Work engagement has been proven to be an empirically distinct construct from workaholism [5], [6]. Work engagement is defined as a positive, fullfilling, work-related state of mind characterized by vigor (energy, concentration, strain, persistence in the face of inconviniencies), dedication (inspiration and challenge, full of work enthusiasm) and absorption (like an enduring flow-experience) [7]. Although the concept of work engagement was clearly defined, the concept of workaholism remained vague because of two reasons. First, work engagement is still considered by some scholars as a possible dimension of workaholism (e.g., [8]). Second, recent conceptualisations of workaholism propose obsessive inner drive as the core characteristic of workaholism and neglect its addictive nature (e.g., [7]). To avoid the vagueness of the concept of “workaholism”, we explicitly differentiate between work engagement and work addiction. Thus, in the present study, we adopted the conceptualization of workaholism as work craving [9], [10].

Wojdylo [9] proposed the theory of work addiction, which she called work craving. She argued that the core characteristics of work addiction are not fully explained by obsessive inner drive but constitute other addictive mechanisms. In her view, a main mechanism of work addiction is the compensatory function of emotions, which explains the inner obsessive drive of workaholics in fulfilling unrealistic standards of perfectionism. Thus, in contrast to previous theories of workaholism (e.g. [11], [7], [8]), the conceptualization of work craving constitutes a synthesis of obsessive-compulsive and addictive elements. Specifically, in addition to the compulsive/behavioral component, work craving theory includes two hedonic/affective components (anticipation of self-worth compensation and anticipation of reduction of negative affect or withdrawal symptoms resulting from working), and a learning component (neurotic perfectionism) [9], [10]. Thus, work craving is defined as an emotional-motivational state oriented at compensation of negative emotions through an obsessive-compulsive work style and a desire for unrealistic (neurotic) perfectionistic standards.

Studies have shown that work engagement and workaholism have distinct regulatory mechanisms and outcomes. Workaholism and work engagement share the behavioral component (working excessively hard, high work involvement), but the emotional and motivational aspects of these phenomena differ fundamentally. Whereas workaholics are motivated by an obsessive inner drive they cannot resist, engaged employees are intrinsically motivated, have a sense of energetic and effective connection with their work activities, and view themselves as able to deal well with the demands of their jobs [12]. These findings imply that workaholism and work engagement are differently related with self-regulation competencies (i.e., self-relaxation and self-motivation). According to work craving theory, the emotion-regulatory components inherent in work craving (i.e., anticipation of reduction of negative emotions and desire for self-worth compensatory incentives) imply that work craving should derive from low rather than high self-regulatory competencies [10].

As mentioned above, studies have also shown that workaholism and work engagement are related with distinct outcomes. Work engagement is related to desirable outcomes like job and life satisfaction, better job performance and negatively associated with ill being [13], [14]). Workaholism, in contrast, is related to negative outcomes like psychosomatic symptoms, mental and physical health complaints [15], [2], [16], [14], [10], poor emotional well-being [17], increased work-family conflict [18], and low life satisfaction [19], [20]). In the present study, we therefore expected work craving to be negatively associated with health (H1a) and work engagement to be positively associated with health (H1b).

The ability to self-regulate one's feelings and thoughts plays an important role in personality functioning [21], [22] and can be differentiated into two major self-regulation competencies: self-relaxation and self-motivation [23], [24], [25]. Individual differences in self-relaxation and self-motivation competencies can be assessed by the failure- and decision-related action control scales, respectively [26]. In these scales, high levels of self-regulation competencies are called action orientation whereas low levels are called state orientation and indicate a low action orientation.

Failure-related action orientation (AOF) is the ability to reduce the negative affect during or after confrontation with failure or threat (high self-relaxation) and to keep up or even enhance access to the self (i.e., implicit representations of own wishes, goals, and preferences). Thus, it promotes the formation of realistic and self-congruent goals and, in turn, well-being especially under threatening conditions (e.g., [27], [28], [29]). In contrast, failure-related state orientation (the low end of the AOF scale) denotes a low ability to self-regulate negative affect (low self-relaxation) and is characterized by ruminative thoughts. Studies found relationships between failure-related state orientation and psychosomatic symptoms, depression, low self-esteem, and self-consciousness [27], [30], [31] as well as workaholism [32].

