Work Characteristics Across Occupations: Consensus, Variance, and Classification

Occupational psychology frequently aggregates self-reported work characteristics to higher-level units like occupations, assuming them to represent a distinct entity. However, concerns persist regarding reliability and generalizability within similar and across distinct occupational contexts. Data from a multi-site sample of 2,070 employees spanning six pairs of both similar blue- and white-collar occupations were analyzed to determine whether occupations can truly function as higher-level entities for self-reported work characteristics. We apply a stepwise approach to evaluate the consistency of self-reports within, and their distinctiveness across different occupations: (1) Interrater agreement via ICC(2) and rWG, (2) variance analyses of homogeneity within and heterogeneity between occupations, (3) and a random forest classifier assessing whether employees can be assigned to their original occupations based on self-reports. Some variables failed to reach recommended cutoffs for interrater agreement, while others displayed divergent results across ICC(2) and rWG, making it challenging to establish a shared understanding. Variance analyses further indicated that some occupations did not exhibit homogeneous within-group variance or clear between-group differences. And although classification performance was above chance, machine learning further supported the notion that, while occupations do capture some degree of shared perceptions of work characteristics, misclassification rates remained substantial. The findings suggest that only select work characteristics, rather than an entire occupation, may form a coherent higher-level unit. The study highlights the need for caution when interpreting aggregated self-reports to inform theory or guide interventions in work design, and for refining measurement and analytical strategies to ensure meaningful conclusions in both research and practice.