 10: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

strain. The job-changing truck drivers who continued working astruck drivers, conversely, did not display this decrease in strain.The finding that turnover can be a consequence as well as anantecedent of strain suggests that job change might, to someextent, reduce the chance of subsequent turnover via decreasingstrain (the turnover 3 strain 3 turnover pathway).

It was taken for granted that truck drivers would not be cogni-zant of the fact that the experienced work-induced strain is intrin-sic to truck driving. Correspondingly, we did not expect, andinclude as a formal hypothesis, that the amount of strain experi-enced prior to the job change would be different for the intraoc-cupational and interoccupational job changers. Observation of thepresent results, however, disproved the casualness of our assump-tion. In particular, compared to the job stayers, the interoccupa-tional job changers experienced elevated fatigue before the jobchange in 1998. The intraoccupational job changers, contrarily, didnot experience more fatigue than the job stayers did. A similarpattern of results emerged when need for recovery in 1998 wascompared between the job stayers, intraoccupational job changers,and interoccupational job changers. Albeit post hoc, these findingssuggest that, to some extent, strain has a dissimilar effect on intra-and interoccupational turnover. Apparently, a comparatively highproportion of strained drivers are cognizant of the fact that theexperienced strain is intrinsic to truck driving. This awareness mayactivate the driver’s decision to search for employment in anotheroccupational grouping.3 Drivers who change jobs within the truck-ing industry may be more likely to turnover for different reasons,for instance, pay dissatisfaction or fear of becoming unemployed.

The dissimilar effect of strain on the two turnover types and thedifferential effect of the two turnover types on subsequent changesin strain suggest that the discrimination between intra- and interoc-cupational turnover may give a more precise representation of theturnover (decision) process. Accordingly, in our opinion, thisturnover differentiation provides an interesting avenue for futureresearch to examine the turnover (decision) process.

Another interesting finding was the effect of organizationaltenure on turnover (r � -.25). Compared to prior research (seemeta-analytical review in Griffeth et al, 2000), this effect wasstrong. Possibly, low-tenure (inexperienced) drivers are psycho-logically unfit for the stressful and strenuous work of truck driving(i.e., healthy worker survivor effect). Alternatively, these driversstill may have to learn about their new work role before there is anadequate person–work environment fit (Edwards, 1991). The con-sequential strain in these drivers might have paved the way toturnover. Additionally, turnover may have been more difficult forthe high-tenure drivers because of restricted alternative employ-ment availability and family concerns (Johns, 1991). Also, thehigh-tenure drivers in this study may have collected more side betsor investments with the organization, which made it more difficultto change jobs (Meyer & Allen, 1984).

Three methodological aspects of this study should be taken intoaccount to appreciate the implications of the results. The firstaspect involves the interpretation of the effects in terms of fullversus partial mediation. According to the standard, strain fullymediates the stress-to-turnover relation when this relation is nolonger significant after addition of strain into the equation. Whenthe relation remains significant but a drop in the contribution ofstressful work to the prediction of turnover occurs, strain is said topartially mediate this relation. Considering our large sample size

this standard may not give a true picture of the mediation effectsin this study. In particular, although strain may account for asubstantial part of the relation from the work characteristics toturnover, these relations may remain statistically significant be-cause of our large sample size. Therefore, we based our interpre-tation of the mediation effects on the reduction of the magnitude ofthe stress-to-turnover relation after the addition of strain into theequation. On the basis of this criterion, we concluded thatstrain—in terms of need for recovery—fully mediates the relationfrom psychological job demands to turnover and that it partiallymediates the relation from the other work characteristics toturnover.

