Scholarly article on topic 'Burnout, work engagement and workaholism among highly educated employees: Profiles, antecedents and outcomes'

Burnout, work engagement and workaholism among highly educated employees: Profiles, antecedents and outcomes Academic research paper on "Psychology"

CC BY-NC-ND
0
0
Share paper
Academic journal
Burnout Research
OECD Field of science
Keywords
{Burnout / "Work engagement" / Workaholism / "The circumplex model of subjective well-being at work" / "Highly educated employees" / "Person-oriented approach"}

Abstract of research paper on Psychology, author of scientific article — Hely Innanen, Asko Tolvanen, Katariina Salmela-Aro

Abstract The present study examined the longitudinal profiles of burnout, engagement and workaholism among highly educated employees. First, the latent profile modeling indicated two latent classes: Engaged and Exhausted-Workaholic. Second, the results revealed that employees with the Engaged profile experienced high levels of energy and dedication, whereas employees with the Exhausted-Workaholic profile experienced exhaustion, cynicism and workaholism. Social pessimism in the transition from high education to work predicted poor subjective well-being at work. Further, workaholism decreased during the career among members of the Exhausted-Workaholic profile suggesting positive direction during career. Finally, Engaged employees experienced detachment and relaxation, life satisfaction and rewards.

Academic research paper on topic "Burnout, work engagement and workaholism among highly educated employees: Profiles, antecedents and outcomes"

ti^^^M AI IIIII.E IN PRESS

Burnout Researchxxx (2014) xxx-xxx

ELSEVIER

Contents lists available at ScienceDirect

Burnout Research

journal homepage www.elsevier.com/locate/burn

Research Article

Burnout, work engagement and workaholism among highly educated employees: Profiles, antecedents and outcomes

qi Hely Innanena*, Asko Tolvanenb, Katariina Salmela-Aroa

a University ofJyväskylä, Department of Psychology, Finland b University ofJyväskylä, Methodology Centre for Human Sciences, Finland

10 11 12

20 21 22

ARTICLE INFO

Article history:

Received 14June 2013

Received in revised form 28 March 2014

Accepted 1 April 2014

Keywords: Burnout

Work engagement Workaholism

The circumplex model of subjective well-being at work Highly educated employees Person-oriented approach

ABSTRACT

The present study examined the longitudinal profiles of burnout, engagement and workaholism among highly educated employees. First, the latent profile modeling indicated two latent classes: Engaged and Exhausted-Workaholic. Second, the results revealed that employees with the Engaged profile experienced high levels of energy and dedication, whereas employees with the Exhausted-Workaholic profile experienced exhaustion, cynicism and workaholism. Social pessimism in the transition from high education to work predicted poor subjective well-being at work. Further, workaholism decreased during the career among members of the Exhausted-Workaholic profile suggesting positive direction during career. Finally, Engaged employees experienced detachment and relaxation, life satisfaction and rewards.

© 2014 he Authors. Published by Elsevier GmbH. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

24 1. Introduction

25Q3 The main aim of this longitudinal study was to identify differ-

26 ent profiles of subjective well-being at work (i.e. burnout, work

27 engagement and workaholism) among highly educated employ-

28 ees. We targeted this group of employees as their work fulfills the

29 criteria proposed by the theory of work engagement and burnout:

30 high education may increase work engagement when faced with

31 emotionally challenging work [1], and it may protect against risk

32 of burnout [2]. As our theoretical framework, we applied the cir-

33 cumplex model of subjective well-being (SWB) at work [3] (see

34 Fig. 1 ).

35 The majority of the existing research on burnout and work

36 engagement has taken a variable-oriented approach (for exam-

37 ple [4-7]). The results of such studies have indicated, for example,

38 that burnout [8], and work engagement [9] are stable over time.

39 However, a stability coefficient does not demonstrate dynamic

40 developmental processes of work-related well-being over time,

41 and it also fails to take into account inter-individual differences.

42 A person-oriented approach, in turn, assumes population hetero-

43 geneity with respect to the phenomenon of interest (here SWB at

Q2 * Corresponding author. Tel.:+358 40805 3491; fax:+358 14260 2801.

E-mail addresses: hely.innanen@jyu.fi, hely.innanen@gmail.com (H. Innanen), asko.tolvanen@psyka.jyu.fi (A. Tolvanen), katariina.salmela-aro@psyka.jyu.fi (K. Salmela-Aro).

work) and changes in it (see [10]). Such an approach may lend addi- 44

tional support to the theory of SWB at work by taking account 45

diversity in SWB on the individual level by identifying different 46

groups of employees who follow a similar pattern of develop- 47

ment of SWB over time. A few previous studies have applied a 48

person-oriented approach in the area of work and organizational 49

psychology [11-14]. The present study contributes to the literature 50

by adopting a person-oriented approach [15,16] in order to gain 51

a deeper understanding of SWB at work on the individual level. 52

Specifically, our aim was to identify the proportion of highly edu- 53

cated employees experiencing low activation and displeasure, and 54

thus at the highest risk for severe burnout. Conversely, we expected 55

to find employees who are engaged, highly activated and experi- 56

encing pleasure [3]. By adding the impact of workaholism into the 57

SWB profiles, we hoped to gain a deeper understanding of the role 58

of all the factors relating to the different profiles of SWB at work, and 59

so contribute to filling an important gap in the research literature. 60

The nature of the present study is explorative. Thus, we are cau- 61

tious about precisely formulating potential longitudinal profiles 62

or means, or possible mean changes in the SWB profiles (includ- 63

ing the variables of simultaneous burnout, work engagement and 64

workaholism) [13]. At the very least, we expected to find hetero- 65

geneity in SWB at work in a sample of highly educated employees 66

with diverse occupational and organizational backgrounds. Fur- 67

thermore, we expected to find at least one longitudinal high-level 68

work engagement profile (with simultaneous low-level burnout 69

and low workaholism) [3] and one longitudinal high-level burnout 70

http://dx.doi.org/10.1016/j.burn.2014.04.001

2213-0586/© 2014 he Authors. Published by Elsevier GmbH. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

jjj^^^M AlIIHI.E IN PRESS

H. Innanen et al. / Burnout Research xxx (2014) xxx-xxx

HIGH ACTIVATION

UNPLEASANT

LOW ACTIVATION

Fig. 1. A two dimensional view ofwork-related subjective well-being (adopted from Bakker and Oerlemans [13]). A simplified model.

profile (with simultaneous high level workaholism and low level work engagement), as examples of these have previously been found in the Finnish working population (for review see [2,17]). In accordance with the model of SWB at work [3], we expected workaholism to be higher in the profiles characterized by burnout, as employees who mainly experience negative emotions at work tend to suffer simultaneously from burnout and workaholism.

Finally, to deepen our understanding of the antecedents and outcomes of SWB at work, we examined the profiles obtained, first, with respect to achievement and social strategies in the transition from university to work, as antecedents of the emergent profiles. Second, we examined the profiles obtained with respect to recovery strategies, life satisfaction, rewards and sense of community as outcomes of the emergent profiles.

1.1. SWB at work

Traditionally, subjective well-being has referred to how an individual evaluates his or her life and well-being [18]. In line with the definition of the more recent circumplex model of SWB at work by Bakker and Oerlemans [3] (see Fig. 1) affective states of SWB at work arise along two continua: the pleasure-displeasure continuum and the high activation-low activation continuum. These affective states can be described as a linear combination of these continua, that is, of pleasure and activation [19]. Consequently, the level of simultaneous activation and pleasure defines an employee's SWB at work.

More specifically, in the circumplex model of SWB at work [3] (Fig. 1) the quadrant characterized by positive emotions and high activation is likely to result in work engagement, and the quadrant characterized by positive emotions and low activation is likely to result in job satisfaction. Further, the quadrant that is characterized by negative emotions and low activation is likely to result in burnout, and the quadrant characterized by negative emotions and high activation is likely to result in workaholism. In the present study we focused on the three quadrants (burnout, work engagement and workaholism) relevant for the present study. Although we adopted the SWB at work model [3], we were not aiming to test the model itself. Instead, we adopted the model for the purpose of examining the SWB profiles, and thus we did not include the quadrant of job satisfaction in our study. The current view of work engagement and burnout on the basis of both Finnish and international studies is that positive and negative states of SWB at work represent different phenomena which employees can nevertheless experience simultaneously [20-22].

1.1.1. Work engagement as a positive form of SWB at work

Work engagement, defined as a positive work-related state of mind [23], has been previously characterized either as the polar of burnout in the same continuum [6] or as a conceptual opposite, i.e. an independent positive construct, negatively related to burnout

[22]. To gain the best understanding of SWB at work, we examined work engagement as an independent three-dimensional positive construct (including energy, dedication, absorption), operational-ized according to the definition by Schaufeli and his colleagues [22]. Energy refers to high levels of mental energy, dedication describes the cognitive dimension of work engagement, including a sense of meaningfulness and challenge, and absorption refers to being fully focused [22] and deeply preoccupied in one's work to the extent that it is difficult to stop working. Engaged employees are enthusiastic and energetic, involved and reasonably committed to their work. They put all their physical, emotional and mental energies into their work, and therefore are capable of optimal performance and feel positive emotions at work [7,24]. As a consequence, they work long hours, but feel pleasure [25] and are not addicted to work

1.1.2. Burnout and workaholism as a negative form of SWB at work

Negative indicators of SWB at work include burnout and workaholism [3] (Fig. 1). The three-dimensional model of burnout comprises high levels of exhaustion and cynicism and a low level of professional efficacy [26]. Exhaustion refers to the stress dimension of burnout and to a lack of mental energy; cynicism, the cognitive dimension of burnout, refers to negative attitudes toward one's work and co-workers; and professional efficacy refers to one's beliefs in one's efficacy at work [27]. Recently, increased research interest has been shown in workaholism, the other negative form of work-related well-being in the circumplex model of SWB [28]. Workaholism is generally described as a tendency to work excessively hard [29,30], and is marked by being obsessed with work and unwillingness to disengage from work (i.e. psychological dependence on work) [31,32]. Workaholics have an extremely powerful desire to achieve [33,34], and hence are unable to resist a compulsive need to work [35]. In pursuit of achievement, workaholics expend a lot of energy on work without sufficient recovery [36,37]. The most obvious characteristic of workaholics is an addiction to work, meaning that they tend to work harder than is required and reject other life contexts [38]. Berglass [39] found workaholics' lack of the capacity or willingness to engage in intimacy with others (i.e. a propensity to social pessimism). Thus, focusing on work may serve as an excuse to avoid participating in social functions at work. As a consequence they have poor relationship quality [40], and poor life satisfaction [41].

