Multilevel and Configurational Perspective on Psychosocial Risk Assessment: Advancing Evidence-Based Workplace Health Evaluation

Psychosocial risk assessment (PRA) has become legally mandated in many European countries to evaluate and manage work-related health risks. Despite its practical relevance, methodological challenges remain in linking scientific evidence to PRA practice. This dissertation addresses three such limitations. First, a level discrepancy exists between predominantly individual-level research evidence and the group-based implementation in practice, where aggregated assessments guide action requirements. Second, although variable-centered approaches provide robust evidence for main effects of psychosocial job characteristics, findings on interaction effects remain inconsistent. Consequently, PRA practice often evaluates job characteristics separately using isolated scale scores. Third, PRA practice typically applies uniform evaluation criteria across occupational contexts, potentially overlooking that the relevance and interplay of psychosocial job characteristics may vary across work domains. Three empirical studies address these limitations using complementary methodological approaches and data from actual PRAs across multiple organizations. Study 1 employs multilevel modeling to assess how group-aggregates of psychosocial demands and resources relate to individual health impairment and to examine the role of within-group agreement in these relationships. Study 2 applies fuzzy-set Qualitative Comparative Analysis (fsQCA) to identify configurations of psychosocial job characteristics associated with burnout and evaluates these configurations in the second half of a split-sample using multilevel model. Study 3 adopts an intraorganizational comparative design to contrast risk configurations in production and non-production domains, providing evidence for the context dependence of psychosocial risk patterns. Across studies, the findings indicate that aggregated group assessments can inform individual risk, that configurational approaches capture complexity, and that context-sensitive evaluation can enhance the precision of risk identification. At the same time, the cross-sectional and self-report design across all studies limits causal inference and may involve shared method variance, which reflects the reliance on real-world PRA data. By integrating multilevel and configurational perspectives, the dissertation strengthens the methodological foundation of PRA and advances theoretical understanding by clarifying how psychosocial job characteristics operate across levels, combine in conjunctural patterns, and vary across occupational contexts.