Presumably, the effect of job control, physical job demands, andsupervisor job demands on turnover is also mediated by jobsatisfaction, which plays a central role in turnover (decision)models (Dipboye, Smith, & Howell, 1994). In fact, previous re-search (e.g., Michaels & Spector, 1982) has established the medi-ating role of satisfaction between stressful work and turnover.Unfortunately, a validated measure of job satisfaction was notincluded in this study. Consequently, the possible (confounding)role of job satisfaction in the stress-to-turnover process could notbe examined. Future longitudinal research is required to elucidatethe combined roles of strain and job satisfaction in the stress-to-turnover process.

The second methodological aspect that requires considerationconcerns the response that, despite the publicity given to theresearch and three repeated mailings, was 61% at the first mea-surement. Unfortunately, we could not retrieve the characteristicsof the nonrespondents at the first measurement. However, in ouropinion, nonresponse did not have a strong influence on theresearch findings. First, at the first measurement, no significantdifferences in work characteristics and strain between nonrespon-dents and respondents at the second measurement were observed.Second, although nonrespondents at the second measurement av-eraged a moderately lower tenure and age, we adjusted the anal-yses for these potential confounders. Third, in a questionnairestudy similar to ours, Van der Beek, Frings-Dresen, Kemper, andVan Dijk (1993) performed a telephone investigation among the150 nonresponding Dutch truck drivers. Of the 150 nonrespondingdrivers, 102 drivers (68%) participated in this investigation. Theinvestigation did not reveal meaningful differences between thenonrespondents and respondents.

The third methodological aspect that warrants comment is thecomposition of the study sample, which was taken randomly fromthe directory of the Dutch Central Bureau of Occupational HealthCare in Road Transport (BGZ Wegvervoer). At that time, the BGZWegvervoer directory comprised about 5,000 road transport com-panies with approximately 100,000 employees of whom the ma-jority are truck drivers. Consequently, we can state with goodreason that the study sample was representative of the entirepopulation of Dutch truck drivers. Moreover, presumably due tothe representativeness of the study sample, the variation in thestudy variables was high. Nevertheless, effect sizes of turnoverpredictors vary across occupations (Griffeth et al., 2000), whichindicates that prudence is required when the present results are

3 We would like to thank an anonymous reviewer for this insightfulcomment.

451STRESSFUL WORK, JOB STRAIN, AND TURNOVER

Page 11: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

generalized to other occupations. As noted previously, work-induced fatigue among truck drivers can have consequences forroad safety, which may compel these workers in particular to findemployment elsewhere. Furthermore, the trucking industry regu-larly faces driver shortages (Keller, 2002), indicating that alterna-tive employment in this industry is highly available. In conformitywith turnover decision models (e.g., Mobley, 1977), this mayactivate quit decisions of strained or dissatisfied drivers.

The results of this study have methodological as well as prac-tical implications. Methodologically, the results indicate that lon-gitudinal organizational stress research may not give a completepicture of the relation between stressful work on the one hand andhealth and well-being on the other hand. Specifically, a selectioneffect may occur among employees in prospective cohort researchthat is comparable with the previously described healthy workersurvivor effect. This effect is brought about by a self-selectionprocess that allows comparatively healthy employees to remain incertain jobs, whereas those who change jobs (and occupation) areless healthy (Rothman, 1986). Similarly, the attrition of strainedemployees may restrict the range of health complaints over time,thereby attenuating prospective relations between stressful workand health complaints.

Practically, the results of this study suggest that managers maybe encouraged to know that an improvement in working conditionsmay simultaneously decrease psychological strain, which, in turn,may reduce turnover. At the individual level, managers and occu-pational health service practitioners may prevent voluntary turn-over by means of extending the psychological resources of em-ployees with elevated psychological strain. For instance, theseemployees may be given the opportunity to attend stress manage-ment courses. When there is no prospect of improvement, thesepractitioners may encourage strained employees to search forsuitable jobs in another occupational setting.

References

Adams, J. (1976). Stress and the risk of illness. In J. Adams, J. Hayes, &B. Hopson (Eds.), Transition. London: Robertson.

