For several decades, occupational health researchers have been preoccupied with factors at work that cause stress and infirmity. However, recent findings indicate that individuals' sense of self and personal worth is increasingly dependent on work, which has become ‘a source of social integration, recreation and especially friendship, in addition to economic security’ (Roos, Trigg, & Hartman, 2006, p. 209). These benefits might have extensive effects on employee health. Work engagement, defined as ‘a positive, fulfilling, affective motivational state of work-related well-being’ (Bakker, Schaufeli, Leiter, & Taris, 2008, p. 187), reflects the recent trend toward ‘positive psychology’, where the concern is the positive aspects of employee health (Seligman & Csikszentmihalyi, 2000; Seligman, Steen, Park, & Peterson, 2005). This more positive vision implies a move beyond pathology to focus on understanding and promoting healthy functioning. Based on conservation of resources (COR) theory (Hobfoll, 1989, 1998, 2001), the present study explored the possible reciprocal relationship between work engagement and physiological symptoms of anxiety and depression. COR theory, a promising explanatory model of burnout (Lee & Ashforth, 1996), encompasses several theories of stress but extends these theories through a resource perspective.
Work engagement and symptoms of anxiety and depression
The basic tenet of COR theory is that people have a deeply rooted motivation to obtain, retain and protect what they value, labelled resources (Hobfoll, 1989). The innovative aspect of COR theory is that it describes not only what individuals do when confronted with stress but also how they behave in the absence of threats. Specifically, when confronted with stress, individuals are predicted by the model to strive to minimize the net loss of resources. Conversely, when not confronted with threats, people strive to develop resource surpluses to offset the possibility of future loss. Whereas the COR theory describes burnout as a state of extreme resource depletion (Hobfoll & Shirom, 2001; Neveu, 2007), work engagement might be regarded as a resource surplus. When people develop resource surpluses, they are likely to experience positive well-being and health.
In occupational health psychology research, work engagement is seen as an antithesis to the more familiar and investigated term, ‘burnout.’ In the present study, work engagement was operationalized and assessed by the Oldenburg Burnout Inventory (OLBI; Demerouti & Bakker, 2008), which agrees with the theoretical argumentation that burnout and engagement are two opposite poles of one continuum (Maslach & Leiter, 1997, 2008). The OLBI includes two dimensions: one ranging from exhaustion to vigour and a second ranging from disengagement to dedication (Demerouti & Bakker, 2008; Halbesleben & Demerouti, 2005). Together, they provide two bi-polar dimensions—energy and identification (Demerouti & Bakker, 2008). Recent research suggests that vigour and dedication constitute the core dimensions of engagement (e.g. González-Romá, Schaufeli, Bakker, & Lloreta, 2006). Vigour is characterized by high levels of energy and mental resilience while working, whereas dedication is characterized by a sense of significance, enthusiasm, inspiration and pride (Demerouti & Bakker, 2008). Research focusing on engagement is still in its infancy, and there are many questions that still need to be answered. According to Bakker et al. (2008), there is particularly a dearth of research on the relationship between work engagement and health.
The European Commission's Green Paper on mental health [COM (2005) 484] states that anxiety and depression are two of the most prevalent mental ill health problems facing European citizens today. Although anxiety and depression are closely related, they are regarded as separate disorders because of differences in presenting characteristics (see Barlow, 2000). There is substantial literature documenting that stress and stressful life events constitute a risk for depression and anxiety disorders. For example, McLaughlin and Hatzenbuehler (2009) found that stressful life events were longitudinally related to anxiety symptoms in a diverse sample of adolescents. Equally, comparisons of depressed patients with non-depressed controls indicates that, besides reporting more stressful events than the controls, depressed people also report more life strain associated with health, family, housing and work (Billings, Cronkite, & Moos, 1983). Demerouti, Bakker, and Bulters (2004) show that a person who feels distressed at one point in time is likely to be distressed at a later point in time unless a significant event changes their emotional state.
