19

Click here to load reader

A daily diary study of coping in the context of the job demands–control–support model

Embed Size (px)

Citation preview

Page 1: A daily diary study of coping in the context of the job demands–control–support model

Journal of Vocational Behavior 66 (2005) 219–237

www.elsevier.com/locate/jvb

A daily diary study of coping in the contextof the job demands–control–support model

Kevin Danielsa,¤, Claire Harrisb

a Business School, Loughborough University, Leicestershire LE11 3TU, UKb Manchester School Of Management, UMIST, UK

Received 8 July 2004Available online 15 December 2004

Abstract

We examined one of the processes thought to underpin Karasek and Theorell’s jobdemands–control–support model (1990). This is that control and support accentuate betterwell-being by fostering problem-focused coping with work demands. We also examinedwhether other forms of coping implemented through control and support are related to indica-tors of well-being. In a daily diary study of 29 workers in a public hospital, we found higherlevels of subsequent goal attainment were associated with problem-focused coping imple-mented by executing control. Problem-focused coping implemented by eliciting support wasassociated with higher concurrent and subsequent levels of pleasant aVect. We found otherforms of coping were also associated with aVect and goal attainment. The results have implica-tions for elaboration of the DCS model. 2004 Elsevier Inc. All rights reserved.

Keywords: Demands–control–support model; Coping; AVect; Goals; Well-being

* Corresponding author. Fax: +44 1509 223962.E-mail address: [email protected] (K. Daniels).

0001-8791/$ - see front matter 2004 Elsevier Inc. All rights reserved.doi:10.1016/j.jvb.2004.10.004

Page 2: A daily diary study of coping in the context of the job demands–control–support model

220 K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237

1. Introduction

Karasek and Theorell’s job demands–control–support model (DCS, 1990) is oneof best known, most widely cited, and reviewed models of well-being and work (e.g.,de Lange, Taris, Kompier, Houtman, & Bongers, 2003). An assumed process under-pinning the model concerns how work control and social support foster active copingwith work demands that, in turn, enhances well-being, health, and productivity. Thecore predictions of the model have been tested mainly by examining statistical inter-actions between demands, control, and support. However, the evidence with respectto these interactions is mixed (de Lange et al., 2003).

These results might indicate that in some circumstances control and support facil-itate eVective coping, but in others, control and support might encourage ineVectiveor even harmful coping responses (Daniels, 1999; Ito & Brotheridge, 2003). The wayin which the DCS has usually been tested implies that mere availability of control orsupport translates into better well-being. Individuals may have a more agentic role,by attempting to execute control or elicit support to facilitate coping with workdemands (de Jonge & Dormann, 2002). However, individuals’ purposes for enactingcontrol or support, and the results of so doing, have rarely been studied in relation tothe DCS. In this paper, we investigate such issues.

2. Job demands–control–support model

First conceived as the job demands–control model (Karasek, 1979), the centralelement was that decision latitude combined with job demands to predict well-beingand health. Demands were deWned as psychological demands such as high workpace,time pressures, and diYcult work. Decision latitude was deWned as comprising of theextent of authority to make decisions concerning the job and the breadth of skillsneeded to perform the job. Many researchers interpreted the notion of a joint eVectbetween demands and decision latitude as a statistical interaction, although manyattempts have failed to Wnd evidence for this interaction (de Lange et al., 2003). How-ever, one methodological reWnement indicates it is the job control component thatbuVers the impact of work demands on well-being (Wall, Jackson, Mullarkey, &Parker, 1996). Another development of the demands–control model has been toextend the model to include social support as a second moderator of work demands.Support is deWned in this context as the levels of helpful social interaction with super-visors and co-workers (Karasek & Theorell, 1990, p. 69). Here too, the evidence withrespect to the buVering eVects of control and support on demands is also mixed (deLange et al., 2003).

Karasek (1989, p. 143) has commented that focusing on statistical interactions ‘isnot the main issue’ and that the practical implications of main eVects models, wherelow control, low support, and high demands are associated with poor well-being, arethe same as buVering models. However, whilst focusing on statistical interactions fortheir own sake might be distracting, simply assessing job demands, control, and sup-port might lead attention away from the issue of exactly how control and support

Page 3: A daily diary study of coping in the context of the job demands–control–support model

K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237 221

enhance well-being. Karasek and Theorell (1990) consider that control, in particular(p. 36), and support (pp. 69–70) both promote well-being and productivity throughfostering active, problem-solving as a means of coping with work demands. In devel-oping the role of individual agency, de Jonge and Dormann (2002) suggest that it isthe execution of control or elicitation of support, rather than there mere availability,that are critical to coping with work demands. Practically, these suggestions mightimply that control and support must be provided in such a way that their enactmentpromotes work-based problem-solving.

There is some support for the idea that job control and social support facilitateactive-problem-focused coping. Ito and Brotheridge (2003) found that co-workersupport and job autonomy were associated with a dimension of active-coping labeledpositive orientation. Co-worker support was also related to seeking advice, assis-tance, and working harder as means of coping. Control and support have also beenfound to interact with problem-focused coping and demands, so that control andsupport can bolster the eVects of problem-focused coping on well-being (Daniels,1999; de Rijk, Le Blanc, Schaufeli, & de Jonge, 1998). However, in some circum-stances, both control and support might make work demands more, rather than lessharmful, to well-being (Buunk & Hoorens, 1992; Kaufman & Beehr, 1986; Mullar-key, Jackson, Wall, Wilson, & Grey-Taylor, 1997; Sargent & Terry, 1998). Togetherwith those studies that indicate no buVering eVects, these studies indicate that controland support might also facilitate less eVective, or even harmful, coping patterns(Daniels, 1999; Ito & Brotheridge, 2003).

Therefore, a more precise examination of the DCS model would be to examine theforms of coping implemented by executing control or eliciting social support inresponse to work demands, and their relations to well-being. If indicators of well-being are associated with control executed to solve problems or support elicited tosolve problems, then the role of active-problem-focused coping, as implied in theDCS model, is supported. If indicators of well-being are associated with control exe-cuted for other coping strategies or support elicited for other coping strategies, thenthere is evidence for how the DCS might be further elaborated.