Arnold, J., & Mackenzie Davey, K. (1999). Graduates’ work experiencesas predictors of organizational commitment, intention to leave andturnover: Which experience really matters? Applied Psychology, 48,211–238.

Aronsson, G., & Rissler, A. (1998). Psychophysiological stress reactions infemale and male urban bus drivers. Journal of Occupational HealthPsychology, 3, 122–129.

Balfour, D. L., & Neff, D. M. (1993). Predicting and managing turnover inhuman service agencies: a case study of an organization in crisis. PublicPersonnel Management, 22, 473–486.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variabledistinction in social psychological research: conceptual, strategic, andstatistical considerations. Journal of Personality and Social Psychology,51, 1173–1182.

Barrick, M. R., & Mount, M. K. (1996). Effects of impression managementand self-deception on the predictive validity of personality constructs.Journal of Applied Psychology, 81, 261–272.

Bedeian, A. G., & Armenakis, A. A. (1981). A path-analytic study of theconsequences of role conflict and ambiguity. Academy of ManagementJournal, 24, 417–424.

Beurskens, A. J. H. M., Bultmann, U., Kant, I. J., Vercoulen, J. H. M. M.,Bleijenberg, G., & Swaen, G. M. H. (2000). Fatigue among workingpeople: Validity of a questionnaire measure. Occupational and Environ-mental Medicine, 57, 353–357.

Bloom, J. R., Alexander, J. A., & Nichols, B. A. (1992). The effect of thesocial organization of work on the voluntary turnover rate of hospitalnurses in the United States. Social Science and Medicine, 34, 1413–1424.

Bultmann, U., De Vries, M., Beurskens, A. J. H. M., Bleijenberg, G.,Vercoulen, J. H. M. M., & Kant, I. J. (2000). Measurement of fatigue inthe working population: Determination of a cutoff point for the checklistindividual strength. Journal of Occupational Health Psychology, 5,411–416.

Bultmann, U., Kant, I. J., Van den Brandt, P. A., & Kasl, S. V. (2002).Psychosocial work characteristics as risk factors for the onset of fatigueand psychological distress: Prospective results from the Maastricht co-hort study. Psychological Medicine, 32, 333–345.

Clarke, L. (1980). Occupational choice: A critical review of research in theUK London HMSO, Careers Service Branch, Department of Employ-ment. London: Department for Work and Pensions.

Demerouti, E., Bakker, A. B., Nachreiner, F., & Schaufeli, W. B. (2001).The Job Demands-Resources Model of Burnout. Journal of AppliedPsychology, 86, 499–512.

De Vries, J., Michielsen, H. J., & Van Heck, G. L. (2003). Assessment offatigue among working people: a comparison of six questionnaires.Occupational and Environmental Medicine, 60, i10–i15.

Dipboye, R. L., Smith, C. S., & Howell, W. C. (1994). Job attitudes. InR. L. Dipboye, C. S. Smith, & W. C. Howell (Eds.), Understandingindustrial and organizational psychology (pp. 139–178). Orlando, FL:Harcourt Brace College Publishers.

Drake, B., & Yadama, G. N. (1996). A structural equation model ofburnout and job exit among child protective service workers. SocialWork Research, 20, 179–189.

Edwards, J. R. (1991). Person-job fit, A conceptual integration, literaturereview, and methodological critique. In C. L. Cooper & I. T. Robertson(Eds.), International Review of Industrial and Organizational Psychol-ogy (pp. 283–357). New York: Wiley.

Evans, G. W., & Carrere, S. (1991). Traffic congestion, perceived control,and psychophysiological stress among urban bus drivers. Journal ofApplied Psychology, 76, 658–663.

Frankenhaeuser, M. (1991). A biopsychosocial approach to work lifeissues. In J. V. Johnson & G. Johansson (Eds.), The psychosocial workenvironment (pp. 49–60). Amityville, NJ: Baywood.