1.1.3. The role of career stages in experiences of SWB at work

Career stage is of relevance in the longitudinal examination of

highly educated employees' experiences of SWB at work. According to the life-span theory of motivation, individuals experience different developmental environments and tasks depending on their age [42]. Havighurst [42], in his traditional theory of developmental tasks, proposed that coping with the tasks of a given life stage provides the basis for coping at the next life stage. Furthermore, in his theory of careers, Super [43] showed that during different life stages individuals accomplish developmental tasks as part of their career decision-making process, and that success in these is related to their experiences of SWB at work. Career stages that have been identified are exploration (age 14-24), establishment (age 25-44), maintenance (age 45-60), and disengagement (age 61 or older) [41].

Many of the transitions in an individual's life course take several years to complete. For example, a career development trajectory

H^^^M AlIIHI.E IN PRESS

H. Innanen et al. / Burnout Research xxx (2014) xxx-xxx 3

typically includes decisions concerning education and the further career (e.g., [44]). It has been suggested that to gain a comprehensive understanding of the role of career stages in SWB at work, a long follow-up period is needed (see also [14]). Applying the cir-cumplex model of SWB at work [3], we focused on the early career (i.e. establishment stage) and on the mid-career (i.e. maintenance stage). We assumed these career stages to be relevant for emerging burnout, work engagement and workaholism.

At the early career stage, young highly educated employees experience many developmental tasks, such as fitting in with the organization, learning the job, and pursuing their career goals. Simultaneously, they have other tasks, such as being a spouse and a parent [43]. Having many simultaneous tasks at this career stage may cause some young employees to experience burnout. Similarly, in pursuit of work-related achievements, some young employees may experience workaholism as a consequence of expending a lot of energy on work without sufficient recovery [36,37]. It has been indicated that focusing excessively on work is related to a low level of SWB at work [37,45]. However, some young employees may be able to cope well with developmental tasks of this career stage and hence experience of work engagement [46]. At the mid-career stage, in turn, highly educated employees are able to maintain work engagement by successfully accomplishing developmental tasks, such as work-related development - that is, determining the possibilities of this stage for further career plateauing - and high productivity. However, for some employees the developmental tasks of the mid-career stage and simultaneous early signs of their own obsolescence may induce experiences of burnout and another negative indicator of SWB, namely workaholism [43,47]. Within this theoretical framework, we addressed the following research question:

Research question 1: What kinds of longitudinal burnout-work

engagement-workaholism profiles can be identified in a sample

of highly educated employees?

As already stated, because of the exploratory role of our main research aim, we were cautious about precisely formulating hypotheses on the number of longitudinal profiles. Similarly, we were not able to hypothesize means and possible mean changes in burnout, work engagement and workaholism in these yet-to-be-identified profiles. Since our sample comprises highly educated employees from diverse occupations and organizations, it is reasonable to expect heterogeneity in their experiences of SWB at work. We expected, first, to find at least one positive profile with low risk for a highly educated employee's well-being, i.e. a profile of high work engagement, low burnout and only moderate workaholism. Furthermore, at least one high risk profile with high burnout and workaholism and low work engagement, was expected. Second, we expected the majority of the employees to report experiences of high engagement and a minority to suffer from a high level of burnout. These expectations are based on our sample, which consisted mainly of highly educated women (75%). The Finnish Health 2000 Study [2] has previously reported that a high level of education may serve as protector against risk for burnout, especially for women. Furthermore, highly educated employees have been shown to have strong self-efficacy beliefs that, in turn, increase their engagement when faced with emotionally demanding tasks [1]. These expectations are also in line with previous empirical studies (for example [49]), according to which highly educated employees generally have experiences of supportive feedback and sense of control over their work, and as a consequence relatively high work engagement. Finally, these expectations are supported by both the Health 2000 Study [2], which reported that some 20% of Finnish employees experience burnout, and the Study of Work and Health in Finland [48], which found that nearly 90% of Finnish employees experience

engagement. Results on the prevalence of workaholism in Finland have not been published.

1.1.4. The role of achievement and social strategies in longitudinal SWB at work

Motivation has been thought to play an important role in individual well-being [42,92]. The life-span models of motivation assume that the challenges, and opportunities individuals experience at particular developmental stages of their lives channel their perceptions over the life stage [55] and personal goals [50,51]; these, in turn, influence the ways in which people direct their development [52], and are also related to SWB at work at the early career stage [53].

The cognitive and behavioral patterns that accompany adaptations to the different situations that arise in this process have been defined as cognitive and attributional strategies [54]. These, in turn, can be divided into achievement and social strategies. Achievement strategies, further, can be subdivided into functional, task-focused strategies (i.e. achievement optimism) and support-seeking in achievement situations (i.e. achievement support strategy) [54], and dysfunctional, pessimistic avoidance strategies (i.e. achievement pessimism) [54,55]. Similarly, social optimism is defined by expectations of positive outcomes in social situations, whereas social pessimism is described by social avoidance [56], and self-handicapping in social situations by social anxiety [57]. Previous studies have shown that employees using optimistic strategies tend to have experiences of work engagement [7,58,59], and strive directly for success on the basis of their high outcome expectations [54]. Employees using pessimistic strategies, in turn, have low outcome expectations and experience anxiety in task-related and social situations [54]; these lead to feelings of low activation and displeasure, and are related to poor well-being. Within this theoretical framework and based on previous findings, we set the following research question:

Research question 2: Do highly educated employees' longitudinal profiles of SWB at work differ in regard to their achievement

and social strategies?

In seeking answers to this question, we focused on three achievement and three social strategies, namely achievement optimism, achievement pessimism and seeking achievement support in achievement situations, and on social optimism, social pessimism and self-handicapping in social situations. We expected, in light of the life-span theory of motivation [50,51] and the theory of careers [43], according to which higher expectations would be related to effective coping and well-being in the future, and lower expectations to poor coping and poor well-being. These expectations were in line with previous findings [25,58] where frequent seeking of support, achievement optimism and social optimism were related to high levels of work engagement. Generally, optimistic people adjusted better to stressors and tended to experience work engagement [61]. Frequent use of achievement and social pessimism and self-handicapping in social situations, in turn, have been related to high burnout [59]. Furthermore, pessimistic social strategies are typical of workaholics, who have a propensity to neglect social relations in their lives [39].

1.1.5. SWB at work and recovery, life satisfaction, rewards and a sense of community as outcomes

We examined the outcomes of the profiles of SWB at work by examining the extent to which these profiles are related, first, to employee recovery strategies, second to life satisfaction, and finally to social areas of worklife such as experiences of rewards and sense of community.

Recovery has been seen as an important variable in a hypothetical causal string of events between the development of

Al IIIII .E IN PRESS

4 H. Innanen et al. / Burnout Research xxx (2014) xxx-xxx

work-related stress and experiences of psychological overload in the longer run [71]. According to the theory of chronic stress and accumulated fatigue, chronic stress can lead to lack of recovery strategies [72]. Employees' recovery strategies are characterized by individual experiences of psychological detachment, relaxation, control, and mastery. Psychological detachment refers to the regulation of one's leisure time [69]; relaxation refers to managing work engagement through positive ways of relaxation, such as reading a book; mastery refers to managing off-job activities [70] by such means as learning new skills during leisure time [85]; and control refers to the extent to which employees are able to choose and make decisions of leisure time, for example to participate in the course and to overcome the impulse to be lazy at home. Psychological detachment from work and relaxation after work has been shown to be important for employees' recovery [85]. Previously, poor psychological detachment has been shown to be associated with exhaustion, and mastery with work engagement [73].

As already stated, coping with career-related tasks may challenge employees' well-being and existing experiences of working life [62]. Thus, working environments may predict the level of SWB at work, which in turn may predict subsequent experiences of working environments [63]. Theoretically and according to the previous results (for example [14,64,65]), we expected the process of SWB and environments at work to be reciprocal rather than uni-directional. For example, depressive employees with poor SWB have perceived their working environment negatively as a consequence of applying the negative perceptual cycle, known as the "gloomy perception mechanism." In contrast, employees with high levels of engagement and comfort might report positively on their working environment by implementing the positive perceptual cycle, known as the "rosy perception mechanism" [66]. Engaged employees thus optimize their work environment (i.e. job crafting) (see [67,68]). In line with this notion, an association between SWB at work and positive outcomes has been indicated in a study applying the circumplex model of SWB at work [3].

Studies on the simultaneous role of different social areas at work are rare (for an exception, see [75,76]). To shed more light on social areas in working environments among highly educated employees, we focused on rewards and sense of community. Rewards include positive feedback from other people as well as personal satisfaction at work [77]. Community, in turn, refers to the overall quality of social interaction at work, and thus includes interpersonal relationships in the organization (with supervisors, colleagues and subordinates) and social support [78].

Within this theoretical framework and previous findings, we set the following research question:

Research question 3: Do employees whose longitudinal profiles

of SWB at work are different also differ with respect to recovery

strategies, life satisfaction, and social areas of worklife, such as

rewards and sense of community, during the mid-career stage?