3. Coping and the DCS model

When faced with threats to well-being, such as high job demands, people evaluatethe options available for coping (Lazarus & Folkman, 1984). This entails determiningthe target of coping and how that target can be achieved. For example, the target ofcoping might be to remove the source of the threat to well-being, and restructuringtasks to avoid the threat might be the means to achieving that target. Coping, then,consists of several components including: coping function—the goal of coping; andcoping behavior, the way function comes to be (Lazarus, 1999). We discuss copingfunction and behavior in turn below.

Coping function is what an individual is trying to achieve through coping. It hasbeen conceived as comprising major functions of problem-focused, emotion-focused,appraisal-focused, and avoidance coping (Lazarus & Folkman, 1984; Skinner, Edge,

Page 4: A daily diary study of coping in the context of the job demands–control–support model

222 K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237

Altman, & Sherwood, 2003). Problem-focused coping corresponds to managing oraltering the source of the problem through, for example, making and executing plansto deal with problems. Emotion-focused coping concerns eVorts to regulate or con-trol emotional distress. Appraisal-focused coping concerns eVorts to change thoughtsabout stressful circumstances, for example by changing expectations or redeWningissues as unimportant. Avoidance coping is a cognitive strategy that entails orientingaway from thoughts and feelings associated with stressful circumstances.

Coping behavior is what people actually do to perform coping functions. In theDCS model, for example, coping behavior is the execution of control or elicitation ofsocial support. That is control and support embedded in the work environment areused to cope with the demands of the job (de Jonge & Dormann, 2002). It is impor-tant to diVerentiate coping behavior from coping function, since it is possible for thesame behavior to fulWll more than one function, perhaps simultaneously (Skinner etal., 2003). For example, one might elicit social support to distract oneself from a situ-ation (avoidance function), help regulate emotions (emotion-focused function), helpchange the interpretation of the situation (appraisal-focused function), or help gener-ate resources for tackling the problem (problem-focused function).

The eVectiveness of each coping function for protecting well-being is determinedby context and the nature of stressful encounters (Lazarus, 1999). In the context ofexecuting control or eliciting support to cope with job demands, problem-focusedcoping is thought to be an eVective function (Karasek & Theorell, 1990). There is evi-dence for this (Daniels, 1999; de Rijk et al., 1998; Ito & Brotheridge, 2003). The eVec-tiveness of other coping functions with respect to job demands is unclear. There isevidence that all of the other functions might be associated with better well-being insome circumstances (Daniels, 1999), but other evidence indicates that avoidance, inparticular, is associated with worse well-being (Ito & Brotheridge, 2003).

4. The present study

The aims of this study are twofold. First, we wanted to examine whether betterwell-being was associated with problem-focused coping implemented in relation towork demands by executing control over work (Hypothesis 1) or eliciting support atwork (Hypothesis 2). These are expectations derived from the implication in the DCSmodel that active-problem-focused coping with work demands is one of the majorpsychological processes that links control and support to well-being. Second, wewanted to examine whether well-being was associated with executing control or elic-iting support in order to use other coping functions, namely emotion-focused,appraisal-focused, and avoidance coping. We had no expectations with respect tothese other forms of coping.

We used two kinds of dependent variable to assess well-being. It is reasonable toassume that one goal of coping is to reduce the impact of work demands on pleasantaVect, and the experience of pleasant aVect is taken as an important component ofwell-being (Daniels, 2000). Coping might also be directed at attaining personal workgoals (Daniels, Harris, & Briner, 2004), and well-being also comprises of attainment

Page 5: A daily diary study of coping in the context of the job demands–control–support model

K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237 223

of major personal goals such as experiencing competence and relatedness (RyV &Keyes, 1995). Therefore, we assessed the impact of coping on aVect and on attain-ment of work goals. We assessed aVect and goals at the end of each working daywhen we assessed coping. However, we also used three indicators assessed on theworking day following our assessment of coping: aVect before work, aVect afterwork, and goals after work. Introducing a lag between assessment of independentand dependent variables establishes temporal precedence, thus strengthening causalinference (Tennen & AZeck, 2002). Also, because the eVectiveness of coping can varywith the time course of a stressor (Lazarus, 1999), coping might have instantaneousand delayed inXuences on well-being.

In assessing coping, we were careful to diVerentiate coping function from copingbehavior. We considered the two coping behaviors implied in the DCS: executingcontrol and eliciting support at work (rather than the availability of control and sup-port, de Jonge & Dormann, 2002). We ensured that assessment of coping behaviorand function were assessed in relation to work demands. We also ensured that assess-ment of each coping function was diVerentiated according to whether control or sup-port was used for that function. In this way, we were able to be conWdent our Wndingsrelated speciWcally to work demands, rather than other workplace stressors, and thecoping functions were related to control and support—rather than other copingresources.

We used a daily diary methodology. Diary studies can capture data in context andclose to changes in well-being, demands, coping behaviors, and coping functions(Bolger, Davis, & Rafaeli, 2003). This enhances the ecological validity of the Wndings,as well as minimizing recall biases and other biases in respect of coping and othervariables (e.g., Todd, Tennen, Carney, Armeli, & AZeck, 2004). Diary methods canalso enable stronger inferences about causality in several ways (Tennen & AZeck,2002). Controls can be included for previous levels of dependent variables, temporalvariation in dependent variables, and stable factors associated with the individual.As noted above, incorporating time lags between assessment of independent anddependent variables can help establish the temporal precedence of the independentvariables.

5. Method

5.1. Design, participants, and procedure

Data were collected with a diary protocol, in which participants recorded entriesbefore and after work, on days when they were in work, over a consecutive period oftwo working weeks. Before commencement of the diary period, participants alsocompleted a questionnaire. Participants were employees who worked in a HumanResource Department of a United Kingdom public hospital. Participants wererecruited through a presentation by the authors to staV in the department. Thirty-three out of 34 members of the department subsequently completed daily diaryassessments.