Frese, M. (1989). Theoretical models of control and health. In S. L. Sauter,J. J. Hurrell, & C. L. Cooper (Eds.), Job control and worker health.Chichester, England: Wiley.

Ganster, D. C. (1989). Worker control and well-being. In S. L. Sauter &C. L. Cooper (Eds.), Job control and worker health (pp. 3–23). Chich-ester, England: Wiley.

Geurts, S., Rutte, C. G., & Peeters M. (1999). Antecedents and conse-quences of work–home interference among medical residents. SocialScience and Medicine, 48, 1135–1148.

Glass, D. C., & Singer, J. E. (1972). Urban stress: Experiments on noiseand social stressors. New York: Academic Press.

Griffeth, R. W., Hom, P. W., & Gaertner, S. (2000). A meta-analysis ofantecedents and correlates of employee turnover: Update, moderatortests, and research implications for the next millennium. Journal ofManagement, 26, 463–488.

Hackman, J. R., & Oldham, G. R. (1975). The development of the JobDiagnostic Survey. Journal of Applied Psychology, 60, 159–170.

Hamelin, P. (2001). Professional drivers’ working time as a factor offlexibility and competitiveness in road haulage. Newsletter of the Euro-pean Trade Union Technical Bureau for Health and Safety. Brussels,Belgium.

Hayes, J., & Hough, P. (1976). Career transitions a source of identitystrain. In J. Adams, J. Hayes, & B. Hopson (Eds), Transition.London: Robertson.

Hedberg, G. E., Jacobsson, K. A., Janlert, U., & Langendoen, S. M. (1993).

452 DE CROON ET AL.

Page 12: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

Risk indicators of ischemic heart disease among male professionaldrivers in Sweden. Scandinavian Journal of Work Environment &Health, 19, 326–333.

Jackofsky, E. F. (1984). Turnover and job performance: An integratedprocess mode. Academy of Management Review, 9, 74–83.

Jackson, P. R., Wall, T. D., Martin, R., & Davids, K. (1993). Newmeasures of job control, cognitive demand, and production responsibil-ity. Journal of Applied Psychology, 78, 753–762.

Jansen, N. W. H., Kant, I. J., & Van den Brandt, P. A. (2002). Need forrecovery in the working population: Description and associations withfatigue and psychological distress. International Journal of BehavioralMedicine, 9, 322–340.

Jansen, N. W. H., van Amelsvoort, L. G. P. M., Kristensen, T. S., Van denBrandt, P. A., & Kant, I. J. (2003). Work schedules and fatigue: Aprospective cohort study. Occupational and Environmental Medicine,60, i47–i53.

Janssen, O. (2000). Job demands, perceptions of effort-reward fairness andinnovative work behaviour. Journal of Occupational and Organiza-tional Psychology, 73, 287–302.

Johns, G. (1991). Substantive and methodological constraints on behaviorand attitudes in organizational research. Organizational Behavior andHuman Decision Processes, 49, 80–104.

Kahn, R. L., & Byosiere, P. (1992). Stress in organizations. In M. D.Dunnette & L. M. Hough (Eds.), Handbook of industrial and organiza-tional psychology: Vol. 3 (pp. 571–650). Palo Alto, CA: ConsultingPsychologists Press.

Kahn, R., Wolfe, D., Quinn, R., Snoek, J., & Rosenthal, R. (1964).Organizational stress: studies in role conflict and ambiguity. New York:Wiley.

Karasek, R. A. (1979). Job demands, job decision latitude, and mentalstrain: implications for job redesign. Administrative Science Quarterly,24, 285–309.

Karasek, R. A. (1985). Job Content Questionnaire and user’s guide.Lowell: University of Massachusetts, Department of WorkEnvironment.

Karasek, R. A., & Theorell, T. (1990). Healthy work: Stress, productivityand the reconstruction of working life. New York: Basic Books.

Keller, S. B. (2002). Driver relationships with customers and driver turn-over: key mediating variables affecting driver performance in the field.Journal of Business Logistics, 23, 39–64.