We expected that employees experiencing a high level of work engagement and a low level of burnout and workaholism would be more likely to experience functional recovery strategies, and a high level of life satisfaction [28,62]. Further, in light of the SWB at work model [3], we expected that membership of the profile with high work engagement would predict a high level of life satisfaction [28,90,91], whereas coping successfully at one career stage would predict a high level of subjective well-being. These expectations are in line with the traditional SWB model [80], where relations between high SWB and life satisfaction as well as happiness were observed. Finally, these expectations arose from the traditional assumption of the importance of work in defining one's identity. Thus, when experiences of work are out of balance because of poor SWB at work, a negative spillover effect

on the individual's life overall, i.e. poor life satisfaction, can be expected [81].

According to the traditional notions [74], we expected that in emotionally demanding jobs, such as in jobs in teaching and jobs with demanding social relations, as is the case in our study, significant relationships might emerge between the SWB profiles and social factors in the work environment. We expected that employees with a high level of work engagement and a low level of burnout and workaholism would experience more rewards, and a stronger sense of community [7,63,76,79] than employees with a high level of burnout (i.e. "gloomy perception mechanism", low activation and displeasure) (see Fig. 1).

2. Method

2.1. Participants and procedure

This study is part of the ongoing Helsinki Longitudinal Student Study (HELS) (1991-2011), with an original sample of 292 university students (77 men, 215 women). We used three measurement points. Profiles of SWB at work were measured at two measurement points, at the early career stage, at Time 2 (year 2005), and at the mid-career stage, six years later, at Time 3 (year 2011). All the respondents available at Time 2, when the SWB profiles were contacted (year2005), 161 (40 men, 121 women; (age32-39years); 54. 8% of original sample), were included in the present study. The same participants were examined using the Mplus estimating procedure for missing values at Time 3 (year 2011) (125 of the participants were the same at both measurement points).

The antecedents of membership of the different SWB profiles were measured at the point of transition from university to early career at Time 1 (year 2001), when the participants filled in the Strategy and Attribution Questionnaire (SAQ) [54]. Subsequently, at Times 2 and 3, the participants filled in the Work engagement [22,81], Work Burnout [82], and Workaholism [83] questionnaires. At Time 3 (year 2011), the participants also filled in the Recovery Strategy, Areas of Work Life, and Life Satisfaction questionnaires.

During their career, 75% of the participants worked in the public sector - in universities, the civil service, and municipalities - and the remaining 25% in private sector organizations. More specifically, at Time 2, the largest proportion (22.3%) were teachers or researchers, mainly in the humanities and social sciences disciplines, while the rest worked in various other fields, such as social work. Women were overrepresented (75% of the sample at Time 2 when the first profile measurements were conducted). Subordinate positions were occupied by 76.3% and managerial positions by 23.7%; 71.3% felt that their education was appropriate for their current job.

2.2. Attrition analysis

Attrition analysis indicated that the final sample (n = 161) did not differ in gender from non-respondents (n = 131). Of those who dropped out during the transition from university to work, three had died, five refused to participate, 30 had an unknown address and 12 had moved abroad. The other participants who dropped out of the study had shown less progress during their last university years than those who continued to participate (M =22.85, SD = 14.11; t(297) = -3.35, p<.001). The participants in the final sample at Times 2 and 3 (n = 161) did not differ from those who left the study after T1 (n = 313) in gender [x2(1) = 1.095,ns] or age [t(314) = .378, ns]. The participants in the final sample were less prone to social optimism than those who left the study after T1 [t(188) = -2.134, p<.05]. No significant differences emerged between the final sample and those who left the study after T1

lll^^^M AIIIIIIE IN PRESS

H. Innanen et al. / Burnout Research xxx (2014) xxx-xxx 5

Table 1

The descriptives of the study variables (N = 161).

Variable items Number of Range Mean (SD) Cronbach's alpha

T1 T2 T3 T1 T2 T3

Social support 6 1-4 3.19 (.40) - - .85 - -

Achievement pessimism 6 1-4 1.81 (.54) - - .75 - -

Achievement optimism 4 1-4 3.55 (.43) - - .81 - -

Self-handicapping 5 1-4 2.49 (.48) - - .81 - -

Social pessimism 5 1-5 2.41 (.60 - - .85 - -

Social optimism 5 0-6 3.14 (.61) - - .83 - -

Workaholism 4 0-6 - 3.50(1.34) 3.32(1.32) - .80 .84

Exhaustion 3 0-6 - 2.29(1.32) 2.04(1.39) - .86 .89

Cynicism 3 0-6 - 1.90(1.38) 1.66(1.46) - .75 .79

Professional efficacy 3 0-6 - 4.55(1.12) 4.55(1.12 - .70 .74

Vigor 3 0-6 - 4.15(1.13) 4.15(1.13) - .83 .83

Dedication 3 0-6 - 4.06(1.28) 4.30(1.24) - .87 .90

Absorption 3 1-5 - 4.17(1.10) 4.71 (1.07) - .74 .78

Rewards 2 1-5 - - 2.75 (3.72) - - .72

Community 2 1-5 - - 3.78(1.03) - - .81

Detachment 2 1-5 - - 3.67 (.87) - - .60

Mastery 2 1-5 - - 2.92 (.94) - - .71

Relaxation 2 1-5 - - 3.62 (.89) - - .71

Control 2 1-5 - - 3.39(1.14) - - .90

Life satisfaction 2 1-5 - - 5.03 (1.32) - - .84

to exhaustion [t(165) = 1.584, ns], cynicism [t(162) = 1.010, ns], or professional efficacy [t(163) = 1.338, ns]. Work engagement and workaholism were not measured at T1.

2.3. Measures

Burnout was assessed with the revised Maslach Burnout Inventory- General Survey (MBI-GS) [82]. The scale includes 3 questions related to exhaustion (e.g., 'I feel emotionally drained from my work'), 3 questions on cynicism (e.g., 'I have become more cynical about whether my work contributes anything'), and 3 questions on professional efficacy (e.g., 'At my work, I feel confident that I am effective at getting things done') to be answered on a Likert-type scale from 0 (never) to 6 (daily). The items from the original scale were as follows: exhaustion 1, 2, 3 (factor loadings .92, .85, .86); cynicism 8, 14, 15 (factor loadings .76, .79, .83); and professional efficacy 5,12,16 (factor loadings .72, .85, .80). The construct validity of the revised MBI-GS was good and the model fitted the data well: x2(23) = 23.562, p = .428, CFI = 1.00, TLI = 1.00, RMSEA = .012, SRMR = .03 [88,89]. Exhaustion correlated significantly with cynicism (r =.38, p<.001), and professional efficacy (r = -.21, p<.01). Cynicism correlated significantly with efficacy (r = -.30, p<.001). Internal consistencies (Cronbach's alpha a) are presented in Table 1. All the a values met the criterion of .70 [93]. The consistency of the scale over time was shown by a values, the levels of which were good at both measurement points.

Work engagement was assessed with the revised Finnish version of the Utrecht Work Engagement Scale (UWES) [22,94]. The UWES-9 has been shown to have good construct validity and rankorder stability in a recent Finnish study [9]. The scale includes 3 questions related to vigor (e.g., 'At my work, I feel bursting with energy'), 3 questions regarding dedication (e.g., 'I find the work that I do full of meaning and purpose'), and 3 questions on absorption (e.g., 'When I am working, I forget everything else around me') to be answered on a Likert-type scale from 0 (never) to 6 (daily).

Workaholism was assessed with 4 questions adopted from the Work Addiction Risk Test [83]. Two questions measured working excessively ('I seem to be in a hurry and racing against the clock', 'I stay busy and keep many irons in the fire'), and two measured compulsion tendency ('I feel guilty if I don't work all the time'; I find myself continuing to work after my coworkers have stopped working'). The questions were rated on a Likert-type scale from 0 (never) to 6 (daily) (factor loadings .99,1.00,1.00, .99). The

construct validity of the scale was excellent: x2(2) = .322, p = .852, CFI = 1.00, TLI = 1.00, RMSEA= .00, SRMR = .00 [72,73]. The internal consistency, convergent validity and discriminant validity of the measurement (five items both) have been shown previously. Furthermore, the two-factor structure has shown good fit to the data in other cultural contexts [84].

Achievement and social strategies were assessed with the extended Strategy and Attribution Questionnaire (SAQ) (for details, see [54]). The participants were asked to rate statements on a 4-point Likert-type scale (1 = strongly agree, and 4 = strongly disagree). Achievement optimism was measured with 5 items (e.g., 'When I get ready to start a task, I am usually certain that I will succeed in it'), achievement pessimism with 6 items (e.g., 'What often happens is that I find something else to do when I have a difficult task in front of me'), and social support strategy with 7 items (e.g., 'When things don't function, it is better to talk with your partners). Social optimism was measured with 4 items (e.g., 'In most cases, I feel I get along well with other people'), social pessimism with 5 items (e.g., 'I often have more important things to do than spend time with my colleagues'), and self-handicapping with 4 items (e.g., 'I often avoid group situations and I'd rather be alone or together with one person at a time').

The items for rewards and a sense of community were drawn from the Areas ofWorklife Survey (AWLS) [75]. In the present study, these were 28 and 29 for rewards (e.g., 'My work is usually noticed') and 31 and 33 for sense of community (e.g., 'I am a member of a supportive team'), which characterize the social dimensions of work environments. Respondents indicated their degree of agreement with these statements on a 5-point Likert-type scale from 1 (strongly disagree) to 5 (strongly agree). The factor loadings for rewards were .75, .69, and for sense of community .88, .54. The inter-item correlations were for good for both rewards (p<.001), and sense of community (p < .001).

