California, expanding nationally
The Dental Board Methodology
This document is the standard The Dental Board holds itself to. It defines what we measure, what we refuse to measure, where our data comes from, how derived scores are constructed, and how the methodology itself changes over time. If a number appears anywhere on this site and this document cannot account for it, that is a defect — report it through the corrections process.
Scope: what is measured, and what is refused
The Dental Board publishes two kinds of information. The first is cited public data — government and other public datasets about the dental market, reproduced with their source, vintage, and known limitations attached. The second is observational research on patient-facing digital experience: properties of practice websites, findability, and appointment access that any member of the public could verify by looking. Both share one property that defines this site: everything published is either traceable to a named source or reproducible by an independent observer following the stated procedure.
Nothing on this site measures clinical quality. We have no access to outcomes, no chart audits, and no clinical review capability — so no score, benchmark, or award here should be read as a statement about the quality of care a practice delivers. Any page that could be misread that way carries this boundary explicitly.
The refusal is deliberate, not modest. Rating clinical quality without clinical evidence is the core dishonesty of most 'best dentist' content on the internet: an authoritative-looking rank built on nothing a reader could check. Our position is that a narrow measurement done honestly is worth more than a broad one done by implication.
Data inputs and their provenance
| Input class | Source type | How provenance is handled |
|---|---|---|
| Public market data | Government statistical datasets (for example, Census business data and population estimates) | Rendered through the site's data layer with the source name, dataset, and vintage attached; suppressed or unavailable values are displayed as such |
| Observable digital signals | Direct observation of public, patient-facing properties — websites, listings, appointment pathways | Each observation is dated; the observation procedure is documented so an independent reviewer could repeat it |
| Practice-verified information | Corrections and verifications submitted by practices with evidence | Marked as verified, with the verification date; superseded values are corrected under the corrections policy, not silently overwritten |
| Derived scores and indices | Computed from the classes above using published formulas | Components and weights are published; a derived value is never displayed where its inputs are unavailable |
Anecdote, unverifiable third-party claims, undisclosed vendor-supplied figures, and any input whose inclusion was influenced by a commercial relationship. If an input class is not in the table above, it is not in the data.
How derived scores are constructed
Where the site publishes a composite measure — a digital-experience score, a benchmark percentile — the construction follows fixed rules. Each component must itself be an observable or cited input. The weighting between components is published alongside the score. Comparisons are made within defined peer groups (geography and practice type), never across unlike populations. And missing data propagates as missing: a practice with an unobservable component receives no score for the affected measure, rather than an imputed or zeroed value that would quietly punish it for our lack of data.
The rules every published score must satisfy
- Every component is independently observable or carries a citation
- Component weights are published with the score
- Peer groups for comparison are defined before scoring, not after
- Missing input produces a 'not scored' state, never a coerced zero
- No commercial relationship — sponsorship, advertising, or otherwise — can alter any component, weight, or result
- The scoring procedure is described in enough detail that an independent analyst could reproduce the result from the same inputs
Versioning and change management
- Methodology versions are datedThe current methodology carries the date at the top of this page. Material changes — a new component, a changed weight, a redefined peer group — produce a new dated version, not a silent edit.
- Changes are logged in plain languageEach material revision is recorded with what changed, why, and what it affects, so a reader comparing a score across time can determine whether a movement reflects the practice or the method.
- Re-scoring is applied consistentlyWhen the method changes, affected measures are recomputed under the new version across the whole peer group at the same time. We do not mix versions within a comparison.
- Errors route through correctionsFactual errors in inputs or computation are handled under the corrections policy, including a visible correction note. Disagreement with the method itself is welcome input to the next version, but it is not a correction.
Known limitations
A methodology that hides its limitations is marketing. These are ours, stated plainly.
Read every measure on this site with these in mind
- Digital-experience measures describe the patient-facing digital layer only — a practice can excel clinically with a weak website, and the reverse
- Public datasets lag: business and population data reflect their collection vintage, not the current month
- Government datasets apply their own suppression and disclosure-avoidance rules, which we surface rather than fill in
- Observations are point-in-time; a practice may have changed since the observation date shown
- Peer-group comparisons depend on classification, and edge cases (multi-specialty, multi-location) are classified by documented rules that cannot be perfect
Frequently asked questions
Do The Dental Board's scores measure the clinical quality of a dentist?
No, categorically. The site has no access to clinical outcomes or records, so its measures are limited to cited public data and observable patient-facing digital signals. A high or low score describes the digital experience and market context, never the standard of care — and any use of our measures to imply otherwise misrepresents them.
Can a practice pay to improve its score or benchmark position?
No. Commercial relationships — sponsorship, advertising, or any other payment — are firewalled from measurement by rule: they cannot alter a component, a weight, or a result. Where sponsorship exists anywhere on the site it is disclosed as such, and the methodology requires that a paying and non-paying practice with identical observable inputs receive identical scores.
What happens to a score when some of the underlying data is missing?
The affected measure is not published for that practice or geography. Missing input produces a 'not scored' or 'not available' state — it is never imputed, averaged in, or treated as zero, because a coerced zero would penalize the subject for a gap in our observation rather than anything about them.
How often are the site's measures refreshed?
On two different clocks. Cited public datasets update when their issuing agencies publish new vintages, and the displayed citation always shows which vintage you are reading. Observational measures carry their observation date and are re-observed on a rolling basis. A displayed date beside a figure is part of the figure — a number without its vintage is not something we publish.
How do I challenge a number I believe is wrong?
Through the corrections process, which covers factual errors in inputs, observations, and computation, and requires evidence we can verify. Corrections that are accepted produce a visible correction note rather than a silent edit. Disagreement with the methodology itself — what we chose to measure or how we weight it — is welcome feedback for the next methodology version, but it is handled as input to the standard, not as a correction to a fact.
Related on The Dental Board
How we handle this information
We keep material limitations visible, separate advertising from editorial judgment, and avoid inventing live scores or recommendations when the underlying evidence is not available.
Related in this network
Related properties may share common ownership. A cross-property link is not an endorsement — see our ownership disclosures.