Item-Level RCT Datasets
Summary
To support my interests in psychometrics and causal inference and to enhance research in this area generally, I have curated a collection of empirical item-level outcome data from RCTs. The datasets are publicly accessible in a harmonized format on the Item Response Warehouse (IRW). At present, the RCT datasets in the IRW span over 100 outcome measures across dozens of RCTs in education, psychology, economics, and health. The datasets all begin with the prefix gilbert_meta in the IRW for ease of access and analysis.
Derivative Research
These item-level datasets have supported research on:
- item-level heterogeneous treatment effects,
- sensitivity of value-added and average treatment effect estimates to measurement and preprocessing choices,
- attenuation bias and outcome scaling,
- socially desirable responding in intervention studies.
Publications using these data include:
- Gilbert et al. (2025), Journal of Policy Analysis and Management
- Gilbert et al. (2025), Journal of Educational and Behavioral Statistics
- Gilbert et al. (2026), Educational Measurement: Issues and Practice
- Gilbert (2025), Methodology
- Halpin and Gilbert (2024), preprint
- Soland and Gilbert (2026), preprint
- Veltri and Gilbert (2026), preprint
Citation Guidance
When using this work, please cite the IRW paper, the original primary studies for each dataset, and my Journal of Policy Analysis and Management paper where I first gathered these data.