Trust & Governance
Methodology
Numbers without methodology are marketing. This page explains how CRI's comparisons, tools, and datasets are built — and each one links back here plus its own specific methodology notes.
Effective and last reviewed:
Comparisons
- Comparison criteria are defined before evaluation begins and are identical for every option compared.
- Performance data comes from standards-body testing, manufacturer technical data sheets (not marketing), field research, and CRI datasets.
- Where options excel under different conditions, we present condition-based guidance rather than forcing a single “winner.”
Decision tools
- Every calculator and estimator publishes its inputs, formula or model logic, assumptions, and data vintage on an accompanying methodology section.
- Outputs are ranges with stated uncertainty, not false-precision point estimates.
- Tools are validated against published reference data before release and re-validated when underlying data updates.
Datasets
- Each dataset page documents: data sources, collection method, sample size or coverage, time period, cleaning rules, and known limitations.
- Proprietary CRI data is aggregated and anonymized; no client- or building-identifiable data is published.
- Datasets carry version numbers and change logs so citations remain stable.
Cost data
- Cost figures state their region, year, and what is included (materials, labor, tear-off, disposal, etc.).
- National averages are presented only with ranges and regional caveats — roofing costs are local.
What we won’t publish
- Point estimates we cannot defend.
- Rankings without published criteria.
- “Best of” lists influenced by commercial relationships.
Questions about any specific methodology? Contact the editorial team — methodology challenges are treated with the same seriousness as correction requests.