Recovery strategies were assessed by the revised version of the Recovery Experience Questionnaire [85]. Each type of recovery strategy was measured by two items: e.g., 'I distance myself from my work' (detachment), 'I spend time relaxing' (relaxation), 'I seek out intellectual challenges' (mastery), and 'I take care of things the way that I want them done' (control). Participants were asked to rate the statements on a 5-point Likert-type scale ranging from 1 (totally disagree) to 5 (totally agree). Factor loadings for control were .92, .88, for mastery: .86, .65, for detachment .57, .79, and for relaxation: .70, .80. The construct validity of the

AIIIIII-E IN PRESS

H. Innanen et al. | Burnout Research xxx (2014) xxx-xxx

Table г

Q9 Fit indices for the longitudinal occupational well-being (burnout, work engagement, workaholism) with different numbers of latent classes (latent profile analyses).

No. ofclasses Log L BIC Entropy BLRT Latent class proportions n (%)

1 -4824.92 10,005.54 - - 161(100)

2 -4778.39 9988.70 .94 .00 136 (84)/25 (16)

3 -4746.75 10,001.63 .82 .00 85 (53)/61 (38)/15 (9)

4 -4717.94 10,002.24 .86 .00 84 (52)/53 (33)/15 ((9)/9 (6)

Notes: BIC = Bayesian Information Criterion; BLRT = Parametric Bootstrapped Likelihood Ratio Test.

scale was moderate: x2(14) = 33.571, p = .002, CFI = .953,TLI = .905, RMSEA = .088, SRMR= .045 [88,89]. Although the p-value was <.05, all the other goodness-of-fit-values showed that the model fitted the data well. Detachment correlated with mastery (r = .24, p <.01, relaxation (r = .43, p < .001), and control (r = .25, p <01). Mastery correlated with relaxation (r =36, p <.001) and control (r = .37, p <001), and relaxation correlated with control (r =.59, p<001). All the upvalues, except detachment, met the criterion of .70 [93]. The a value for detachment was at a moderate level. Moreover, the inter-item correlations for separate variables were good (p< .001).

Life satisfaction was assessed by the Satisfaction with Life Scale (SWLS) [86]. The scale comprises 5 questions (e.g., 'The circumstances in my life are excellent') to be answered on a 5-point Likert-type scale ranging from 1 (totally disagree) to 5 (totally agree).

530 2.4. Analytical strategy

Employees with similar longitudinal profiles of burnout, workaholism, and work engagement were identified by using Latent Profile Analysis (LPA) [87], applying the missing data method, where the program estimates the values of the participants from Time 2 at Time 3. The estimation method used was maximum likelihood performed by Mplus program version 5.21 [88]. LPA analysis is a type of finite mixture analysis that seeks to identify the smallest number of homogenous latent classes (profiles) that adequately describe the associations among the observed continuous variables (burnout, work engagement, workaholism [89]. The advantage of LPA over traditional cluster analysis is that it is model-based and permits the use of statistical criteria for deciding the number of latent classes for the core indicators of the SWB at work model. The estimation was performed step by step, starting from the one-class

solution to estimate the parameters for 2, 3.....k-class solutions

[13]. To ensure the validity of each class solution, a large set of random starting values (500) for the parameters were used in the present study.

The Bayesian Information Criterion (BIC), Bootstrapped Likelihood Ratio Test and classification quality test (Entropy) - as implemented in the Mplus statistical program - were used as the statistical criteria for choosing the best-fitting model. Models with a lower BIC value and higher Entropy value fit the data better. Changes in burnout, workaholism and work engagement between measurement points were measured by multivariate analyses of covariance (MANCOVA) for repeated measures. Using the SPSS 16 and PASW Statistics 18 programs, we examined whether the latent classes would differ in background factors (gender and age). In the attrition analysis, either the chi-squared test or t-test was used. The bivariate logistic regression analysis of the PASW 18 program was used, first, to examine achievement and social strategies at Time 1 as antecedents of the longitudinal SWB profiles, and second to analyze associations between achievement and social strategies (at Time 2 and at Time 3) and the longitudinal SWB at work profiles. Multivariate analysis of covariance (MANCOVA) was used to analyze the differences between the longitudinal SWB profiles in, first,

recovery strategies, second, life satisfaction, and, third, rewards and sense of community (at Time 3).

3. Results

Design of the study is presented in Fig. 2. Descriptive informations on all the study variables (number of items, ranges, means, standard deviations, reliabilities) are shown in Table 1. Table 2 reports the tested latent class solutions for the longitudinal patterns of burnout, work engagement and workaholism included in the LPA analysis (see also Appendix A). The fit indices did not support the one-class solution as compared to multi-class solutions. We compared the profiles of the different latent multi-class solutions and found theoretically meaningful patterns that remained unchanged in their content in most of the multi-class solutions (i.e. solutions with 2-4 classes).

With respect to the fit indices (see Table 2), the BIC value supported the two-class solution. Entropy was very good (.94) in the two-class solution, indicating that the two-class model provides a clear classification. On the basis of the above-described stability of the patterns as well as the information provided by the fit indices, we decided to choose a two-class solution for our subsequent analyses. We made this choice for four reasons: first, the two-class solution included the theoretically meaningful SWB profiles, which remained unchanged in all the multiclass solutions; second, unlike the three- and four-class solutions, the two-class solution did not produce several profiles with few members; third, the two-class solution produced theoretically justified profiles; and fourth the average latent class probabilities for the most likely latent class membership by latent class showed good levels. The two-class solution, containing two SWB profiles, is shown in Table 2. Profile I represented the low-risk profile. The participants in this profile 1 (n = 136; 84%) reported high work engagement and professional efficacy, low cynicism and moderate workaholism. The participants in the profile 2 (n = 25; 16%) reported high exhaustion, low energy and rather high workaholism. In both profiles, the employees reported high levels of professional efficacy and absorption.

Personal strategies Time 1

Longitudinal occupational well-being profiles from Time 2 to Time 3

Outcomes

600 601

Figure 2. Design of the study. T1, Time 1; T2, Time 2; T3, Time 3.

ti^^^M AI IIHI.E IN PRESS

H. Innanen et al. / Burnout Research xxx (2014) xxx-xxx 7

Table 3

Q10 Changes in burnout, engagement and workaholism according to two longitudinal occupational well-being profiles, using multivariate analyses of covariance (MANCOVA) for repeated measures (gender and age were controlled for).

Engaged(n =136) Exhausted-Workaholic (n = 25) Pattern differences Time effect Pattern x Time

M (SE) M (SE) F F F

T2 T3 T2 T3

Exhaustion 1.98 (.10) 1.71 (.12) 4.53 (.22) 4.07 (.26) 153.48*** 1.21 ns .31ns

Cynicism 1.64 (.12) 1.55 (.15) 3.93 (.26) 3.54 (.32) 76.93*** 1.62 ns .49 ns

Professional efficacy 4.56 (.11) 4.16 (.12) 4.59 (.23) 4.18 (.26) 3.28 ns .23 ns .01 ns

Energy 4.35 (.11) 4.41 (.11) 3.09 (.24) 3.28 (.24) 27.11*** 1.36 ns .30 ns

Dedication 4.30 (.12) 4.33 (.12) 2.63 (.24) 3.05 (.24) 32.89*** 1.95 ns .28 ns

Absorption 4.24 (.11) 4.70 (.12) 3.94 (.27) 4.42 (.26) 1.27 ns .01 ns .22 ns

Workaholism 3.42 (.14) 3.17 (.13) 4.61 (.30) 4.01 (.28) 13.30*** 18.95*** 13.30***

* p<.05. " p<.01. ■" p<.001

602 These two SWB profiles were labeled Engaged and Exhausted-

603 Workaholic.

604 The means and mean changes for exhaustion, cynicism, pro-

605 fessional efficacy, energy, dedication, absorption and workaholism

606 in the two longitudinal SWB profiles are shown in Table 3. Sig-

607 nificant and stable differences between the profiles were found

608 in the means of exhaustion, cynicism, energy, dedication, and

609 workaholism. A significant profile x time interaction was observed

610 for workaholism; decreasing workaholism was found in the

611 Exhausted-Workaholic profile.

612 The chi-squared tests revealed no significant differences

613 between the SWB profiles in either employees' reports regarding

614 the appropriateness of their education for their current job

615 (options Yes or No) [x2(1, N =161) = 0.01,ns] or gender [x2 (1,

616 N = 161) = 0.01, ns]. Independent samples t-test revealed no age dif-

617 ferences between the SWB at work profiles [t(159) = -1.09, ns].

618 3.1. Achievement and social strategies as antecedents of SWB at

619 work profiles

620 The results revealed that (Time 1) social pessimism in the

621 final model, determined by bivariate logistic regression analysis,

622 was a significant antecedent of membership of the Exhausted-

623 Workaholic profile. To examine the role of achievement and social

624 strategies from early to mid-career, we analyzed the simultaneous

625 associations between achievement and social strategies and the

626 SWB profiles at the early career and at the mid-career stages. The

627 results showed that achievement optimism was negatively asso-

628 ciated to membership of the Exhausted-Workaholic profile and

629 supported membership of the Engaged profile at the early career

630 stage, at Time 2, and at the mid-career stage, at Time 3, in the final

631 model (Table 4).

632 3.2. SWB at work and recovery strategies, life satisfaction,

633 rewards, and sense of community

634 At Time 3, of the four recovery strategies, the longitu-

635 dinal SWB profiles showed significant differences in, first,

Table 4

The results of binary logistic regression analyses (with Exhausted-Workaholic profile as the reference category) in the years 2001, 2005 and 2011 (final models).