Kemery, E. R., Mossholder, K. W., & Bedeian, A. G. (1987). Role stress,physical symptomatology, and turnover intentions: A causal analysis ofthree alternative specifications. Journal of Occupational Behaviour, 8,11–23.

Kivimaki, M., Vahtera, J., Thomson, L., Griffiths, A., Cox, T., & Pentti, J.(1997). Psychosocial factors predicting employee sickness absence dur-ing economic decline. Journal of Applied Psychology, 82, 858–872.

Klenke-Hamel, K. E., & Mathieu, J. E. (1990). Role strains, tension, andjob satisfaction: Influences on employees’ propensity to leave: A multi-sample replication and extension. Human Relations, 43, 791–807.

Kristensen, T. S. (1991). Sickness absence and work strain among Danishslaughterhouse workers: An analysis of absence from work regarded ascoping behaviour. Social Science and Medicine, 32, 15–27.

Kuiper, J. I., Van der Beek, A. J., & Meijman, T. F. (1998). Psychosomaticcomplaints and unwinding of sympathoadrenal activation after work.Stress Medicine, 14, 7-12.

Lang, D., Wittig-Berman, U., & Rizkalla, A. (1992). The influences of rolestress, physical symptoms, and job satisfaction on turnover intentions: Atwo-sample test of a modified Bedeian and Armenakis model. Journal ofSocial Behavior and Personality, 7, 555–568.

Leiter, M. P. (1991). Coping patterns as predictors of burnout: The functionof control and escapist coping patterns. Journal of Organizational Be-havior, 12, 123–144.

Leiter, M. P. (1993). Burnout as a developmental process: Consideration of

models. In W. B. Schaufeli, C. Maslach, & T. Marek (Eds.), Profes-sional burnout: Recent developments in theory and research (pp. 237–250). Washington, DC: Taylor & Francis.

Levi, L. (1981). Preventing work stress. Reading, MA: Addison-Wesley.Manlove, E. E., & Guzell, J. R. (1997) Intention to leave, anticipated

reasons for leaving, and 12-month turnover of child care center staff.Early Childhood Research Quarterly, 12, 145–167.

Maslach, C. (1982). Understanding burnout: Definitional issues in analyz-ing a complex phenomenon. In W. S. Paine (Ed.), Job stress and burnout(pp. 29–40). Beverly Hills, CA: Sage.

McFadden, M., & Demetriou, E. (1993). The role of immediate workenvironment factors in the turnover process: a systematic intervention.Applied Psychology: An International Review, 42, 97–115.

Meijman, T. F., & Kompier, M. A. J. (1998). Bussy business: How urbanbus drivers cope with time pressure, passengers, and traffic safety.Journal of Occupational Health Psychology, 31, 109–121.

Meijman, T. F., Mulder, G., Van Dormolen, M., & Cremer, R. (1992).Workload of driving examiners: A psychophysiological field study. InH. Kragt (Ed.), Enhancing industrial performance (pp. 245–258). Lon-don: Taylor & Francis.

Meijman, T. F., & Schaufeli, W. (1996). Psychische vermoeidheid enarbeid: ontwikkelingen in de A&O Psychologie [Fatigue at work: de-velopments in the I & O Psychology]. De Psycholoog, 3, 236–241.

Meyer, J. P., & Allen, N. J. (1984). Testing the side-bet theory of organi-zational commitment. Journal of Applied Psychology, 69, 372–378.

Michaels, C. E., & Spector, P. E. (1982). Causes of employee turnover: Atest of the Mobley, Griffeth, Hand and Meglino Model. Journal ofApplied Psychology, 67, 53–59.

Mobley, W. H. (1977). Intermediate linkages in the relationship betweenjob satisfaction and employee turnover. Journal of Applied Psychology,62, 237–240.