Page 6: A daily diary study of coping in the context of the job demands–control–support model

224 K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237

Participants reported that they were present at work on 272 occasions out of apossible 330 (i.e., n D 33 £ 10 days of the study). Diary entries for each day weredeleted from the analyses if they were not fully completed. One participant provideddata on most days of the study, but had a large number of missing data within eachentry. Three participants only provided one complete diary entry, and omitted toprovide any data on most days of the study. For these four participants, data weredeleted for all analyses. The minimum number of days of complete data was four forthe remainder of the sample (two participants). Complete diary entries were availablefor 221 out of 258 possible entries from 29 of the original 33 participants.

In the Wnal sample of 29, 25 participants were female. The average age of the sam-ple was 35.7 years (SD D 9.9). The majority of participants described their ethnicity as‘White British’ (28). Five of the participants described themselves as secretaries orclerical assistants; 16 as human resource assistants or oYcers; and eight as managersor directors. Four worked part-time (mean hours for full-time workers D 37.4,SD D .3; mean hours for part-time D 24.6, SD D 6.2). Participants had been workingin their current job for an average of 2.3 years (SD D 2.5) and had been working forthe hospital for an average of 5.0 years (SD D 5.3). There were no signiWcant diVer-ences on demographic data or trait aVect between those retained in the Wnal sampleand those deleted.

On the Wednesday before the start of diary data collection, participants weregiven a presentation on how to complete the diary and also required to complete ashort questionnaire. The participants were then given a diary booklet in which torecord their responses twice daily, before work and after work for two workingweeks. It was emphasized in the presentation that it was important participantsadhere to the protocol and to complete diary entries before work and after Wnishingwork. Participants were instructed not to complete an entry retrospectively if theyhad forgotten to do so at the required time. Participants were instructed to start com-pleting their diary on a speciWed date (the Monday following the presentation). Atthe end of the two working weeks, participants sealed their diaries in envelopes, andthese were collected by the second author.

To enhance compliance, a number of strategies were used (Bolger et al., 2003).First, there was a lengthy recruitment process to secure commitment from partici-pants. This included participants and one of the authors counter-signing agreementsconcerning data conWdentiality, the right to withdraw from the research at any time,and other ethical issues in research. Participants were instructed of the importanceof completing the diary as requested, and that if they had forgotten to complete anentry they were to leave the section and continue with the rest of the schedule asrequested. Participants were also instructed that their responses were conWdential tothe research team, and that they were to keep the diary in a safe place. Diary book-lets were small (A5 sized) and bound, and the day of the study and when partici-pants were expected to complete discrete sections were given at the top of every pageof the diary booklet. Participants were given both authors’ contact details if theyhad any questions during the study, and the second author also visited the partici-pants personally to check compliance and other issues during several days of thestudy.

Page 7: A daily diary study of coping in the context of the job demands–control–support model

K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237 225

5.2. Measures

Trait aVect was assessed by questionnaire before commencement of the diaryaspect of data collection. In the before-work section of the diary, measures weretaken of aVect before work. In the after-work section of the diary, an assessment wasmade of daily demands, aVect at work for that day, goal attainment for that day, andthe extent to which control was executed and support elicited in order to implementproblem-focused, emotion-focused, appraisal-focused, and avoidance coping.

5.2.1. Trait aVectParticipants indicated how they normally felt at work using 15 items (e.g., happy,

worried, annoyed, gloomy, active) found to be valid indicators of the major elementsof aVective well-being at work (Daniels, 2000). Each adjective was rated on a six-pointscale (1 D ‘Never’ to 6 D ‘All of the time’). Items were coded such that high scoresindicated trait pleasant aVect, summed, and divided by 15 to give an overall score.

5.2.2. AVect before workA 10-item scale was used (e.g., ‘at ease,’ ‘anxious,’ ‘gloomy,’ and ‘motivated’). The

items were selected also from aVects found to be valid indicators of aVective well-being at work (Daniels, 2000). Participants were asked to indicate how they felt atthat moment in time on a scale of 1–6 (1 D ‘not at all’ to 6 D ‘very much’). Items werescored so that high scores indicated pleasant aVect, summed, and divided by 10 togive an overall score. AVect before work was used as a control variable in analyseswith that day’s coping and as a dependent variable for the previous day’s coping.

5.2.3. Daily demandsAfter work, participants rated daily demands on a two-item scale with a six-point

response format (Daniels, Harris, & Briner, 2002). These items were how often theyexperienced high demands at work (‘every minute’ ‘to not at all’) and for how longhigh demands at work lasted during the course of that day (‘all of the day’ to ‘not atall’). Participants were instructed that high demands could include a ‘heavy work-load; working to deadlines; having tasks interrupted; or having to perform multipletasks.’ The items were summed and then divided by two to give an overall rating ofdemands. Higher ratings indicated greater incidence of demands.

5.2.4. Daily copingAfter work, participants were asked to indicate whether they had executed control

or elicited support to deal with high work demands on that day. Participants wereable to check both options, check just one of the two options, or check neither. Exe-cuting control was operationalized as changing the way individuals worked. Exam-ples given to participants in instructions for completing this section wererescheduling activities, changing work methods, or changing work objectives. Elicit-ing support was operationalized as talking to other people at work. Examples givenin the instructions were talking to others to seek advice, conWde in someone, or to askfor help with tasks.

Page 8: A daily diary study of coping in the context of the job demands–control–support model

226 K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237

If a participant indicated s/he changed the way s/he worked in order to deal withwork demands, s/he was then asked to rate nine items assessing why s/he had done so.The nine items corresponded each to one of the four coping functions assessed. Twoof these were for problem-focused coping (e.g., ‘to solve the problem’); two for emo-tion-focused coping (e.g., ‘to get your emotions oV your chest’), three for appraisal-focused coping (e.g., ‘to change your expectations of the problem’); and two foravoidance coping, that reXected the cognitive purpose of this function (e.g., ‘to turnyour attention away from the problem’). Many of the nine items were adapted fromEdwards and Baglioni’s measure (1993), and examined by us in a pilot study. If a par-ticipant indicated s/he had talked to other people at work, s/he was then asked to ratethe same nine items, except this time assessing why s/he had talked to other people atwork. Ratings were made on a Wve-point Likert-type scale (1 D ‘not at all,’ 5 ‘to alarge extent’). These ratings were then recoded as 0 to 4. An item was also scored zeroif the corresponding behavior was not endorsed. To form scale scores for each behav-ior for each participant on each day, items were summed.