Achievement and social strategies ß (SE) OR (95% Cl)

Exhausted-Workaholic

n = 25(16%)

Social pessimism 2001 1.52(0.67) 4.56* (1.24,16.77)

Achievement optimism 2005 -1.31 (0.55) .27* (0.09, 0.79)

Achievement optimism 2011 -1.27(0.48) .28** (0.11,0.72)

Notes: OR=odds ratio, Cl = confidence interval. * p<.05 ** p<.01.

psychological detachment and relaxation (Table 5). The employ- 636

ees in the Engaged profile, i.e. the low-risk profile, reported 637

more psychological detachment and relaxation than those in the 638

Exhausted-Workaholic profile, i.e. the high-risk profile. Experi- 639

ences of mastery and control, however, did not differ between 640

the two longitudinal SWB profiles. Second, the employees in the 641

Engaged profile reported significantly more life satisfaction, and 642

experiences of rewards than those in the Exhausted-Workaholic 643

profile. Experiences of sense of community, in turn, did not differ 644

between the two SWB profiles. 645

4. Discussion 646

This study contributes to the previous studies (e.g., [14]) by uti- 647

lizing a person-oriented approach, a follow-up period of six-years 648

(the profiles), and data gathered in two waves among highly edu- 649

cated Finnish employees to identify longitudinal profiles of SWB 650

at work (simultaneous burnout, work engagement, workaholism). 651

Academically highly educated employees formed our target group, 652

since work requiring higher education is regarded as not only as 653

challenging [4] but also as engaging [9]. High education may, fur- 654

thermore, increase work engagement [1], and protect against risk 655

of burnout [2]. 656

The results showed heterogeneity in the levels of, and changes 657

in highly educated employees' SWB at work. The three quadrants 658

(burnout, work engagement and workaholism), relevant to our 659

aims, were included simultaneously in the LPA analysis. Overall, 660

in line with the circumplex model of SWB at work [3], we expected 661

to differentiate at least two clear longitudinal SWB profiles among 662

highly educated employees: one low risk profile characterized 663

by high work engagement (energy, dedication, absorption), and 664

one high risk profile characterized by high burnout (exhaustion, 665

Table 5

Meandifferences in recovery strategies, experiences of rewards, sense ofcommunity Q11 and life satisfaction between the two longitudinal occupational well-being profiles using multivariate analyses of covariance (MANCOVA) (gender and age were controlled for).

Engaged (n = 136) M (SE) Exhausted-Workaholic (n=25) M (SE) F-test Partial r2

Detachment 3.79 (.82) 3.28 (.80) 9.505** .05

Relaxation 3.73 (.83) 3.25 (.88) 6.43* .04

Mastery 2.97 (.89) 2.58(1.14) 3.13 ns .02

Control 3.48(1.14) 3.20(1.25) 1.20 ns .01

Rewards 3.56 (.87) 2.93(1.12) 8.13** .06

Community 3.84(1.00) 3.45(1.12) 2.50 ns .02

Life satisfaction 5.05 (1.20) 4.20(1.64) 7.18** .06

* p<.05. ** p<.01. ***p <.001.

lll^^^M AIIIIII.E IN PRESS

8 H. Innanen et al. / Burnout Research xxx (2014) xxx-xxx

cynicism, low professional efficacy) and high workaholism. Further, the achievement and social strategies of the participants had been examined during their transition from high education to early career, four years before the first analysis of their profiles of SWB at work. This had been done in order to reveal the role of achievement and social strategies as antecedents of their subsequent membership of different profiles of SWB at work. The simultaneous associations of achievement and social strategies with membership of the SWB profiles during the career were also analyzed. Furthermore, at the mid-career stage, the outcomes of membership of the two SWB profiles were examined.

First, we found as expected, the longitudinal paths of SWB at work from early to mid-career. Two distinctive profiles of SWB at work were observed: an Engaged profile (84% of employees) and an Exhausted-Workaholic profile (16% of employees). Second, the deployment of social pessimism in the transition from education to work anticipated membership of the longitudinal Exhausted-Workaholic profile, while achievement optimism during the career was negatively associated with membership of the Exhausted-Workaholic profile, and linked to the Engaged profile. Finally, as outcomes of the two profiles, the Engaged employees experienced higher levels of psychological detachment and relaxation, rewards, and life satisfaction than their Exhausted-Workaholic counterparts.

4.1. Profiles of SWB at work from early to mid-career

Examining SWB profiles according to sub-dimensions of burnout and work engagement, provided important new knowledge of heterogeneity in SWB profiles. The results contributed to the circumplex model of SWB at work and showed longitudinally that frequent positive emotions and fewer negative emotions, typical of work engagement, indicated membership of the Engaged profile. These employees were in activation state in regard to high engagement, low burnout and moderate workaholism. The Engaged profile was characterized by high energy and dedication. Frequent negative emotions and fewer positive emotions, on the contrary, tend to be related to the Exhausted-Workaholic profile. The Exhausted-Workaholic profile was observed as a risk profile in terms of experiences of displeasure; characterized by high exhaustion, cynicism and a high level of workaholism. This result was in line with our expectations and with previous findings indicating that highly educated employees are likely to be classified in the Engaged profile [1], and are likely to be work-engaged and committed to their organizations [11,49]. This result was in line with general findings on burnout and work engagement in Finnish working population [2,48]. One reason for the inclusion in the Engaged profile of the majority of the present sample of highly educated employees might also be their high level of education, as this may help them to pursue appropriate goals and maintain adaptive coping at different stages of their careers [46,54,55].

Interestingly the two profiles detected in the present study were quite the opposite of each other. When measuring burnout and work engagement simultaneously with separate measurements, exhaustion and cynicism, the indicators of burnout [27], were at a high level in the Exhausted-Workaholic profile, and at a low level in the Engaged profile. In the Engaged profile, in turn, energy and dedication were high and in the Exhausted-Workaholic profile they were low. Professional efficacy and absorption were at the same level in both profiles, which may suggest that they are more independent factors in subjective well-being (see [27]). According to the perspective of Maslach et al. [6], low scores on exhaustion and cynicism, and high scores on efficacy, would be indicative of job engagement in the burnout-work engagement continuum. It would be interesting to examine the SWB profiles in future studies from the perspective of the burnout-engagement continuum proposed by Maslach et al. [6]. It has been suggested that the practical

significance of the burnout-work engagement continuum is that, when measuring occupational well-being among employees, work engagement represents a goal requested by applicants for burnout interventions [6].

Noteworthy significant differences in the levels of absorption and professional efficacy between the longitudinal Engaged and Exhausted-Workaholic profiles were not observed in the present study. The levels of absorption and professional efficacy were rather high in both profiles. This result supported the theoretical framework [22,29,30], according to which high absorption in work could be a factor in both high and poor SWB at work (i.e. related to burnout, work engagement, and workaholism [3], and that professional efficacy and absorption could represent more independent factors in subjective well-being [27,94]. This result may also further support the view that the role of education in SWB at work is important. Further, a favorable change in the experience of worka-holism was observed among those in the Exhausted-Workaholic profile from early to mid-career. At the beginning of their career, some young highly educated employees experienced a high level of workaholism. During their career, they may continue to experience burnout but, however, less workaholism, through appropriate utilization both of their education and work experience and of more adaptive coping strategies. This result showed that those highly educated employees in the Exhausted-Workaholic profile turned to more a positive direction during their career. As expected of highly educated employees, they also showed a good level of professional efficacy during their career. For this reason, it might be possible to diminish their propensity to workaholism during their later career, and turn them more toward high pleasure/positive high activation in the circle described by the circumplex model [3], and along the continuum of burnout [15]. This result may emphasize the importance of high quality education as a basis for longitudinal SWB at work. Furthermore, this result emphasizes the importance of being able to cope successfully with developmental and career-related tasks during the early career stage in order to provide the basis for coping with career-related tasks in the mid-career stage [42,43,46].

4.2. The role of achievement and social strategies in the longitudinal profiles of SWB at work as antecedents and during the career

Our results supported our expectations, the life-span models of motivation [50,52,53], and previous findings (for example [59]) according to which social pessimism is related to negative, unpleasant emotions and low expectations and further poor SWB at work [3]. Our results were in line with the life span theory of motivation and previous findings [58] according to which achievement optimism is associated to high work engagement and low burnout. The results revealed, first, the strong role of social pessimism, measured in the transition from education to work, as an antecedent of membership of the Exhausted-Workaholic profile, i.e. the high risk profile, later in the career. This result is significant when revealing the role of social pessimism in the face of risk for young employees' further marginalization in working life. The result suggested that frequent social pessimism in student life, and thus fewer social relationships [33,34] may have far-reaching consequences for young adults. This result strongly suggests that further attention should be paid to motivation and adaptive social strategies in education, and especially in the transition from education to work, in order to experience further work engagement [50,51,54].

Notably, the results showed that later on, in the early and in mid-career, achievement optimism was negatively associated with the Exhausted-Workaholic profile and had the strongest association with membership of the Engaged profile in both career stages. This result was in line with the propositions of the life-span model of motivation (for further information, e.g., [13,50,52,53]). Motivation

llll^^^M AlIIHI.E IN PRESS

H. Innanen et al. / Burnout Research xxx (2014) xxx-xxx

and successful coping in career-related tasks could increase positive expectations of outcomes, which, in turn, could increase further achievement optimism and support membership of the Engaged profile later on [46,50]. Thus, optimistic attitudes in achievement situations enabled better adjustment to stressors in career-related tasks [25,61].

4.3. Outcomes of longitudinal SWB at work profiles

service is well organized. Thus, this study may be of great value for occupational health care service experts interested in identifying different profiles of SWB at work and developing burnout interventions during the early career stage, and thus reducing the prospect of poor SWB during the later career. The study may also be valuable for the management and Human Resources departments of organizations seeking to support work engagement.

860 861 862

In line with our expectations and previous results [27,85], highly educated employees in the Engaged profile experienced higher levels of detachment and relaxation, a greater life satisfaction, and a stronger sense of sufficient rewards than employees in the Exhausted-Workaholic profile. This result strengthened the findings, based on the model of SWB at work proposed by Bakker and Oerlemans [3] and Russell [19], that positive emotions lead to work engagement, which, in turn, has positive outcomes. These results emphasized the role of high SWB at work for psychological detachment from work and relaxation. Experiences of mastery and control, however, did not differ between the two longitudinal SWB profiles, suggesting that highly educated employees might be able to master and control the relationships between work and leisure-time. Recovery has been shown to be an important variable in a hypothetical causal string of events between the development of work-related stress and experiences of psychological overload in the longer run [71]. Our results supported the theoretical evidence and previous results according to which chronic stress may lead to lack of recovery strategies [72]. These results may support the phenomenon known as the cumulative cycle, according to which young adults' social pessimism in the transition from higher education to work predicts poor SWB at work, which, in turn, predicts poor psychological detachment and relaxation, and furthermore poor life satisfaction as a consequence of the negative life spiral. Most of all, our results were in line with the results of Sonnentag and Fritz [85], who found that feelings of psychological detachment from work were crucial for SWB at work. Finally, the results supported the theoretical argument [81] that poor SWB at work causes negative spillover over to the individual's life overall, leading to poor life satisfaction [28].