Moore, J. E. (2000). Why is this happening? A causal attribution approachto work exhaustion consequences. Academy of Management Review, 25,335–349.

Newton, T. J., & Keenan, A. (1990). Consequences of changing employersamongst young engineers. Journal of Occupational Psychology, 63,113–127.

Nicholson, N., & West, M. (1988). Managerial job change. Men andwomen in transition. Cambridge, England: Cambridge University Press.

North, F., Syme, L., Feeney, A., Shipley, M., & Marmot, M. (1996).Psychosocial work environment and sickness absence among Britishcivil servants: The Whitehall II study. American Journal of PublicHealth, 86, 332–340.

Orris, P., Hartman, D. E., Strauss, P., Anderson, R. J., Collins, J., Knopp,C., et al. (1997). Stress among package truck drivers. American Journalof Industrial Medicine, 31, 202–210.

Rothman, K. J. (1986). Objectives of epidemiological study design. In K. J.Rothman (Ed.), Modern epidemiology (pp. 77–97). Boston: LittleBrown.

Schaufeli, W. B., & Van Dierendonck, D. (1999). UBOS - De UtrechtseBurnout Schaal Handleiding [UBOS - The Utrecht Burnout Scale Man-ual]. Lisse, The Netherlands: Swets & Zeitlinger.

Selye, H. (1956). The stress of life. New York: McGraw-Hill.Sluiter, J. K., De Croon, E. M., Meijman, T. F., & Frings-Dresen, M. H. W.

(2003). Need for recovery from work-related fatigue and its role in thedevelopment and prediction of subjective health complaints. Occupa-tional and Environmental Medicine, 60, i62–i70.

Sluiter, J. K., Van der Beek, A. J., & Frings-Dresen, M. H. W. (2001). Therelation between work-induced neuroendocrine reactivity and recovery,subjective need for recovery, and health status. Journal of Psychoso-matic Research, 50, 29–37.

Spector, P. E. (1994). Using self-report questionnaires in OB research: Acomment on the use of a controversial method. Journal of Organiza-tional Behavior, 15, 385–392.

453STRESSFUL WORK, JOB STRAIN, AND TURNOVER

Page 13: Stressful Work, Psychological Job Strain, and Turnover: A 2-Year  Prospective Cohort Study of Truck Drivers

Spector, P. E. (1998). A control theory of the job stress process. In C. L.Cooper (Ed.), Theories of organizational stress (pp. 153–169). NewYork: Oxford University Press.

Spector, P. E., & Michaels, C. E., (1986). Personality and employeewithdrawal: Effects of locus of control on turnover. PsychologicalReports, 59, 63–66.

Steptoe, A. (2001). Job control, perceptions of control, and cardiovascularactivity. An analysis of ambulatory measures collected over the workingday. Journal of Psychosomatic Research, 50, 57–63.

Swaen, G. M. H., Kant, I. J., Van Amelsfoort, L. G. P. M., & Beurskens,A. J. H. M. (2002). Job mobility, its determinants, and its effects:Longitudinal data from the Maastricht Cohort Study. Journal of Occu-pational Health Psychology, 7, 121–129.

Taris, T. W., Schreurs, P. J. G., & Van Iersel-Van Silfhout, I. J. (2001). Jobstress, job strain, and psychological withdrawal among Dutch Universitystaff: Towards a dual-process model for the effects of occupationalstress. Work and Stress, 15, 283–296.

Taylor, A. H., Daniel, J. V., Leith, L., & Burke, R. J. (1990). Perceivedstress, psychological burnout and paths to turnover intentions amongsport officials. Applied Sport Psychology, 2, 84–97.

Tetrick, L. E., & Larocco, J. M. (1987). Understanding, prediction, andcontrol as moderators of the relationships between perceived stress,satisfaction, and psychological well-being. Journal of Applied Psychol-ogy, 72, 538–543.