Decision-related action orientation (AOD) is the ability to up-regulate positive affect, to overcome feelings of listlessness, and to foster confidence and enthusiasm despite the presence of challenging demands and difficulties (high self-motivation). Thus, it is a decisive precondition for an efficient translation of intentions into action. In contrast, decision-related state orientation (the low end of the AOD scale) denotes a low ability to self-generate positive affect (low self-motivation) and it characterized by hesitation. Studies have linked decision-related state orientation to reduced well-being [27], a low ability to behave according to one's preferences [33], procrastination [34] and workaholism [32].

Hence, we hypothesize that self-relaxation and self- motivation competencies are negatively associated with work craving (H2a) and positively with work engagement (H2b). Furthermore, we hypothesize that self-relaxation and self-motivation competencies are positively related with health indices (H3). Finally, we consider work styles as mediators between self-regulation competencies and health. Despite the correlational nature of many of the reviewed findings, self-regulation competencies are better conceived of as antecedents rather than consequences of work styles and health because failure- and decision-related action orientation are personality dispositions that develop during early childhood and are rather stable over time [25], [26]. In addition, longitudinal findings indicate that work styles precede rather than follow from health states [14]. We hypothesize that the relationship between self-relaxation and self-motivation competencies and good health is partially mediated by work engagement (H4a), and that the relationship between self-relaxation deficits and self-motivation deficits and poor health is partially mediated by work craving (H4b).

According to the view of lay people, working a lot can be equated to workaholism. From our point of view, working hours and workaholism have to be separated. Wojdylo et al. [10] empirically showed that work cravers and work engagers did indeed work an equal number of hours. Thus, differences between work craving and work engagement cannot be attributed to different hours of working but, instead, to different work styles alone. We aspired to replicate these findings, so we also explored the relationship between working hours and the two different work styles of work craving and work engagement. We hypothesize that the relationship between work craving and number of worked hours is low (H5).

The present study is a part of the Work Craving International Project (WCIP), a large cohort research realized in Poland and Germany (“Work craving – personality antecedents and regulatory mechanisms”). The WCIP aims at examining the new conceptualization of workaholism as work craving and its personality mechanisms [10].

Four hundred and sixty-nine teachers from the Federal State Rheinland-Pfalz of Germany participated in the present study (345 women and 121 men, 3 participants gave no information on gender). Participants were recruited via e-mail to all school principals across Rheinland-Pfalz. The principles were asked to distribute the information among the staff. Participants completed the questionnaire voluntarily online which took about 45 minutes. Participants' average age was 45 years (SD  =  10.57), with a range of 21 to 64 years. 62.5 % were married, 20.5 % were living in a relationship, 14.9 % were living alone, and 1.9 % did not provide information about their marital status. 36.7 % were teaching in primary schools, 14.7 % in secondary/graduate schools, 16.2 % in special schools, 15.6 % in vocational schools, and 16.8% in other school forms. 63.1 % were employed full-time, 17.3 % worked 30 hours a week, 11.7 % were employed part-time, 6 % worked less than part-time, and 1.9 % did not give information regarding their employment.

All questionnaires were administered online in a German version.

Table 1 presents an overview of the descriptive results and correlations for the study variables. As expected, work craving correlated negatively (H1a) and work engagement positively with health (H1b). Second, self-regulatory competencies (AOF and AOD) correlated negatively with work craving (H2a) and positively with work engagement (H2b). Third, self-relaxation (AOF) and self-motivation (AOD) showed significantly positive correlations with health (H3) and with each other. Contrary to previous findings [10], there was a significant relationship between work craving and working hours, but the relationship was very weak (r  = .10) indicating that work craving cannot be inferred from working hours (H5). Finally, in compliance with previous findings [10], the correlation between work engagement and work craving was very small indicating that they are distinct work styles.

