Job Demands and Resources May Not Measure Up Equally Across Academic Career Stages, Study Finds

Burnout among researchers has become one of the most talked-about crises in modern science, but a sweeping new study suggests that the psychological tools used to understand it may not work the same way for everyone. In a large cross-sectional survey of 3,101 researchers across 27 European countries, investigators tested whether the widely used Job Demands-Resources (JD-R) framework — a model that explains burnout and work engagement through the balance of workplace stressors and supports — operates equivalently for junior and senior academics. The answer, published in the International Journal of Public Health, is more complicated than anyone expected.

The JD-R framework has long been a cornerstone of occupational health psychology. It proposes two parallel processes: a health-impairment pathway, in which excessive job demands such as heavy workloads, conflict, and insecurity gradually deplete employees’ physical and psychological reserves and push them toward burnout; and a motivational pathway, in which job resources such as autonomy, role clarity, and community support fuel engagement and buffer the damage caused by demands. Meta-analytic evidence has supported these core relationships across diverse employment sectors, from healthcare to industry to public service. But whether the same measurement model holds true across the very different worlds of a PhD candidate and a full professor has remained an open question.

To find out, researchers from the Researcher Mental Health Observatory (ReMO), a European Union COST Action, collected data between September 2023 and August 2024 using the Sustainable Working Conditions in Academia (STAIRCASE) survey. Participants were researchers holding active positions at universities and non-university research organisations, classified into four career stages: R1 (PhD candidates), R2 (postdoctoral researchers), R3 (senior researchers), and R4 (full professors). The team measured burnout using the 12-item Burnout Assessment Tool (BAT-12), work engagement with the ultra-short Utrecht Work Engagement Scale (UWES-3), and four job demands — illegitimate tasks, job insecurity, interpersonal conflict, and work-family conflict — alongside four job resources: role clarity, job control, sense of community, and predictability at work.

The raw numbers told a familiar story. Burnout prevalence was higher among junior researchers, at 29.5 percent, than among senior researchers, at 25.1 percent, a statistically significant difference. Junior researchers were also far more likely to hold contracts ending within 12 months — 31.6 percent of them, compared with just 5.6 percent of senior researchers — and 38.1 percent of junior researchers held contracts shorter than one year. These figures align with a growing body of evidence that early-career academics face a uniquely precarious employment landscape, with short-term contracts, limited funding opportunities, and delayed career stability contributing to elevated occupational stress.

But the technical heart of the study — structural equation modelling, or SEM — revealed something unexpected. When the researchers tested whether their measurement instruments functioned equivalently across career stages, using a multigroup SEM approach that examined configural, metric, and scalar invariance, the models failed. The scalar invariance model, which assumes that both factor loadings and item intercepts are equal across groups, showed poor fit: RMSEA of 0.101, CFI of 0.770, and TLI of 0.746, all well below accepted thresholds. In practical terms, this means that the survey items did not measure the same underlying constructs in the same way for junior and senior researchers. A question about job insecurity, for instance, may carry different psychological weight for a PhD candidate facing a contract ending in six months than for a professor with a permanent position.

Because full measurement invariance was not established, the team could not make valid direct comparisons of latent factor means across groups. Instead, they estimated separate single-group SEMs for junior and senior researchers. Within each group, the structural associations were broadly consistent with JD-R theory: job demands were positively associated with burnout (β = 0.17 for juniors, β = 0.24 for seniors), while job resources were negatively associated with burnout (β = −0.38 for juniors, β = −0.33 for seniors) and positively associated with work engagement (β = 0.46 for juniors, β = 0.47 for seniors). However, model fit was only moderate to poor in both groups — CFI values of 0.82 and 0.84, RMSEA of 0.09 — indicating that the specified models did not fully reproduce the observed covariance structure in the data.

The authors are careful about what these findings do and do not mean. The lack of measurement invariance, they stress, does not demonstrate that the JD-R framework itself lacks generalizability to academic researchers. Rather, it indicates that the specific operationalization used in this study — a particular set of survey items combined into latent constructs — did not function equivalently across the two career-stage groups. The framework is deliberately broad and flexible, allowing demands and resources to be defined according to occupational and organizational context, and previous multi-sector studies have found that the strength of JD-R relationships does not differ significantly across employment settings even when levels of demands and resources do. The present results reflect challenges in applying a common measurement model to a highly heterogeneous, multi-country sample, not evidence that the underlying theory operates differently across career stages.

Several factors may have contributed to the measurement problems. The sample was recruited through convenience and snowball sampling across 27 countries, and country-level clustering, institutional characteristics, and disciplinary context were not modelled. Such heterogeneity may influence how specific demands and resources are experienced and interpreted. Additionally, two indicators — overwork and emotional support — were excluded from the final measurement model due to excessive missing data, potentially narrowing the content coverage of the respective constructs. A sensitivity analysis using full-information maximum likelihood did not converge to a stable solution, supporting the exclusion-based approach. The cross-sectional design also prevents causal inference and cannot rule out reverse causation, in which burnout itself increases perceived demands or reduces access to resources.

Despite these limitations, the study’s strengths are considerable. The large sample, cross-national coverage, use of established and validated measures, and — crucially — the explicit formal testing of measurement invariance rather than assuming it, provide a valuable assessment of the challenges involved in applying the JD-R framework to academic researchers. The findings also carry practical weight: they suggest that interventions aimed at reducing burnout in academia may need to be tailored to career stage, since the psychological meaning of job insecurity, work-family conflict, and other demands may differ substantially between a doctoral student and a full professor. Role clarity and sense of community showed relatively high loadings on the job resources construct in both groups, consistent with previous academic research linking clarity about responsibilities and belonging to a scholarly community with lower burnout and greater wellbeing.

The authors call for career-stage-specific, longitudinal, and multilevel approaches to establish measurement equivalence before testing the JD-R framework’s generalizability in academia. Future research should examine variation in JD-R processes across career stages while accounting for country, institutional, disciplinary, and employment contexts, and should explore whether JD-R processes vary by academic discipline, given differing workload structures and funding pressures across fields. Until such equivalence is established, the study stands as a cautionary tale for occupational health researchers: even the most widely supported frameworks in psychology may require careful recalibration when applied to the diverse and rapidly evolving world of academic work, where the experience of a PhD candidate and that of a full professor may be separated by far more than seniority.

Subject of Research: Do job demands and resources generalize across academic career stages? Implications for burnout and work engagement

Article Title: Do job demands and resources generalize across academic career stages? Implications for burnout and work engagement

Article References: Do job demands and resources generalize across academic career stages? Implications for burnout and work engagement. (n.d.). https://doi.org/10.3389/ijph.2026.1610023

Image Credits: AI Generated

DOI: 10.3389/ijph.2026.1610023

Keywords: demands, resources, generalize, across, academic, career, stages, Implications, burnout, work, engagement, scientific research