These results also supported the theoretical notion [76] that individuals with poor well-being perceive their environment negatively as a consequence of the negative perceptual cycle. Individuals with high levels of engagement, instead, perceive their working environments positively by implementing the positive perceptual cycle, and thus optimizing their work environment [67,68]. Thus, SWB at work would be a significant predictor of experiences of social areas of worklife, for example internal rewards such as a sense of work well-done and external rewards such as positive feedback of coworkers and supervisors. In line with previous findings, the present study revealed that feelings of reward, such as positive feedback from other people and personal satisfaction at work were high among members of the Engaged profile [77]. This result emphasized the role of SWB in experiences of rewards.

Finally, from the viewpoint of practice, the results supported the circumplex model of SWB [3], according to which burnout and workaholism comprise the negative forms of SWB at work. Work engagement, on the contrary, is the positive form of SWB at work. Furthermore, research utilizing also sub dimensions of burnout and work engagement provides information that may be useful in the development of interventions on an individual and an organizational level. This study emphasizes the importance of paying attention to motivation and adaptive social strategies during the stage of education, before pressures from working life start to accumulate. In Finland the occupational health care

5. Limitations 866

As well as contributions, our study also has some limitations. 867

First, the sample was strongly female-dominated and restricted 868

to employees who had graduated from the social sciences. Also, 869

the sample consisted of graduates from only one university. How- 870

ever, the university in the present study was a large university 871

in Finland. The third limitation is that all of the measures used 872

were self-reports. Fourth, the sample size was rather small. Fifth, 873

only social working environments were examined as outcomes of 874

the profiles of SWB at work. Using the complete Areas of Work 875

Life Scale would give useful further knowledge of the outcomes of 876

these profiles. Future research on the SWB at work of highly edu- 877

cated employees could greatly benefit from using larger samples 878

from several universities, also examining vocational tracks, and by 879

gathering responses not only from employees but also from super- 880

visors, co-workers and spouses. Replicating the findings among 881

blue-collar workers would also bring further benefits. Moreover, a 882

longitudinal study with more than two measurement points would 883

offer more possibilities for estimating changes in burnout-work 884

engagement-workaholism profiles. 885

6. Conclusions 886

The present study examined SWB at work from early to 887

mid-career using a person-oriented approach. To profiles were 888

identified: an Engaged profile and an Exhausted-Workaholic pro- 889

file. SWB at work was at a good level among the sample of 890

highly educated employees in Finland. The majority of employ- 891

ees were members of the longitudinal Engaged profile. Second, the 892

results revealed continuity in both profiles, with only a minority 893

of employees experienced changes in their profile. It is notewor- 894

thy that decreasing workaholism from early to mid-career was 895

observed in the Exhausted-Workaholic profile. Third, adopting a 896

high level of social pessimism during the transition from educa- 897

tion to work would predispose newcomers to membership of the 898

Exhausted-Workaholic profile. Fourth, membership of the Engaged 899

profile was related to experiences of psychological detachment 900

from work and relaxation, sufficient rewards, and overall life satis- 901

faction. 902

Conflict of interest statement 903

The authors declare that there are no conflicts of interest. 904

Uncited reference Q4 905

[60]. 906

Acknowledgements 907

We are grateful to The Finnish Cultural Foundation/Central QQ56 908 Finland Regional Fund for financial assistance. This study was also Q6 909

supported by a research project of the Academy of Finland (Nr. 910

139931). 911

AIIIHI.E IN PRESS

10 H. Innanen et al. / Burnout Research xxx (2014) xxx-xxx

Appendix A. Average latent class probabilities for most likely latent class membership by latent class

Most likely latent class membership Latent class

1 0.990 0.010

2 0.031 0.969

Note. Values in italics represent the average posterior probability associated with the clusters to which participants are assigned.

References

[1] Xanthopoulou D, Bakker BA, Fischbach A. Work engagement among employees facing emotional demands: the role of personal resources. J Pers Psychol 2013:12:74-84, http://dx.doi.org/10.1027/1866-5888/a000085.

[2] Ahola K, Honkonen T, Kalimo R, Nykyri E, Aromaa A, Lonnqvist J. Tyou-upumus Suomessa. Terveys 2000 - tutkimuksentuloksia [Burnout in Finland -results from the Finnish Health 2000 Study]. Suom laak lehti 2004:59:4109-13 http://www.julkari.fi/handle/10024/97657

[3] Bakker AB, Oerlemans WGM. Subjective well-being at work in organizations. In: Cameron K, Spreitzer G, editors. Handbook of positive organizational scholarship. Oxford: Oxford University Press: 2011. p. 178-89 http://www.pop-lab.com/beheer/userupload/papers/27.pdf

[4] Bakker AB, Demerouti E, Euwema MC. Job resources buffer the impact of job demands on burnout. J Occup Health Psychol 2005;10(2):170-80, http://dx.doi.org/10.1037/1076-8998.10.2.170.

[5] Freudenberger HJ. Staff burn-out. J Soc Issues 1974;30(1):159-65, http://dx.doi.org/10.1111/j.1540-4560.1974.tb00706.x.

[6] Maslach C, Leiter M, Schaufeli MW. Measuring burnout. In: Cooper CL, Cartwright S CL, editors. The Oxford handbook of organizational well-being. Oxford, UK: Oxford University Press; 2009. p. 86-108.

[7] Kahn WA. Psychological conditions of personal engagement and disengagement at work. Acad Manag J 1990;33(4):692-724.

[8] Hakanen JJ, Schaufeli W, Ahola K. The job demands-resources model: a three-year cross-lagged study of burnout, depression, commitment, and work engagement. Work Stress 2008;22:224-41, http://dx.doi.org/10.1080/026783 708023 79432.

[9] Seppala P, Mauno S, Feldt T, Hakanen JJ, Kinnunen U, Tolvanen A, et al. The construct validity of the Utrecht Work Engagement Scale: multisample and longitudinal evidence. J Happiness Stud 2009;10:459-81, http://dx.doi.org/10.1007/s10902-008-9100-y. http://www.jstor.org/stable/10.2307/256287 [accessed 21.01.13].

[10] Laursen B, Hoff E. Person-centered and variable-centered approaches to longitudinal data. Merrill-Palmer Q 2006;52:377-89.

[11] Boersma K, Lindblom K. Stability and change in burnout profiles overtime: a prospective study in the working population. Work Stress 2009;23(3):264-83, http://dx.doi.org/10.1080/02678370903265860.

[12] Makikangas A, Feldt T, Kinnunen U, Tolvanen A. Do low burnout and high work engagement always go hand in hand? Investigation of the energy and identification dimensions in longitudinal data. Anxiety Stress Coping 2012;25(1):93-116, http://dx.doi.org/10.1080/10615806.2011.565411.

[13] Feldt T, Huhtala M, Kinnunen U, Hyvonen K, Makikangas A, Sonnentag S. Long-term patterns of effort-reward imbalance and over-commitment: investigating occupational and recovery experiences as outcomes. Work Stress 2013;27(1):64-87, http://dx.doi.org/10.1080/02678373.2013.765670.

[14] Makikangas A, Feldt T, Kinnunen U. Warr's scale of job-related affective well-being: a longitudinal examination of its structure and relationships with work characteristics. Work Stress 2007;21(3):197-219, http://dx.doi.org/10.1080/02678370701662151.

[15] Bergman L, Trost K. The person-oriented vs. variable-oriented approach: are they complementary, opposite, or exploring different worlds? Merrill-Palmer Q2006;52:601-32.

[16] Laursen BP, Hoff E. Person-centered and variable-centered approaches to longitudinal data. Merrill-Palmer Q 2006;52(3):377-89. http://web.ebscohost.com/ehost/pdfviewer/pdfviewer?sid=a93e2318-261b-413c-a199-322e9c596b19%40sessionmgr110&vid=2&hid=126 [accessed 15.12.12].

[17] Kauppinen T, Mattila-Holappa P, Perkio-Makela M, Saalo A, Toikkanen J, Tuomivaara S, et al. Tyo ja terveys Suomessa. Helsinki: Tyoterveyslaitos; 2012 http://www.ttl.fi/fi/verkkokirjat/tyo-ja-terveys-suomessa/Documents/Tyo-ja-Terveys_2012.pdf

[18] Diener E, Sandvik E, Pavot W. Happiness is the frequency, not the intensity, of positive versus negative affect. In: Strack F, Argyle M, Schwarz N, editors. Subjective well-being at work: an interdisciplinary perspective. New York: Pergamon; 1991. p. 119-38.

[19] Russell JA. Emotion, core affect, and psychological construction. Cogn Emot 2009;23:1259-83, http://dx.doi.org/10.1080/02699930902809375.

[20] Hakanen JJ, Bakker AB, Demerouti E. How dentists cope with theirjob demands and stay engaged: the moderating role of job resources. Eur J Oral Sci 2005;113:479-87.

