Psyche: Packed Dataset of Psychology Studies Including Unfinished Work

A compressed snapshot of the Psyche dataset containing every study regardless of status, including unfinished and known-broken ones. The dataset is maintained by psyche-org and was last updated on 2026-07-21. It includes multiple file formats such as Parquet and JSONL, with compressed versions offering significant size reductions.

Use Cases

• Assessing data completeness and study status based on the included STATUS.md files
• Analyzing experimental designs and prompts from the compressed JSONL files
• Comparing data processing efficiency using the different compressed file formats (Parquet, JSONL.gz)
• Researching the lifecycle of psychological studies, including broken or unfinished work

Strengths

• Contains every study regardless of status, providing a complete raw snapshot
• File compression offers significant size reductions, with Parquet files noted as ~15-22× smaller than original CSVs
• Includes explicit warnings to check study status before use, indicating transparency

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

psyche-org

Collection Method

Likely aggregated from multiple psychology studies, though the specific collection method is not detailed.

Freshness

Last updated 2026-07-21 04:19:18; freshness should be verified