Before claiming your research is reproducible, verify that another researcher could independently replicate your results using only the information you've provided—this means checking your documentation, data availability, code functionality, and methodological transparency.
Documentation Completeness
Review whether your methods section includes sufficient detail for replication. This means specifying exact versions of software, libraries, and tools used, not just their names. Document all parameter settings, random seeds, and preprocessing steps. If you made any judgment calls during analysis, explain the reasoning so others can make consistent decisions.
Data and Materials Accessibility
Confirm that all data, code, and materials are actually available in the format you claim. Test your data repository links to ensure they work. If using restricted-access data, provide clear instructions for how others can obtain the same dataset. For computational work, include sample datasets that demonstrate your pipeline without requiring full data access.
Code Verification
Run your analysis code in a fresh environment—ideally on a different machine or in a clean container. This catches undocumented dependencies and hard-coded paths that only work on your system. Check that your code produces the same outputs (within acceptable numerical tolerance) when rerun. If you're working with experimental protocols, consider having a colleague attempt the procedure using only your written instructions.
Environmental Specifications
Document your computational environment completely: operating system, hardware specifications (especially for GPU-dependent work), and any system-level configurations. For wet-lab work, specify equipment models, reagent lot numbers, and environmental conditions like temperature and humidity where relevant.
Statistical Reproducibility
Distinguish between computational reproducibility (same code + data = same output) and statistical reproducibility (same methods + different sample = consistent conclusions). Report confidence intervals and effect sizes, not just p-values. If your results depend on random sampling, demonstrate stability across multiple runs.
For complex experimental setups, PaperClaw can help by parsing published methods sections and generating structured reproduction plans—useful for verifying that your own documentation meets reproducibility standards before submission.
