How similar is similar enough? Job profile similarity benchmarks using occupational information network data

Abstract

Job comparison research is critical to many human resources initiatives, such as transporting validity evidence. Job analysis methods often focus on critical attribute (e.g., tasks, work behaviors) overlap when assessing similarity, but profile similarity metrics represent an alternative or complementary approach for job comparisons. This paper utilizes Occupational Information Network (O*NET) data to establish a distribution of job profile correlations across all job pairs for five attributes – generalized work activities, knowledge, skills, abilities, and work styles. These correlations represent effect sizes, or degree of shared variance between jobs. Practitioners may reference these correlational distributions as benchmarks for gauging the practical significance of the observed degree of similarity between two jobs of interest compared to the broader world of work.