Just as life strain associated with health, family, housing and work may lead to depression and anxiety in the long run (Billings et al., 1983; McLaughlin & Hatzenbuehler, 2009) we argue that joy and satisfaction associated with these same life areas might ease symptoms of depression and anxiety. Judge and Locke (1993) suggest that job satisfaction influences subjective well-being through the centrality of work to an individual's life. Since most individuals spend the majority of their waking hours at work, a positive evaluation will have extensive effect on judgments of happiness and well-being overall. For example, Fredrickson's (1998) ‘broaden-and-build model’ of positive emotions suggests that positive emotions broaden the individual's attentional focus and behavioural repertoire and, as a consequence, build social, intellectual and physical resources. Similarly, the COR theory anticipates that positive experiences or resources are likely to accumulate, creating a positive spiral of resources, which is likely to have positive health-promoting effects. Such resource caravans might provide health-promoting effects because they may do the following: (1) undo some of the negative physiological effects associated with negative emotions; (2) positively influence the neuroendocrine system; and/or (3) interrupt and short-circuit the rumination spiral of stressful circumstances and prevent the decline into clinical depression (see Folkman & Moskowitz, 2000). In a study among older adults, Glass, Mendes deLeon, Bassuk, and Berkman (2006) found social engagement not only to be independently associated with depressive symptoms when examined cross-sectionally but also to be associated with change in depressive symptoms among people who were not depressed at the baseline.
Previous research has found support for both reciprocal gain spirals (Llorens, Schaufeli, Bakker, & Salanova, 2007) and resource caravans (Hakanen, Perhoniemi, & Toppinen-Tanner, 2008) of work engagement. However, neither Llorens et al. nor Hakanen, Perhoniemi et al. have examined the possible health-promoting effect of work engagement on symptoms of anxiety and depression.
Most closely related is Hakanen, Schaufeli, and Ahola's (2008) study that tested the longitudinal association between burnout and depression and between work engagement and commitment, respectively. Unfortunately, they did not test the possible association between work engagement and depression directly even though their correlational data indicated a negative relationship between them. To the authors' knowledge, the possible analogue relationship between work engagement and anxiety has not been examined previously. The aim of the present study was to explore the longitudinal relationship between work engagement (vigour and dedication) and symptoms of anxiety and depression. Based on the assumption of the COR theory and previous studies, we formulated the following hypotheses:
Hypothesis 1. Vigour at Time 1 has negative cross-lagged effects on anxiety at Time 2.
Hypothesis 2. Vigour at Time 1 has negative cross-lagged effects on depression at Time 2.
Hypothesis 3. Dedication at Time 1 has negative cross-lagged effects on anxiety at Time 2.
Hypothesis 4. Dedication at Time 1 has negative cross-lagged effects on depression at Time 2.
However, the study of the association between work engagement and depression and between work engagement and anxiety is complicated by the possibility of reversed causation and the effect of unmeasured third variables (see de Lange, Taris, Kompier, Houtman, & Bongers, 2004; Dormann, 2001; Zapf, Dormann, & Frese, 1996). People with depression or anxiety often lose interest in previously valued social roles, and the feeling of engagement in work might fade. Billings et al. (1983) suggest that depressed patients are doubly disadvantaged since not only do they suffer from more social isolation, but also their lower level of social resources illustrates less stress-buffering potential relative to that available to community controls. Therefore, initial losses beget further losses (Hobfoll, 1989, 1998, 2001). Judge and Locke (1993) supported a reciprocal relationship between job satisfaction and subjective well-being among a sample of clerical staff working at the university. Based on the cognitive theory of depression, which focuses on an individual's thought processes, they argued that subjective well-being affects job satisfaction through the way individuals collect and recall information about their jobs. For example, people suffering from anxiety or depression might store, evaluate or recall information about their jobs in a dysfunctional matter ensuing in less work engagement.
Previous studies on self-reported stressor–strain relations hypothesized that the results may be explained partly by a third variable such as negative affectivity (e.g. Burke, Brief, & George, 1993). Unmeasured third variables may threaten the internal validity of non-experimental studies. In most instances, relevant third variables are unknown. An unknown third variable can have a stable character such as time pressure or unstable character such as the weather. For instance, the impact of good weather on an individual's emotional state (i.e. occasion factors) may reduce the level of depressive symptoms or increase the feeling of work engagement, respectively. Dormann (2001) proposed a ‘less restrictive synchronous common factor’ (LRSCF) model to investigate the potential impact of unmeasured third variables as a source of spuriousness. The LRSCF models are described in more detail in the ‘Method’ section.
By utilizing a 2-year longitudinal panel design and advanced statistical analyses such as structural equation modelling (SEM), the present study also tested for reversed cross-lagged relationship between the study variables and the potential effect of unmeasured third variables. Hence, the following hypotheses were formulated:
Hypothesis 5. Anxiety at Time 1 has negative cross-lagged effects on vigour at Time 2.
Hypothesis 6. Anxiety at Time 1 has negative cross-lagged effects on dedication at Time 2.
Hypothesis 7. Depression at Time 1 has negative cross-lagged effects on vigour at Time 2.
Hypothesis 8. Depression at Time 1 has negative cross-lagged effects on dedication at Time 2.
Hypothesis 5. Hypotheses 1–4 are valid even if unmeasured third variables are taken into account.