Document is current

Article History

Received: 20 September 2024

Accepted: 22 January 2026

First Online: 3 February 2026

Competing interests

: The authors declare no competing interests.

: This study used secondary data consisting of employee reviews and associated metadata (for example, star ratings and helpful-vote counts) that were publicly accessible on Glassdoor at the time of collection. The research involved no direct interaction with human participants, no intervention or manipulation, and no collection of direct personal identifiers. On this basis, the study was not submitted for formal institutional ethics review because it was designed as a secondary analysis of publicly available online text and metadata and posed no more than minimal risk to individuals. Nevertheless, the study was conducted in line with widely used research-ethics principles for internet-mediated research in the social sciences and humanities, including proportionality, data minimization, and protection against re-identification. Privacy safeguards included analyzing and reporting findings in aggregate, avoiding any attempt to identify or contact reviewers, and not reproducing verbatim review text in ways that could reasonably enable traceability to specific individuals.

: Because the dataset comprised reviews and metadata that were already publicly posted on Glassdoor prior to the research and because the authors had no direct contact with human participants, informed consent was not sought. This approach is consistent with the use of secondary analysis of publicly accessible materials where there is no interaction with individuals, no intervention, and no collection of direct identifiers, and where risk is minimized through appropriate safeguards. To further reduce privacy risk, we did not attempt to infer identities, we present results at an aggregated level, and any shared materials for transparency or replication are de-identified and limited to derived measures (for example, linguistic features and anonymized firm identifiers) rather than redistribution of verbatim review texts.