Double-edged effects of work task stress on safety performance: a cognitive appraisal perspective

• Work task stress has double-edged effects on safety performance. • Challenge and hindrance appraisals mediate the work task stress–safety performance relationship. • Vocational identity weakens the work task stress–hindrance appraisal pathway. • No significant moderation of the stress–challenge appraisal pathway. In high-risk industries, enhancing employees’ safety performance remains a central concern for both practitioners and scholars. However, evidence on the effects of work task stress is inconsistent: some studies suggest that stress can motivate better safety performance, whereas others provide evidence that the accumulation of industry-inherent pressures and role demands undermines it. This debate underscores uncertainty about the mechanisms through which work task stress influences safety outcomes, particularly the role of cognitive appraisals. Building on cognitive appraisal theory and the job demands-resources (JD-R) model, we propose a dual-pathway mediation model in which work task stress influences safety compliance and safety participation through two competing appraisal processes. A field survey of 308 frontline employees in high-risk industries reveals that work task stress exhibits double-edged effects: it improves both safety compliance and safety participation through challenge appraisals while simultaneously diminishing them through hindrance appraisals. Moreover, vocational identity significantly attenuates the hindrance pathway but does not reliably enhance the positive effects of challenge appraisals. This research challenges the traditional either-or assumption of stress appraisals by adopting a dual-pathway framework, which offers a new lens for explaining the mechanisms linking stress to safety performance in safety critical contexts. Simultaneously, the asymmetric moderating effect of vocational identity enriches theoretical understanding of the distinct functions of personal resources in the JD-R model.