Research methodology

Evidence-confidence language

All operational evidence is labeled Verified (directly supported by reliable evidence), Estimated (a reasonable estimate with known limitations), Unknown (not currently known), or Data Gap (missing evidence that itself becomes a tracking priority). We never guess to fill a data gap — in research or on any page of this site.

Definitions before numbers

Every published metric must carry its formula, denominator, cohort, and time period. A number whose denominator is undisclosed is not publishable. This is the same standard our tools enforce on their own outputs.

Data sources and consent

Future original studies will use only appropriately consented, aggregated, de-identified operational data. No client names, Medicare identifiers, policy numbers, health information, or other sensitive personal data are collected by public tools on this site, and none will appear in research in any form.

Disclosure requirements

Any published study must disclose:

• sample size, composition, and how participants were selected;
• operational definitions for every metric;
• known limitations and sources of bias;
• what the data cannot support — stated as plainly as what it can;
• the review date on which the findings will be re-examined.

External sources

Where we cite third-party research, we cite the original source, link to it, name its publication date, and quote its wording rather than rounding it up. If a claim cannot be traced to a source that supports its exact wording, it does not ship — see the editorial & evidence policy and corrections process.