5.2.5. Daily aVectDaily aVect was assessed after work with the same 10 items and response format

used in the before-work section. However, in this instance, participants were asked torate how work had made them feel that day. Same day and following day ratings ofdaily aVect were used as dependent variables.

5.2.6. Daily goal attainmentA Wve-item scale that assessed daily attainment of major personal goals was used

(Harris, Daniels, & Briner, 2003). Example items were ‘got on with people at work’and ‘felt a sense of purpose at work.’ Items were rated on a four-point Likert-typescale (1 D ‘not at all,’ 4 D ‘all day’). The items were summed and divided by Wve. Sameday and following day ratings of goal attainment were used as dependent variables.

5.3. ConWrmatory factor analyses of coping scales

To examine the structure of the coping scales, conWrmatory factor analyses(CFAs) were carried out on the diary data (number of observations D 221 from 29people). The CFAs were used to investigate the extent to which the supposed mea-surement models could account for correlations between items. Because the observa-tions contained within- and between-person variance and the data were skewed,standard goodness of Wt statistics were unsuitable. Instead, we used Satorra and Ben-tler’s (1994) robust methods, which were appropriate because we did not wish to dis-aggregate the variance into its within- and between-person components. In all cases,factor models were identiWed by Wxing the variance of factors to 1, analyses were con-ducted on the covariance matrices, and all factors were allowed to correlate.

An eight-factor model was Wtted to the data. The eight factors corresponded to thefour coping functions implemented by exercising control and the four coping func-tions implemented by eliciting support. The analysis indicated acceptable Wt for thismodel (Satorra–Bentler rescaled �2 (df 107) D 138.69, p < .05, CFI D .99). All factor

Page 9: A daily diary study of coping in the context of the job demands–control–support model

K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237 227

loadings factors were signiWcant in the hypothesized direction (p < .000001). A four-factor model, corresponding to the four coping functions only, did not Wt the datawell (Satorra–Bentler rescaled �2 (df 129) D 906.75, p < .00001, CFI D .63). A two-fac-tor model comprising of control and support factors also did not Wt the data well(Satorra–Bentler rescaled �2 (df 134) D 603.70, p < .00001, CFI D .78).

5.4. Data analysis

We used hierarchical linear modeling with the HLM5 package (Raudenbush,Bryk, Cheong, & Congdon, 2000). This is suitable for regression analysis with multi-level diary data. For the analyses, there were Wve dependent variables: same day aVectafter work, next day aVect before work, next day aVect after work, same day dailygoal attainment, and next day goal attainment. Introducing a time lag in the analysesstrengthens causal inference. However, this also meant some data were excluded inanalyses with time lags. For aVect and goal attainment measures taken concurrentlywith coping assessments, 221 observations were available from the 29 participants forthe analyses. Some 161 daily observations were available from the 29 participants foranalyses with next day aVect before work as the dependent variable. In the analysesinvolving following day ratings of aVect after work and goal attainment as dependentvariables, 153 observations were available.

In all analyses, trait aVect, work demands, day of the week (coded as dummy vari-ables), and week of data collection were controlled. For analyses involving aVect andsame day goal attainment as dependent variables, daily aVect ratings before worktaken on the same day as coping ratings were controlled. For analyses involving nextday goal attainment, goal attainment measures taken concurrently with coping rat-ings were controlled. The person level measure of trait aVect was centered at thegrand mean for the sample. Coping ratings, daily aVect before work, goal attainment,and work demands were centered at the mean for each person, and their slopesallowed to vary between people. The dummy variables representing day and weekwere left in their raw metric and relations with dependent measures Wxed to be invari-ant across people (Hofmann & Gavin, 1998 give more information on centering deci-sions).

The coping variables were skewed. Therefore, robust statistics were used to test thesigniWcance of coping terms. This skew also meant that variance in the dependentmeasures was greater when control or support were not used than when they were.Therefore, two dummy variables were used to represent whether control was exe-cuted or not, or support elicited or not (note the coping function variables were con-tinuous). These dummy variables were then used to model heterogeneity in thevariance in the dependent measures, allowing better parameter estimates to beobtained (Snijders & Bosker, 1999).

One-tailed statistical inference tests were used on the regression slopes of problem-focused coping on the dependent variables, because of expectations from the DCSmodel. All other tests were two-tailed. We used a sequential Bonferroni procedure tocontrol for type I error across the analyses (Holm, 1979). In this procedure, testsare placed in ascending order of signiWcance within a family of tests. The smallest

Page 10: A daily diary study of coping in the context of the job demands–control–support model

228 K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237

probability is then multiplied by the number of tests in the family. The second proba-bility is then multiplied by the remaining number of tests, and so on. Tests are judgedto be signiWcant if the product is less than .05. We deWned a family of tests as the testsof each coping function for each separate coping behavior across a single dependentvariable. For example, one family of tests comprised of tests of the four coping func-tions implemented by executing control on aVect after work. Therefore, a family oftests consisted of four tests.

6. Results

Table 1 shows the means, modes, standard deviations, reliabilities, and correla-tions for the measures. Table 2 shows the results of the multilevel regressions.

For the multilevel regression of problem-focused coping on the dependent vari-ables, the results show some support for Hypothesis 1, concerning problem-focusedcoping implemented through control. One out of Wve tests revealed a statistically reli-able relationship. This was a positive relationship with next day ratings of goalattainment (p < .05). The results show stronger support for Hypothesis 2, concerningproblem-focused coping implemented by eliciting support. Here, there were two sta-tistically reliable positive relationships, with concurrent ratings of aVect after work(p < .01) and next day ratings of aVect after work (p < .05). The relationship betweennext day ratings of goal attainment and problem-focused coping implemented byeliciting support evinced some variation between people (p < .05).

