LLM-based Qualitative Coding of Psychological Safety (Replication Package – Anonymous Version 1.1)

This replication package contains all datasets, prompt configurations, model-generated results, and the pseudo code supporting the study “LLM-based Qualitative Coding of Psychological Safety in Software Engineering Communities on Stack Exchange: An Exploratory Study.”

The package includes:

The human-coded baseline dataset (DataBase_PsychologicalSafety_GoldStandard.xlsx) based on Edmondson’s (1999) behavioral framework.

Six prompt configurations (P01–P06) used for closed, open, and thematic coding phases.

Corresponding model outputs from ChatGPT-4o and DeepSeek-V3, including agreement scores, subcodes, and thematic structures.

A detailed pseudo code file documenting the full methodological workflow.

This material enables full transparency and reproducibility of the LLM-based exploratory workflow with human oversight.During double-blind review, all identifying information has been removed.The public version with DOI will be released upon paper acceptance.