[21] Folkman S. The case for positive emotions in the stress process. Anxiety Stress 986 Coping 2008;21:3-14, http://dx.doi.org/10.1080/10615800701740457. 987

[22] Schaufeli WB, Salanova M, Gonzalez-Roma V, Bakker AB. The 988 measurement of engagement and burnout: a two sample confirm- 989 atory factor analytic approach. J Happiness Stud 2002;3(1):71-92. 990 http://www.academia.edu/861114/The_Measurement_of_Engagement_and 991 Burnout_A_Two_Sample_Confirmatory_Factor_Analytic_Approach [accessed 992 15.12.12]. 993

[23] Bakker AB, Demerouti E. Towards a model of work engagement. Career Dev Int 994 2008;13(3):209-23, http://dx.doi.org/10.1108/13620430810870476. 995

[24] Gorgievski MJ, Hobfoll SE. Work can burn us out or fire us up: conservation of Q7 996 resources in burnout and engagement. In: Halbesleben JRBD, editor. Handbook 997 ofstress and burnout in health care. 2008. p. 7-22. 998

[25] Aspinwall LG, Taylor SE. Modeling cognitive adaptation: a longitudinal investi- 999 gationofthe impact of individual differences and coping on college adjustment 1000 and performance. J Pers Soc Psychol 1992;63:989-1003. 1001

[26] Maslach C, Schaufeli WB, Leiter MP. Job burnout. Annu Rev Psychol 1002 2001;52(1):397-422. 1003

[27] Maslach C, Leiter MP. Early predictors of job burnout and engagement. J Appl 1004 Psychol 2008;93:498-512, http://dx.doi.org/10.1037/0021-9010.93.3.498. 1005

[28] Hakanen JJ, Schaufeli WB. Do burnout and work engagement pre- 1006 dict depressive symptoms and life satisfaction? A three-wave 1007 seven-year prospective study. J Affect Disord 2012;13(2-3):415-24, 1008 http://dx.doi.org/10.1016/jjad.2012.02.043. 1009

[29] Oates WE. On being a workaholic. Pastor Psychol 1968;19(8):16-20, 1010 http://dx.doi.org/10.1007/BF01785472. 1011

[30] McMillan LHW, O'Driscoll MP, Marsh NV, Brady EC. Under- 1012 standing workaholism: data synthesis, theoretical critique, and 1013 future design strategies. Int J Stress Manag 2001;8(2):69-91. 1014 http://link.springer.com/article/10.1023/A%3A1009573129142#page-1 1015 [accessed 15.12.12]. 1016

[31] McMillan LHW, O'Driscoll MP. Workaholism and health: implica- 1017 tions for organizations. J Organ Change Manag 2004;17:509-19, 1018 http://dx.doi.org/10.1108/09534810410554515. 1019

[32] Shimazu AWB, Schaufeli WB. Is workaholism good or bad for employee 1020 well-being? The distinctiveness of workaholism and work engage- 1021 ment among Japanese employees. Ind Health 2009;47(5):495-502. 1022 http://www.ncbi.nlm.nih.gov/pubmed/19834258 [accessed 15.01.13]. 1023

[33] Mudrack PE. Understanding workaholism: the case of behavioral tendencies. 1024 In: Burke R, editor. Research companion to working time and work addiction. 1025 Cheltenham, UK: Edward Elgar; 2006. p. 108-28. http://scholar.google.fi/ 1026 scholar?hl=en&q=Mudrack+2006&btnG=&as_sdt=1%2C5&as_sdtp= [accessed 1027 15.12.12]. 1028

[34] Mudrack PE, Naughton TJ. The assessment of workaholism 1029 as behavioral tendencies: scale development and prelimi- 1030 nary empirical testing. Int J Stress Manag 2001;8(2):93-111 1031 http://link.springer.com/article/10.1023/A%3A1009525213213#page-1 1032

[35] Taris TW, Schaufeli WB,Shimazu A. The push and pull ofwork: the differences 1033 between workaholism and work engagement. In: Bakker AB, Leiter MP, editors. 1034 Work engagement: a handbook of essential theory and research. New York: 1035 Psychology Press; 2010. p. 39-53. 1036

[36] Sonnentag S, Zijlstra FRH. Job characteristics and off-job activities as predictors 1037 of need for recovery, well-being, and fatigue. J Appl Psychol 2006;91(2):330-50, 1038 http://dx.doi.org/10.1037/0021-9010.91.2330. 1039

[37] Porter G. Workaholic tendencies and the high potential for stress among 1040 co-workers. Int J Stress Manag 2001;8(2):147-64. http://sfx.nelliportaali.fi/ 1041 nelli09b?sid=google&auinit=G&aulast=Porter&atitle=Workaholic+tendencies+ 1042 and+the+high+potential+for+stress+among+co-workers.&id=doi:10.1023/ 1043 A:1009581330960 [accessed 15.01.13]. 1044

[38] Schaufeli WB, Taris TW, Bakker AB. It takes two to tango. Worka- 1045 holism is working excessively and working compulsively. In: Burke 1046 RJ, Cooper CL, editors. The long working hours culture: causes, conse- 1047 quences and choices. UK: Emerald Group Publishing; 2008. p. 203-26. 1048 http://www.google.fi/books?hl=en&lr=&id=AJXnY9lqYvoC&oi=fnd&pg= 1049 PA203&dq=Schaufeli+Taris+Bakker+2008&ots=ovHxoU4u9N&sig= 1050 QDkYio9mHTNz-ibJSLb3PNTd5v4&redir_esc=y#v=onepage&q=Schaufeli% 1051 20Taris%20Bakker%202008&f=false [accessed 21.01.13]. 1052

[39] Berglass S. Treating workaholism. In: Cooms RH, editor. Handbook of addic- 1053 tive disorders: a practical guide to diagnosis and treatment. Hoboken, NJ: John 1054 Wiley & Sons; 2013. p. 383-407. 1055

[40] Bakker AB, Demerouti E, Burke R. Workaholism and relationship quality: a 1056 spillover-crossover perspective. J Occup Health Psychol 2009;14:23-33. 1057

[41] Taris TW, Schaufeli WB, Verhoeven LC. Workaholism in the Netherlands: 1058 measurement and implications for job strain and non-work conflict. J 1059 Appl Psychol 2005;54:37-60 http://www.mpopa.ro/statistica_master/ 1060 workaholism_fact_an.pdf 1061

[42] Havighurst R. Social and developmental psychology: trends influencing the 1062 future of counceling. PersGuidJ 1980;58:328-33. 1063

[43] Super DE. A life-span, life-space approach to career development. In: Brown 1064 D, Brooks DL, editors. Career choice and development: applying contem- 1065 porary theories to practice. 2nd ed. San Francisco, CA: Jossey-Bass; 1990. 1066 p. 197-261. 1067

[44] Nurmi JE, Salmela-Aro K, Koivisto P. Goal importance, and related agency 1068 beliefs and emotions during the transition from vocational school to work: 1069 antecedents and consequences. J Vocat Behav 2002;60:241-61. 1070

ti^^^M AI IIHI.E IN PRESS

H. Innanen et al. / Burnout Research xxx (2014) xxx-xxx 11

1071 [45] Burke RJ. Workaholism in organizations: psychological and phys- [70] Fritz C, Sonnentag S. Recovery, well-being, and performance-related out- 1152

1072 ical well-being consequences. Stress Med 2000;16(19):11-6, comes: the role of workload and vacation experiences. J Appl Psychol 1153

1073 http://dx.doi.org/10.1002/(SICI)1099-1700(200001). 2006;91(4):936-45. http://www.ub.uni-konstanz.de/kops/volltexte/2008/ 1154

1074 [46] Salmela-Aro K, Nurmi JE. Employees' motivational orientation and well- 5585/[accessed 20.01.13]. 1155

1075 being at work. A person-oriented approach. J Organ Change Manag [71] SluiterJK, van der Beek AJ, Frings-Dresen MW. The influence of work charac- 1156

1076 2004;17(5):471-89. teristics on the need for recovery and experienced health: a study on coach 1157

1077 [47] Greenhaus JH, Callanan GA, Godshalk VM. Career management. UK: drivers. Ergon 1999;42:573-83. 1158

1078 Sage Publisher; 2009. p. 21-37. http://www.google.fi/books?hl=en& [72] McEwen BS. Protective and damaging effects of stress mediators. N Engl J Med 1159

1079 lr=&id=XchJNXEue3oC&oi=fnd&pg=PR1&dq=+well-being+at+management+ 1998;338:171-9. 1160

1080 level+at+mid+career+stage&ots=4gtclD3-R-&sig=C3LsId_edSKPlbDnmXAo [73] Siltaloppi M, KinnunenU, FeldtT. Recovery experiences as moderators between 1161

1081 P4cSdBI&redir_esc=y#v=onepage&q&f=false [accessed 10.12.12]. psychosocial work characteristics and occupational well-being. Work Stress 1162 10Q8 [48] Hakanen J, Seppälä P. Työn imu, psykologisten perustarpeitten tyydyttymi- 2009;23(4):330-48, http://dx.doi.org/10.1080/02678370903415572. 1163

1083 nen työssäja työnyksilöllinen muokkaus. In: KauppinenT, Mattila-Holappa P, [74] Lee RT, Asfort BE. Further examination of managerial burnout: toward an inte- 1164

1084 Perkiö-Mäkelä M, Saalo A, ToikkanenJ, Tuomivaara S, et al., editors. Työja ter- grated model. J Organ Behav 1993;14:3-20. 1165

1085 veys Suomessa 2012. Seurantatietoatyöoloistaja työhyvinvoinnista. Helsinki: [75] Leiter MP, Maslach C. Areas of worklife: a structured approach to orga- 1166

1086 Finnish Institute of Occupational Health; 2013. nizational predictors of job burnout. In: Perrewe P, Ganster DC, editors. 1167

1087 [49] Pines AM. Burnout. In: Goldberg L, Breznitz S, editors. Handbook of stress: Research in occupational stress and well-being. Emotional and physi- 1168

1088 theoretical and clinical aspects. New York: The Free Press; 1993. p. 386-402. ological processes and positive intervention strategies, vol. 3. Oxford, 1169

1089 [50] Little BR, Salmela-Aro K, Phillips S, editors. Personal project pursuit: goals, England: Elsevier; 2003. p. 91-134. http://scholar.google.fi/scholar?q= 1170