Emotion-focused coping implemented by executing control was associated withless pleasant aVect the following morning (p < .05). Emotion-focused coping throughsupport was associated with lower concurrent ratings of goal attainment (p < .01).However, ratings of goal attainment made on the following day were associated withmore use of emotion-focused coping through support (p < .05). This relationship var-ied between people (p < .01). Overall, there are no associations between the dependentvariables and appraisal-focused coping, irrespective of how it was implemented.However, there was evidence for variation between people in the relationshipbetween appraisal-focused coping implemented through support and next day rat-ings of goal attainment (p < .05). Avoidance by executing control was associated withmore pleasant aVect on the following morning (p < .05). Eliciting support for avoid-ance was associated with lower concurrent ratings of goal attainment (p < .05). Therelationship between next day ratings of goal attainment and avoidance copingimplemented by eliciting support evinced some variation between people (p < .01).

7. Discussion

The results indicate that control and support can facilitate the eVectiveness ofproblem-focused and some other forms of coping. However, the results also indicatethat some forms of coping implemented through control and support can sometimesbe associated with less well-being. Further, the results indicate there can be variability

Page 11: A daily diary study of coping in the context of the job demands–control–support model

K. D

aniels, C. H

arris / Journal of Vocational B

ehavior 66 (2005) 219–237229

r these data.

6 7 8 9 10 11 12 13

—.75 —.85 .89 —.59 .77 .75 —

.40 .34 .37 .25 —

.27 .41 .29 .30 .39 —

.49 .55 .61 .45 .66 .55 —

.30 .52 .47 .60 .33 .59 .63 —

Table 1Means, standard deviations, reliabilities, and correlations

SigniWcance of correlations not shown, since standard signiWcant tests of correlations are inappropriate fo

Variable Mean Mode SD � 1 2 3 4 5

1. Trait aVect 4.26 4.00 .66 .91 —2. AVect before work 4.28 3.90 .85 .89 .52 —3. Daily demands 3.24 3.00 1.31 .92 .04 .04 —4. Daily goal attainment 2.83 3.00 .54 .77 .37 .51 .03 —5. AVect after work 4.25 5.30 .97 .91 .55 .70 ¡.11 .68 —

Daily control coping6. Problem-focused .71 0.00 1.26 .96 .11 ¡.03 .22 .11 .007. Emotion-focused .38 0.00 .87 .90 .09 ¡.06 .26 .05 ¡.068. Appraisal-focused .49 0.00 .99 .95 .08 ¡.04 .26 .12 .009. Avoidance .25 0.00 .70 .88 .07 ¡.20 .25 ¡.06 ¡.05

Daily support coping10. Problem-focused 1.74 0.00 1.50 .94 .13 ¡.08 .08 ¡.04 .0111. Emotion-focused 1.17 0.00 1.31 .95 ¡.10 ¡.19 .18 ¡.20 ¡.2912. Appraisal-focused 1.07 0.00 1.11 .90 ¡.06 ¡.24 .19 ¡.04 ¡.1613. Avoidance .53 0.00 .81 .89 ¡.23 ¡.36 .13 ¡.23 ¡.36

Page 12: A daily diary study of coping in the context of the job demands–control–support model

230K

. Daniels, C

. Harris / Journal of V

ocational Behavior 66 (2005) 219–237

ol variables are day, week, trait aVect, daily demands, results are available on request. SigniWcance tests are

Same day goal attainment

Next day goal attainment

B SE VC B SE VC

.02 .01 .00 .06* .02 .03

.00 .02 .00 .05 .04 .02*

.00 .03 .00 .01 .02 .03¡.09** .03 .00 .15* .05 .04**

.00 .02 .00 .05 .03 .00¡.03 .03 .00 .12 .05 .03*

.01 .03 .01 ¡.05 .07 .03¡.11* .04 .01 .19 .09 .14**

Table 2Summarized multilevel regressions

SE, robust standard error; VC, variance component. Each coping function set entered separately. ContraVect before work (all except next day goals) or same day goal achievement (next day goals only). Fullcorrected for type I error using a sequential Bonferroni procedure.

* p < .05.** p < .01.

Same day aVect after work

Next day aVectafter work

Next day aVect before work

B SE VC B SE VC B SE VC

Controls +Problem-focusedControl coping .04 .03 .00 .03 .05 .00 ¡.03 .02 .00Support coping .10** .03 .01 .17* .07 .03 ¡.01 .04 .01

Emotion-focusedControl coping ¡.06 .07 .02 .02 .07 .04 ¡.18* .06 .02Support coping ¡.07 .03 .00 .13 .10 .10 .07 .06 .04

Appraisal-focusedControl coping ¡.06 .08 .04 .03 .09 .06 ¡.14 .06 .01Support coping .07 .04 .01 .17 .12 .18 .06 .07 .03

AvoidanceControl coping ¡.05 .09 .05 ¡.27 .18 .01 .34* .12 .09Support coping ¡.10 .06 .01 .07 .18 .42 ¡.18 .10 .11

Page 13: A daily diary study of coping in the context of the job demands–control–support model

K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237 231

between individuals in the relationships between coping and indices of well-being.With respect to the DCS model, the results indicate support for the implied role ofactive-problem-focused coping as the process by which control and support attenu-ate the negative impact of demands on well-being, since in some instances, problem-focused coping implemented through control and support are positively associatedwith indicators of daily well-being. The instances when problem-focused coping isassociated with better daily well-being, and when other forms of coping are associ-ated with well-being, indicate some ways in which the DCS model can be elaborated.

7.1. Control, support, and problem-focused coping

There is some support for Hypothesis 1, that better well-being is associated withproblem-focused coping implemented by executing control over work in relation towork demands. It might be argued that theoretically and practically, this carries fewimplications because only one indicator of well-being out of Wve was associated withproblem-focused coping implemented through control, the signiWcance only justreached conventional levels (p < .05, one-tailed), and the regression coeYcient wasrelatively small (B D .06). However, there are several counters to this argument.Although only one test evinced the hypothesized relationship, namely the relation-ship with next day goal attainment, this relationship did involve longitudinal predic-tion with lagged levels of the dependent variable controlled, and therefore establishestemporal precedence. Together with sequential Bonferroni correction, this resultarguably provides strong support for a causal inXuence of problem-focused copingimplemented through control on subsequent goal attainment. Moreover, the non-corrected probability of achieving this result by chance was smaller than the cor-rected probability by an order of magnitude (p < .005, one-tailed).