To further test our hypotheses regarding meditational effects (H4a and H4b), we used structural equation modeling (SEM). We designed a path-model (Figure 1, Model A) allowing us to test the four mediation assumptions at the same time: that work craving mediates between self-relaxation deficits and poor health and between self-motivation deficits and poor health whereas work engagement mediates between self-relaxation competencies and good health and between self-motivation competencies and good health. The Model Fit was estimated by means of Mplus6 [39]. The regression coefficients were adjoined to the hypothetical Model A in Figure 1. For Model A, good fit indices were achieved (CFI  = .96; SRMR  = .04; χ²(1)  =  17.997). The cut off ranges for the absolute fit indices were achieved (cf. [40], [41]). Looking at the path coefficients of Model A, work craving was associated with lower general health whereas AOD, AOF, and work engagement were associated with higher general health. Most important for our theoretical expectations, only the paths from AOF to work craving and from AOD to work engagement were significant, whereas the paths from AOF to work engagement and from AOD to work craving were low and not significant.

According to these results, we tested an alternative model (Figure 1, Model B) with only two indirect paths at the same time: that work craving mediates between self-relaxation deficits and poor health whereas work engagement mediates between self-motivation competencies and good health. Model B reached with a CFI  = .997, SRMR  = .019, and χ²(2)  =  3.27 a generally good fit. We found significant direct effects from AOF and AOD on health (β_AOF  = .20, p <.01; β_AOD  = .11, p <.01). Consistent with expectations, we also found significant indirect effects from AOF through work craving on health (H4a: β_{AOF_WCS}  =  -.19, p <.05) and from AOD through work engagement on health (H4b: β_{AOD_UWES}  =  −.08, p <.05).

When comparing the model fit of the competing models A and B, Model B reached a better fit. According to the chi square difference test, the difference between the models was significant (Δχ²(1)  =  14.72, p <.001. The findings indicated that the relationship between self-relaxation deficits (but not by self-motivation deficits) and poor health was partially mediated by work craving whereas the relationship between self-motivation competencies (but not self-relaxation competencies) and good health was partially mediated by work engagement.

The main limitation of the current study is its cross-sectional nature that precludes cause–effect relationships being uncovered. Emotional self-regulation competencies are usually considered as relative stable personality dispositions that develop in early childhood [24], [25]. Therefore, it is unlikely that the scores on our self-regulation measures are caused by work craving/work engagement or health. In contrast, the causal direction of the association between health and work craving/work engagement is less clear. Although previous research longitudinally established that high workaholism preceded poor health and high work engagement preceded good health [14], we cannot exclude the possibility that unhealthy workers display more workaholic behaviors and healthy workers display more enthusiastic behaviors. These notions can only be tested adequately using a longitudinal design.

Second, the current data set was drawn from a very specific group of workers (i.e., teachers). Although this particular group has frequently been assumed to be a high-risk group for health problems [48], [49], it should be noted that the fact that all participants belonged to this high-risk group could well have led to a restriction in the variance of the study variables. The unique nature of the present sample underlines the need to replicate the current findings on samples of workers from different occupations.

Third, in the present studies, only self-reports were used to measure the constructs. Thus, we cannot preclude an impact of social desirability on the data. For instance, respondents could try to minimize or maximize their problems by under- or over-reporting the severity and the frequency of symptoms. Workaholics might also deny their condition in the same way as people suffering from alcoholism deny drinking. Thus, our present findings should be validated with more objective measures or with additional ratings from family members regarding criteria for work craving.

As for the practical implications of the present study, it is informative to see that not only current motivational states such as controlled/autonomous motivation [43] or introjected/identified motivation (cf. [42]) but also individual differences in the ability to self-regulate emotions (i.e., action vs. state orientation) are strongly associated with workaholism and work engagement. The role of self-regulation in the etiology of work craving/work engagement has received less attention than would be warranted when considering the strength of the relationships. Especially a lack of self-relaxation was strongly associated with work craving, suggesting that the concepts of self-regulatory competencies may be a good starting point for interventions at the individual and organizational level to prevent and to treat workaholism. The theory and findings on self-regulation can help to further understand the causes of workaholic behavior, while psychotherapeutic techniques - especially those focused on working with emotions, like emotion-focused therapy [50], introvision methods [51], or schema-focused therapy [52] - may be used in assisting workaholic clients to overcome their self-regulatory deficits and internal conflicts.

In conclusion, the present findings may contribute to the development of effective interventions that may help people suffering from work craving to reduce their tendency to “live to work” and promote their ability to “love to work”. The fact that work craving and work engagement exhibit different patterns of possible causes and consequences implies that different intervention strategies should be used when work craving is to be reduced or work engagement is to be enhanced.