Abstract :
[en] Purpose: The present study aimed to validate the job demands-resources model (Bakker & Demerouti, 2017) among a representative sample of the worker population in Luxembourg. Moreover, our purpose was to identify which specific job demands and resources contribute the most to burnout and work engagement, respectively. Design: Data were collected via computer assisted telephone and web interview in a large sample of 1689 employees working in Luxembourg (55.2% male, Mage = 44.1, SDage = 10.3). Most participants worked in academic professions (31.4%, n = 531), followed by technicians and associate professionals (24.0%, n = 406), clerical support workers (11.7%, n = 197) and others (32.86%, n = 555). We employed the Quality of Work Index – Luxembourg (QoWIL) to measure several areas of work, including work intensity, job design, physical and social conditions (Sischka & Steffgen, 2019). Additionally, different employment conditions were measured to get an indicator of the employment quality in Luxembourg. Findings: Results of latent moderated structural equation modelling (LMS) indicated a good fit of the model to the data, χ2(411) = 1738.017, RMSEA = .04 (95% CI = 0.04 - 0.05), CFI =.92, SRMR = .06. All job resources (i.e. social support, autonomy and job security) significantly predicted work engagement, whereas all demands (i.e. workplace mobbing, work-life inference, emotional demands) significantly predicted burnout. Particularly, social support was the most important resource (ß = .29, p < .001, R2 = 11.4%), followed by job security (ß = .17, p < .001, R2 = 3.9%) and autonomy (ß = .11, p < .001, R2 = 1.4%). Workplace mobbing explained the largest percentage of variance in burnout (ß =. 47, p < .001, R2 = 41.6%), followed by work home inference (ß = .30, p < .001, R2 = 13.0%) and emotional demands (ß = 15, p < .001, R2 = 2.2%). While burnout had a negative impact on job performance (ß = -18, p < .001), work engagement did not predict the latter (ß = .07, p > .05). Besides, only one out of nine hypothesized interaction effects had a significant effect on work engagement (i.e. social support x mobbing, ß = 0.15, p < .001) and on burnout (i.e. social support x emotional demands, ß = -0.08, p < .05). Conclusion: Whereas the present findings provided strong support for the motivational and health impairment processes proposed by the JDR model, we found limited support for the interaction hypotheses. The results outline the importance of social conditions in explaining employees’ health, illustrating important starting points for organizational interventions that aim to promote well-being. Contributions: While other studies have tested the propositions of the JDR model by focusing on different work sectors, the present study includes a more comprehensive range of occupations,classified according to the ISCO-08. Given its large data set, it provides enough statistical support to detect interaction effects and allows for the correction of measurement errors using LMS. Furthermore, it follows the parsimony principle by specifying the most important starting points for interventions across occupations.