1090 action, and human flourishing. Mahwah, NJ: Lawrence Erlbaum Associates; areas+of+worklife&btnG=&hl=en&as_sdt=0%2C5 [accessed 15.12.12]. 1171

1091 2007. [76] de Lange A, TarisTW, KompierM, Houtman I, Bongers P. Different mechanisms 1172

1092 [51] NurmiJE. Age differences in adult life goals, concerns, and theirtemporalexten- to explain the reversed effects of mental health on work characteristics. Scand 1173

1093 sion: a life course approach to future-oriented motivation. Int J Behav Dev J Work Environ Health 2005;31:3-14, http://dx.doi.org/10.5271/sjweh.843. 1174

1094 1992;15:487-508, http://dx.doi.org/10.1177/016502549201500404. [77] Richardsen AM, Burke RJ, Leiter MP. Occupational demands, psychological 1175

1095 [52] Brandtstädter J. Personal and social control overdevelopment: some implica- burnout, and anxiety among hospital personnel in Norway. Anxiety Stress 1176

1096 tions of an action perspective in life-span developmental psychology. In: Baltes Coping 1992;5:62-78. 1177

1097 PB, Brim Jr OG, editors. Life-span development and behavior, vol. 6. New York: [78] Leiter MP, Shaughnessy K. The areas of worklife model of burnout: tests of 1178

1098 Academic Press; 1984. p. 1-32. mediation relationships. Ergon Int J 2006;28:327-41. 1179

1099 [53] Heckhausen J, Chang EE, Greenberger E, Chen C. Striving for educational and [79] Rice RW. Organizational work and the overall quality of life. Appl Soc Psychol 1180

1100 careergoals during the transition after high school: what is beneficial? J Youth Annu 1984;5:155-78. 1181

1101 Adolesc 2013;42:1385-98. [80] Diener E, Sandvik E, Pavot W. Happiness is the frequency, not the intensity, of 1182

1102 [54] NurmiJE, Salmela-Aro K, Haavisto T. The strategy and attribution ques- positive versus negative affect. In: Strack F, Argyle M, Schwartz N, editors. Sub- 1183

1103 tionnaire: psychometric properties. Eur J Psychol Assess 1995;11(2):108-21, jective well-being: an interdisciplinary perspective. Elmsford, NY: Pergamon 1184

1104 http://dx.doi.org/10.1027/1015-5759.11.2.108. Press; 1991. p. 119-39. 1185

1105 [55] van Heuvel M, Demerouti E, Schaufeli W, Bakker AB. Personal resources and [81] Kantak D, Futrell C, Sager J. Job satisfaction and life satisfaction in a sales force. 1186

1106 work engagement in the face of change. In: HodmontJ, Stavroul L, editors. Con- J Pers Sell Sales Manag 1992;12:1-7. 1187

1107 temporary occupational health psychology: global perspectives on research [82] Maslach C, Jackson SE. The measurement of experienced burnout. J Organ Behav 1188

1108 and practice. UK: John Wiley & Sons; 2010. p. 124-50. 2007;2(2):99-113, http://dx.doi.org/10.1002/job.4030020205. 1189

1109 [56] Eronen S. Achievement and social strategies and the cumulation of [83] Robinson BE. The Work Addiction Risk Test: development of a ten- 1190

1110 positive and negative experiences during young adulthood. Helsinki: tative measure of workaholism. Percept Mot Skills 1999;88:199-210, 1191

1111 University of Helsinki; 2000. http://www.doria.fi/bitstream/handle/10024/ http://dx.doi.org/10.2466/pms.1999.88.1.199. 1192

1112 3718/achievem.pdf?sequence=2 [accessed 15.02.13]. [84] Schaufeli WB, Shimazu A, Taris TW. Being driven to work exces- 1193

1113 [57] Snyder C, Smith TW, Augelli RW, Ingram RE. On the self-serving function of sively hard: the evaluation of a two-factor measure of workaholism 1194

1114 social anxiety: shyness as a self-handicapping strategy. J Pers Soc Psychol in the Netherlands and Japan. Cross-Cult Res 2009;43(4):320-48, 1195

1115 1985;48(4):970-80. http://dx.doi.org/10.1177/1069397109337239. 1196

1116 [58] Salmela-Aro K, Tolvanen T, Nurmi JE. Achievement strategies during uni- [85] Sonnentag S, Fritz C. Recovery experience questionnaire: devel- 1197

1117 versity studies predict early career burnout and engagement.J Vocat Behav opment and validation of a measure for assessing recuperation 1198

1118 2009;75(2):162-72,http://dx.doi.org/10.1016/j.jvb.2009.03.009. and unwinding. J Occup Health Psychol 2007;12(3):204-21, 1199

1119 [59] Salmela-Aro K, Tolvanen A, Nurmi JE. Social strategies during university studies http://dx.doi.org/10.1037/1076-8998.12.3.204. 1200

1120 predictearlycareerworkburnoutandengagement: 18-yearlongitudinalstudy. [86] Diener E, Emmons RA, Larsen RJ, Griffin S. The satisfaction with life 1201

1121 JVocatBehav2011;79(1):145-57,http://dx.doi.org/10.1016/j.jvb.2011.01.002. scale. J Pers Assess 1985;49(1):71-5. http://scholar.google.fi/scholar?hl= 1202

1122 [60] Emmons RA, Kaiser HA. Goal orientation and emotional well-being: linking en&q=diener+emmons+1985&btnG=&as_sdt=1%2C5&as.sdtp= [accessed 1203

1123 goals and affect through the self. In: Martin LL, Tesser A, editors. Striving 10.02.13]. 1204

1124 and feeling. Interactions among goals, affect, and self-regulation. Mahwah, NJ: [87] Vermunt JK, Magidson J. Latent class models for classification. Com- 1205

1125 Lawrence Erlbaum Associates; 1996. p. 79-98. put Stat Data Anal 2003;41(3):531-7. http://www.statisticalinnovations. 1206

1126 [61] Scheier MF, Carver CS. Effects of optimism on psychological and physi- com/articles/csda2003.pdf [accessed 10.02.13]. 1207

1127 cal well-being: theoretical overview and empirical update. Cogn Ther Res [88] Muthen LK, Muthen BO. Mplus user's guide. 7th ed. Los Angeles, 1208

1128 1992;16:201-28. CA: Muthen & Muthen; 1998-2012. http://www.statmodel.com/download/ 1209

1129 [62] Karasek RA, Theorell T. Healthy work: stress, productivity and the reconstruc- usersguide/Mplus%20user%20guide%20Ver_7_r6_web.pdf [10.02.13; accessed 1210

1130 tion of working life. New York, NY: Basic Books; 1990. p. 89-103. 15.01.13]. 1211

1131 [63] Bakker AB, Albrecht SL, Leiter MP. Key questions regarding [89] Muthen B. Latent variable analysis: growth mixture modeling and related 1212

1132 work engagement. Work Organ Psychol 2011;20(1):4-28, techniques for longitudinal data. In: Kaplan D, editor. Handbook of quanti- 1213

1133 http://dx.doi.org/10.1080/1359432X.2010.485352. tative methodology for the social sciences. Newbury Park, CA: Sage; 2004. 1214

1134 [64] de Lange AH, Taris TW, Kompier MAJ, Houtman ILD, Bongers PM. Different p. 345-68. 1215

1135 mechanisms to explain the reversed effects of mental health on work charac- [90] Hakanen JJ, Schaufeli WB. Do burnout and work engagement pre- 1216

1136 teristics. Scand J Work Environ Health 2005;31(1):3-14. dict depressive symptoms and life satisfaction? A three-wave 1217

1137 [65] Salanova M, Bakker AB, Llorens S. Flow at work: evidence for an upward spi- seven-year prospective study. J Affect Disord 2012;141(2-3):415-24, 1218

1138 ral of personal and organizational resources. J Happiness Stud 2006;7:1-22, http://dx.doi.org/10.1016/jjad.2012.02.043. 1219

1139 http://dx.doi.org/10.1007/s10902-005-8854-8. [91] Hayes CT, Weathington BL. Optimism, stress, life satisfaction, and job burnout 1220

1140 [66] de Lange A, Taris T, Kompier M, Houtman I, Bongers P. The relationships in restaurant managers. J Psychol 2007;141(6):565-79. 1221

1141 between work characteristics and mental health: examining normal, reversed [92] Mauno S, Kinnunen U, Ruokolainen M. Job demands and resources as 1222

1142 and reciprocal relationships in a 4-wave study. Work Stress 2004;18:149-66, antecedents of work engagement: a longitudinal study. J Vocat Behav 1223

1143 http://dx.doi.org/10.1080/02678370412331270860. 2007;70(1):149-71, http://dx.doi.org/10.1016/jjvb.2006.09.002. 1224

1144 [67] Bakker AB. An evidence-based model of work engagement. Curr Dir Psychol Sci [93] NunnalyJC, BernsteinIH. Psychometrictheory.3rded. New York: McGraw-Hill; 1225

1145 2011;20:265-9. 1994. 1226

1146 [68] Tims M, Bakker AB. Job crafting: towards a new model of individual [94] Schaufeli WB, Bakker AB, Salanova M. The measurement of work engage- 1227

1147 job redesign. S Afr J Ind Psychol 2010;36(2):1-9 http://www.scielo.org.za/ ment with a short questionnaire a cross-national study. Educ Psychol Meas 1228

1148 pdf/sajip/v36n2/v36n2a03.pdf 2006;66(4):701-16. 1229

1149 [69] Sonnentag S, Kuttler L, Fritz C. Job stressors, emotional exhaustion, and

1150 need for recovery: a multi-source study on the benefits of psycho-

1151 logical detachment. J Vocat Behav 2010;76(3):355-65. http://kops.ub.uni-konstanz.de/volltexte/2010/11946/ [accessed 10.11.12].