The core rule
Use Mellio as a research tool. Before using any result in a manuscript, report, presentation, decision, or public claim, review the output against your data, analysis plan, assumptions, and field-specific reporting standards.
What Mellio is for
Mellio is built for statistical workflows around results, tables, and figures. Depending on the workflow, you may analyze data in the browser, send fitted results from R with mellio_open(), edit tables, create figures from supported results, or upload a workbook that already contains values you want to visualize.
Mellio supports selected research workflows while additional object support is expanded. Mellio for R 1.0.0 supports common models, tests, tables, figures, and selected optional-package results.
The product goal is to preserve the connection between analysis, presentation, and interpretation so outputs are easier to inspect and reuse.
Accuracy and validation
Mellio is developed with comparison tests against established statistical behavior, including R or reference implementations where appropriate. Those tests are important, but no statistical application can guarantee that every result is correct for every dataset, model, package version, option, or reporting convention.
Differences can occur because of rounding, missing-data handling, contrast coding, factor levels, default options, model specification, package versions, data cleaning choices, or unsupported result formats. If something looks different from another tool, treat it as something to inspect rather than ignore.
Your responsibilities
- Choose analyses that match your research design and data structure.
- Check assumptions, exclusions, missing-data handling, variable coding, and model specifications.
- Review tables, figures, APA-style text, exported files, and generated result cards before using them outside Mellio.
- Do not rely on Mellio as the sole basis for clinical, legal, financial, safety, policy, or other high-stakes decisions.
- Keep source data, scripts, and analysis notes so results can be independently checked when needed.
Reporting possible issues
If you find a discrepancy, please report it with the analysis type, Mellio output, expected output, browser, and enough non-sensitive example data or source output to reproduce the issue. Until a discrepancy is understood, use your verified source analysis as the authority.