United States
Dental Market Data Methodology
This page documents how the numbers on this site actually work: where they come from, how they are stored, what travels with each value, and what happens when a source withholds a number. It describes the system as built — not an aspiration — and it exists so any figure here can be checked, reproduced, or challenged.
Sources: public, named, and linked
Every cited figure on this site originates from one of a small set of public datasets, each chosen because it is authoritative for its unit of analysis and publicly checkable. We do not blend in commercial data-broker estimates, model our own market sizes, or fill gaps with interpolation. If a number appears with a citation, the citation resolves to a public table or API query you can open yourself.
| Source | Publisher | What we take from it | Unit of analysis |
|---|---|---|---|
| County Business Patterns (CBP) | U.S. Census Bureau | Establishment counts, employment, and payroll for NAICS 621210 | Establishments — physical locations, not firms or dentists |
| American Community Survey (ACS) | U.S. Census Bureau | Population estimates used as denominators for density figures | Population |
| Occupational Employment and Wage Statistics (OEWS) | U.S. Bureau of Labor Statistics | Employment and wage distributions for dentist occupations | Wage-and-salary workers; excludes most self-employed owners |
| Dental HPSA designations | Health Resources & Services Administration | Federal dental-shortage designations | Designations for geographies or population groups |
| NAICS 2022 manual | OMB / Census Bureau | The classification definition of code 621210 | Industry classification text, not measurements |
Each source is the primary publisher for its measure. Adding secondary sources that repackage the same federal data would add citation surface without adding information — and would introduce exactly the vintage and unit ambiguities this methodology exists to prevent.
Provenance travels with the value, not the page
The core design decision: provenance is a property of each individual data point, not a footnote on the page. Every value in the data layer is stored with its source identifier, the dataset's reference vintage, the ISO date our ingestion retrieved it, the unit of measure, the geography level it describes, an optional note on any transformation applied, and a deep link to the exact source table or API query. When a page renders a figure, its citation line is generated from that record by a single shared formatter — no page hand-writes attribution, so a citation cannot drift from the value it describes.
- Ingest from the named sourceAn ingestion script pulls from the public source's API or published tables and writes versioned data files into the repository, stamping each value with its full provenance record including retrieval date.
- Build from the committed files onlySite builds read exclusively from those committed files — never from live endpoints. This makes every published figure reproducible: the page you see corresponds to a recorded retrieval, not to whatever an API returned at render time.
- Render through one citation pathComponents display values together with citations generated from the provenance record. Derived figures — like establishments per 100,000 residents — state both parent sources and the fact of derivation.
- Refresh by re-ingestion, visiblyWhen a source publishes a new vintage, re-running ingestion updates the files, the vintages, and the retrieval dates together. Updates are deliberate, dated events in version control, not silent drift.
Suppression is rendered, never repaired
Federal sources routinely withhold values — most often to protect the confidentiality of individual businesses in small geographies, sometimes because a measure simply isn't published at a given level. Our data layer represents these as explicit states: a value is either reported, suppressed (with the source's reason), withheld, or not collected at that geography. The structure forces every consuming component to handle the non-reported cases — there is no code path by which a suppressed value renders as a zero or a blank cell that looks like data. On the page, you'll see the limitation stated in words, with the citation still attached, because the fact that a source withheld a value is itself sourced information.
A useful audit for any data product, including ours: find a small geography and check what it shows where the source suppressed the value. An honest system says so. A dishonest or careless one shows zero — which asserts, falsely, that the count was measured and found to be nothing. This site is built to fail that coercion structurally, not by editorial vigilance.
Units, derivations, and the limits we accept
Units are encoded in the data layer as distinct types — establishments, employees, annual payroll, wages, population, designation counts — so a figure cannot silently change what it counts between storage and display. Where we derive a figure, the derivation is simple, disclosed, and built from cited parents: density divides a CBP establishment count by an ACS population estimate, and its citation names both. We accept the limits that come with this discipline. Establishment data publishes with a lag of roughly two years, and we show its vintage rather than pretending currency. OEWS wage figures exclude most practice owners, and we say so where they appear. CBP counts locations, not clinical capacity. Small-area values are often suppressed, and we render the suppression. A site willing to interpolate, blend, and estimate could show more numbers than we do. It could not show you where they came from — and that trade is the whole methodology.
What you can verify about any figure on this site
- Which public dataset it came from, via the named source in its citation
- Which reference period it describes, via the stated vintage
- When we retrieved it, via the stated retrieval date
- What it counts, via the stated unit and geography level
- The source's own record, via the deep link accompanying the citation
Frequently asked questions
Why do figures on this site sometimes look older than numbers quoted elsewhere?
Usually because we state vintages and others don't. Federal business data publishes with a lag — establishment counts describe a reference year one to two years back — and we display that vintage rather than presenting the newest available data as if it described today. Sources quoting fresher-sounding numbers are often citing the same federal vintage without the label, or using modeled estimates whose methods aren't published. We prefer a dated true number to an undated impressive one.
Do you estimate or model values that sources don't publish?
No. Where a source suppresses or doesn't publish a value, the page says so explicitly with the reason and citation. The only computed figures are simple disclosed derivations from cited parents — such as dividing an establishment count by a population estimate to get density — and those state their derivation where they appear. There is no interpolation, no gap-filling, and no proprietary scoring anywhere on the site.
Why don't the establishment counts match dentist counts from state licensing boards?
Different units measuring different things. County Business Patterns counts establishments — physical business locations — while a licensing board counts people holding active licenses, including some who don't practice, practice part-time, or work at multiple locations. Neither is wrong; they answer different questions. This site keeps units explicit on every figure precisely so these comparisons get made knowingly rather than accidentally.
How often is the data refreshed?
When underlying sources publish new vintages, we re-run ingestion, which updates the stored values, vintages, and retrieval dates together as a recorded change. Because pages build from committed data files rather than live API calls, a figure never changes silently between visits — any update corresponds to a deliberate re-ingestion you can see reflected in the citation's retrieval date.
Can I cite this site's figures in my own work?
You can, but the better practice — and the one this site is designed to enable — is to cite the primary source directly: every figure links to the exact public table or query it came from, with vintage and unit stated. Cite Census CBP, ACS, BLS OEWS, or HRSA as appropriate, carry the vintage and unit into your citation, and use this site as the finding aid that got you there.
Related on Offices of Dentists
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.