It might also be argued that the results are conservative, because they do not takeinto account cumulative eVects of coping over time. To examine potential accumula-tion of eVects over time with the current data, we calculated the partial correlationbetween each participants’ average level of problem-focused coping implementedthrough control in the Wrst week of the study and each participants’ average level ofgoal attainment in the second week of the study, controlling for each participants’average level of goal attainment in the Wrst week. The partial correlation came tor D .35 (p < .05 one-tailed), indicating over 12% of the variance in accumulatedchanges in goal attainment from one week to the next might be accounted for byproblem-focused coping implemented by executing control in the previous week.

Our results with respect to control used for problem-focused coping raise at leasttwo issues. First, why is the result speciWc to goal attainment and not aVect? Second,why is there only a lagged association? In respect of the Wrst issue, progressingtoward and attaining goals is associated with aVect (Diener, 1984), but the relation-ships between work-related goals and aVect are moderated by several factors, such asthe importance of the goals (Harris et al., 2003). Therefore, simply solving problemsthat prevent goal attainment may not enhance pleasant aVect, unless such other fac-tors are also present. In respect of the second issue, the lack of a relationship betweengoal attainment and coping both assessed on the same day, taken together with the

Page 14: A daily diary study of coping in the context of the job demands–control–support model

232 K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237

lagged relationship, may reXect the time needed to obtain feedback on whether thecoping attempt had been successful. That is, there is not necessarily any immediatefeedback on the success or likely success of coping in attaining goals in this instance.

There were two statistically reliable relationships that oVered support for the sec-ond hypothesis. The Wndings indicated that problem-focused coping implementedthrough social support was associated with higher ratings of pleasant aVect duringthe working day on the same day as coping and also on the following day. The sec-ond of these relationships oVers the strongest support for Hypothesis 2, since thisestablishes temporal precedence of support-related problem-focused coping overpleasant aVect. Unlike the results with problem-focused coping implemented throughcontrol, there is both a contemporaneous and a lagged relationship. This contempo-raneous relationship might reXect instantaneous feedback on the likelihood of cop-ing success provided by the donor of support. Also, and again unlike the Wndingswith control, the Wndings with support are speciWc to aVect after work. These resultsmight suggest that support is a more eVective resource for problem-focused copingtargeted at demands when they cause, or are anticipated to cause, unpleasant aVect.However, support might be not as eVective a resource for solving problems withattaining personal goals.

7.2. Control, support, and other forms of coping

Executing control used for emotion-focused coping was associated with lowermomentary ratings of pleasant aVect on the morning after coping, although no otherrelationship was evident. This relationship might reXect the cognitive eVort expendedin aVect regulation (Butler et al., 2003), which might have a carry-over eVect that dis-sipates quickly. In contrast, avoidance by executing control was associated withhigher momentary ratings of pleasant aVect on the following morning—indicatingthis form of coping executed through control can have short-term palliative beneWts,perhaps because it limits exposure to demands (Miller, 1979).

Both emotion-focused coping and avoidance coping implemented by eliciting sup-port were associated with lower concurrent goal attainment. Discussing or expressingemotions with others may be thought of as unprofessional or indicative of poor per-formance (Ashforth & Humphrey, 1995), as might openly distracting oneself fromwork-related problems by talking to others. In these instances, then, provision of sup-port might lead to perceptions of personal incompetence or inability to attain per-sonal goals (Blaine, Crocker, & Major, 1995). The results indicate, however, that sucheVects might dissipate quickly. Moreover, emotion-focused coping through supportwas associated with higher ratings of goal attainment on the following day. It mightbe that using support to regulate aVect extracts less cognitive eVort than using con-trol, so that cognitive activity can be directed toward attaining goals in the longerterm. Thus, short-term impediments to goal attainment might be a consequence ofusing support as a means for longer-term resolution of goal-related problemsthrough other means.

Appraisal-focused coping had no relations with indicators of well-being, regard-less of how it was implemented. Demands might be frequently occurring and familiar

Page 15: A daily diary study of coping in the context of the job demands–control–support model

K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237 233

stressors for many people (Karasek & Theorell, 1990). In these circumstances, it ispossible that attempts to re-appraise the impact of demands are hindered by pre-existing and stable beliefs that also inXuence appraisals of current levels of demands(Daniels et al., 2004). These stable beliefs are thought to be inXuenced by long stand-ing prior exposure to stressors, and consequently diYcult to change (Daniels et al.,2004).

For eliciting social support, there was evidence of variation between people in therelationships between each coping function and goal attainment on the followingday. The Wndings might reXect the nature of support networks. There is an expecta-tion of subsequent reciprocation of support (House, 1981). The variation betweenpeople might reXect diVerences in the expected time scale of reciprocation. For exam-ple, some supportive relationships might require following day reciprocation of sup-port, and therefore direction of eVort toward providing support and away fromattaining personal goals. Other supportive relationships might not require suchimmediate reciprocation.

7.3. Limitations and future research

Diary and other event sampling methods are particularly useful for investigatingdynamic processes such as aVect and coping. However, one potential issue is non-compliance with data collection protocols. We measured many of the variables assummative assessments over the whole working day. Such summative assessmentsare less prone to problems with compliance than if we had asked just for momentaryassessments (Bolger et al., 2003, p. 594). We did not consider such summative assess-ments for many of our variables as a problem, because daily estimates made at theend of the working day have good correspondence to summated momentary ratingsmade across the day (e.g., Todd et al., 2004).