Todd, C. M., & Deery-Schmidt, D. M. (1996). Factors affecting turnoveramong child care providers: A longitudinal study. Early ChildhoodResearch Quarterly, 11, 351–376.

Vahtera, J., Kivimaki, M., Pentti, J., & Theorell, T. (2000). Effect ofchange in the psychosocial work environment on sickness absence: Aseven year follow up of initially healthy employees. Journal of Epide-miology and Community Health, 54, 484–493.

Van der Beek, A. J., Frings-Dresen, M. H. W., van Dijk, F. J. H., Kemper,H. C. G., & Meijman, T. F. (1993). Loading and unloading by lorrydrivers and musculoskeletal complaints. International Journal of Indus-trial Ergonomics, 12, 13–23.

Van der Doef, M., & Maes, S. (1999). The job demand-control (-support)model and psychological well being: A review of 20 years of empiricalresearch. Work and Stress, 13, 87–114.

Van der Velde, M. E. G., & Feij, J. A. (1995). Change of work perceptionsand work outcomes as a result of voluntary and involuntary job change.Journal of Occupational and Organizational Psychology, 68, 273–290.

Van Veldhoven, M. J. P. M. (1996). Psychosociale arbeidsbelasting en

werkstress [Psychosocial workload and occupational stress]. Lisse, TheNetherlands: Swets & Zeitlinger.

Van Veldhoven, M. J. P. M., & Broersen, J. P. J. (2003). Measurementquality and validity of the “need for recovery scale.” Occupational andEnvironmental Medicine, 60, i3–i9.

Van Veldhoven, M. J. P. M., & Meijman, T. F. (1994). Het meten vanpsychosociale arbeidsbelasting met een vragenlijst: de VragenlijstBeleving en Beoordeling van de Arbeid (VBBA) [The Dutch Question-naire on the Experience and Assessment of Work]. Amsterdam: NIA.

Van Yperen, N. W., & Snijders, T. A. B. (2000). A multilevel analysis ofthe demands-control model: Is stress at work determined by factors atthe group level or the individual level. Journal of Occupational HealthPsychology, 5, 182–190.

Vercoulen, J. H. M. M., Hommes, O. R., Swanink CM, Jongen, P. J.,Fennis, J. F. M., Galama, J. M. D., Van der Meer, J. W. M., &Bleijenberg, G. (1996). The measurement of fatigue in patients withmultiple sclerosis. A multidimensional comparison with patients withchronic fatigue syndrome and healthy subjects. Archives of Neurology,53, 642–649.

Vercoulen, J. H. M. M., Swanink, C. M. A., Fennis, J. F. M., Galama,J. M. D., Van der Meer, J. W. M., & Bleijenberg, G. (1994). Dimen-sional assessment of chronic fatigue syndrome. Journal of Psychoso-matic Research, 38, 383–392.

Wall, T. D., Jackson, P. R., & Mullarkey, S. (1995). Further evidence onsome new measures of job control, cognitive demand and productionresponsibility. Journal of Organizational Behavior, 69, 153–166.

Wall, T. D., Jackson, P. R., Mullarkey, S., & Parker, S. K. (1996). Thedemand-control model of job strain: A more specific test. Journal ofOccupational and Organizational Psychology, 69, 153–166.

Werbel, J. D. (1983). Job change: A study of an acute job stressor. Journalof Vocational Behavior, 23, 242–250.

White, R. (1959). Motivation reconsidered: The concept of competence.Psychological Review, 66, 297–333.

Wright, T. A., & Bonett, D. G. (1992). The effect of turnover on worksatisfaction and mental health: Support for a situational perspective.Journal of Organizational Behavior, 13, 603–615.

Wright, T. A., & Cropanzano, R. (1998). Emotional exhaustion as apredictor of job performance and voluntary turnover. Journal of AppliedPsychology, 83, 486–493.

Received October 10, 2002Revision received June 6, 2003

Accepted July 10, 2003 �

454 DE CROON ET AL.