Our sample consisted of just one department in one organization. We consider thecontrol of organizational context and departmental type that enhances internal andstatistical conclusion validity, beyond that already obtained by using a diary method-ology, to outweigh the disadvantages of so doing (Cook & Campbell, 1976). This isbecause our research was concerned with examining aspects of a theoretical modelthat has already been examined in other ways in many other occupations and coun-tries. Nevertheless, our Wndings are limited in the extent to which they might general-ize to other organizational and departmental settings. Therefore, replication in othercontexts would be useful.

We also used a general measure of work demands, rather than one tailored to spe-ciWc work demands. Using a general measure facilitated comparison across diVerenttasks, between and within individuals, during the course of the study. The limitationis that a general measure might be more prone to measurement error. Using a diarymethodology, in itself, minimizes error in measurement, because measures are takenclose to when demands are experienced. Moreover, we used two other strategies tominimize error. First, in the rubric for the demands measures, we deWned demandsfor participants. Second, we used a fully anchored frequency-based rating scale inorder to minimize errors in recording (cf. Frese & Zapf, 1988).

Page 16: A daily diary study of coping in the context of the job demands–control–support model

234 K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237

Here, our questions about coping behavior concerned control that had been exe-cuted and support that had been elicited. Our questions about coping function con-cerned the reasons for using control and support. We did not address theconsequences of attempting a coping behavior that fails. It might be that attemptingto execute control for problem-focused coping when no control is available might bemore harmful than not attempting to execute control at all. Examining such pro-cesses in detail would elaborate the psychological processes underpinning the DCSmodel further. However, in this study, our analyses did account for stable diVerencesin availability of control and support, although these were not measured. This isbecause we allowed the intercepts of the regression analyses to vary between people,as is common in multilevel regression, and also we centered daily variables at themean for that person. The eVects of these processes are to separate out stablebetween-person from daily within-person variance (Hofmann & Gavin, 1998). Ouranalyses, then, examined the extent to which variations from mean levels of copingand demands for each person were related to variations from usual levels of well-being. Stable between person diVerences in availability of control and support areaccounted for by diVerences in regression intercepts that reXect stable diVerences indaily indicators of well-being between people.

We have not examined the dynamics of how one coping function, attemptedthrough support or control, might enable another, more directly eVective copingfunction, to be attempted through either support or control. Of course, it is possiblethat control through avoidance and emotion-focused coping through support canhave longer-term beneWts, if they facilitate subsequent problem-focused coping. Nev-ertheless, by adopting a daily diary design, we were able to shed some light on someof the dynamic processes by which control, support, and coping combine.

8. Conclusions

The DCS model has been traditionally tested in a mechanistic way, in which the psy-chological processes underpinning the beneWcial eVects of control and support are rarelygiven detailed theoretical or empirical attention. Nevertheless, it is implied in the DCSmodel that social support and control enable eVective problem-focused coping as meansof coping with work demands, which in turn fosters well-being. The results of this studyelaborate upon the DCS model in three main ways. First, and most generally, the resultsindicate individual agency in how control and support are used. Second, the results indi-cate that problem-focused coping implemented by executing control or support havediVerential relationships with indicators of well-being. For control, problem-focusedcoping is related to greater subsequent attainment of personal goals. In contrast, prob-lem-focused coping through support appears to be related directly to greater pleasantaVect. The third way in which these results elaborate the DCS model is by indicatingthat other forms of coping can be eVective for well-being. SpeciWcally, using control toavoid work demands appears to have a subsequent but short-term beneWt for pleasantaVect. Using support for emotion-focused coping seems to be associated with attain-ment of personal goals subsequently, although at the cost of short-term goal attainment.

Page 17: A daily diary study of coping in the context of the job demands–control–support model

K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237 235

Practically, the results indicate that job redesign interventions to increase controland support, for example through team-working, might increase their chances of suc-cess if problem-solving skills training is introduced alongside job redesign. Indeed,although evidence for the eVectiveness of job redesign interventions is mixed (Briner& Reynolds, 1999), such interventions appear to be more successful when introducedwith training (Cotton, 1993). Other interventions, such as mentoring or counselingstaV in solving demands-related problems through control or support, might also beuseful. Training managers might facilitate such individually focused interventionstargeted at their subordinates (Doherty & Tyson, 1998). Given that problem-focusedcoping through control was related to goal attainment, then well-being might beimproved by incorporating processes in developmental appraisal systems to identifyhow organizations can provide the latitude and support for individuals to align theirown goals with organizational goals, and then how to attain these goals (Armstrong,2001). Finally, the results indicate that as well as encouraging problem-solvingthrough control and support, two other forms of coping need not necessarily be dis-couraged. These are avoidance through control and emotion-focused coping throughsupport, both of which can have some beneWcial eVects. However, given norms ofrationality that pervade work organizations, the use of such coping functions in con-junction with problem-focused coping through control or support may need to beemphasized.

Acknowledgments

We thank the anonymous reviewers for their constructive and encouraging comments.

References

Ashforth, B. E., & Humphrey, R. H. (1995). Emotion in the workplace: A reappraisal. Human Relations,48, 97–125.

Armstrong, M. (2001). A handbook of human resource management practice (8th ed.). London: Kogan Page.Blaine, B., Crocker, J., & Major, B. (1995). The unintended negative consequences of sympathy for the stig-

matized. Journal of Applied Social Psychology, 25, 889–905.Bolger, N., Davis, A., & Rafaeli, E. (2003). Diary methods: Capturing life as it is lived. Annual Review of

Psychology, 54, 579–616.Briner, R. B., & Reynolds, S. (1999). The costs, beneWts, and limitations of organizational level stress inter-

ventions. Journal of Organizational Behavior, 20, 647–664.Butler, E. A., EgloV, B., Wilhelm, F. H., Smith, N. C., Erickson, E. A., & Gross, J. J. (2003). The social con-

sequences of expressive suppression. Emotion, 3, 48–67.Buunk, B. P., & Hoorens, V. (1992). Social support and stress: The role of social comparison and social

exchange processes. British Journal of Clinical Psychology, 31, 445–457.Cook, T. D., & Campbell, D. T. (1976). The design and conduct of quasi-experiments and true-experiments

in Weld settings. In M. D. Dunnette (Ed.), Handbook of industrial and organizational psychology. NewYork: Wiley.

Cotton, J. L. (1993). Employee involvement. Newbury Park: Sage.Daniels, K. (1999). Coping and the job demands–control–support model: An exploratory study. Interna-

tional Journal of Stress Management, 6, 125–144.

Page 18: A daily diary study of coping in the context of the job demands–control–support model

236 K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237

Daniels, K. (2000). Measures of Wve aspects of aVective well-being at work. Human Relations, 53, 275–294.Daniels, K., Harris, C., & Briner, R. B. (2002). Understanding the risks of stress: A cognitive approach. Sud-

bury: HSE Books.Daniels, K., Harris, C., & Briner, R. B. (2004). Linking work conditions to unpleasant aVect: Cognition,

categorisation and goals. Journal of Occupational and Organizational Psychology, 77, 343–364.de Jonge, J., & Dormann, C. (2002). The DISC model: Demand induced strain compensation mechanisms

in job stress. In M. F. Dollard, H. R. WineWeld, & A. H. WineWeld (Eds.), Occupational stress in the ser-vice professions. London: Taylor & Francis.

de Lange, A. H., Taris, T. W., Kompier, M. A. J., Houtman, I. R. D., & Bongers, P. M. (2003). “The verybest of the millennium”: Longitudinal research and the demand–control–(support) model. Journal ofOccupational Health Psychology, 8, 282–305.

de Rijk, A. E., Le Blanc, P., Schaufeli, W., & de Jonge, J. (1998). Active coping and need for control as mod-erators of the job-demand-control model: EVects on burnout. Journal of Occupational and Organiza-tional Psychology, 71, 1–18.

Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95, 542–575.Doherty, N., & Tyson, S. (1998). Mental well-being in the workplace: A resource pack for management train-

ing. Bedford, UK: Human Resource Research Centre, CranWeld School of Management.Edwards, J. R., & Baglioni, A. J. (1993). The measurement of coping with stress: Construct validity of the

ways of coping checklist and the cybernetic coping scale. Work and Stress, 7, 17–31.Frese, M., & Zapf, D. (1988). Methodological issues in the study of work stress: Objective vs subjective

measurement work stress and the question of longitudinal studies. In C. L. Cooper & R. Payne (Eds.),Causes, coping, and consequences of stress at work. Chichester: Wiley.

Harris, C., Daniels, K., & Briner, R. B. (2003). A daily diary study of goals and aVective well-being at work.Journal of Occupational and Organizational Psychology, 76, 401–410.

Hofmann, D. A., & Gavin, M. B. (1998). Centering decisions in hierarchical linear models: Implications forresearch in organizations. Journal of Management, 24, 341–623.

Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics,6, 65–70.

House, J. S. (1981). Work stress and social support. Reading, MA: Addison-Wesley.Ito, J. K., & Brotheridge, C. M. (2003). Resources, coping strategies, and emotional exhaustion: A conser-

vation of resources perspective. Journal of Vocational Behavior, 63, 490–509.Karasek, R. A. (1979). Job demands, job decision latitude, and mental strain: Implications for job redesign.

Administrative Science Quarterly, 24, 285–308.Karasek, R. A. (1989). Control in the workplace and its health-related aspects. In S. L. Sauter, J. J. Hurrell,

& C. L. Cooper (Eds.), Job control and worker health (pp. 129–159). New York: Wiley.Karasek, R. A., & Theorell, T. (1990). Healthy work. New York: Basic Books.Kaufman, G. M., & Beehr, T. A. (1986). Interactions between job stressors and social support—Some

counterintuitive results. Journal of Applied Psychology, 71, 522–526.Lazarus, R. S. (1999). Stress and emotion: A new synthesis. New York: Springer.Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer.Miller, S. M. (1979). Controllability and human stress: Method, evidence, and theory. Behavior Research

and Therapy, 17, 287–304.Mullarkey, S., Jackson, P. R., Wall, T. D., Wilson, J. R., & Grey-Taylor, S. M. (1997). The impact of tech-

nology characteristics and job control on worker mental health. Journal of Organizational Behavior, 18,471–489.

Raudenbush, S., Bryk, A., Cheong, Y. F., & Congdon, R. (2000). HLM5: Hierarchical linear and nonlinearmodeling. Lincolnwood, IL: ScientiWc Software International.

RyV, C. D., & Keyes, C. L. M. (1995). The structure of psychological well-being revisited. Journal of Per-sonality and Social Psychology, 69, 719–727.

Sargent, L. D., & Terry, D. J. (1998). The eVects of work control and job demands on employee adjustmentand work performance. Journal of Occupational and Organizational Psychology, 71, 219–236.

Satorra, A., & Bentler, P. M. (1994). Corrections to test statistics and standard errors in covariance struc-ture analysis. In A. von Eye & C. C. Clogg (Eds.), Latent variables analysis: Applications for develop-mental research (pp. 399–419). Thousand Oaks: Sage.

Page 19: A daily diary study of coping in the context of the job demands–control–support model

K. Daniels, C. Harris / Journal of Vocational Behavior 66 (2005) 219–237 237

Skinner, E. A., Edge, K., Altman, J., & Sherwood, H. (2003). Searching for the structure of coping: Areview and critique of category systems for classifying ways of coping. Psychological Bulletin, 129, 216–269.

Snijders, T., & Bosker, R. (1999). Multilevel analysis: An introduction to basic and advanced multilevel mod-eling. Thousand Oaks: Sage.

Tennen, H., & AZeck, G. (2002). The challenge of capturing daily processes at the interface of social andclinical psychology. Journal of Social and Clinical Psychology, 21, 610–627.

Todd, M., Tennen, H., Carney, M. A., Armeli, S., & AZeck, G. (2004). Do we know how we cope. Relatingdaily coping reports to global and time-limited retrospective assessments. Journal of Personality andSocial Psychology, 86, 310–319.

Wall, T. D., Jackson, P. R., Mullarkey, S., & Parker, S. (1996). The demands–control model of job strain: Amore speciWc test. Journal of Occupational and Organizational Psychology, 69, 153–166.