What does this dataset measure?

For every calculator on Do The Calculation, it records whether the page names the published formula it implements, which body issued that formula, and what kind of source it is: a government publication, a technical standard, a peer-reviewed paper, or a clinical guideline.

Can I reuse the data?

Yes, under the Creative Commons Attribution 4.0 licence. Copy, redistribute, adapt, and build on it for any purpose including commercial use, provided you credit Do The Calculation and link to this report.

Why publish the calculators that are not yet cited?

Publishing only the cited ones would turn an audit into an advertisement. The uncited rows are the more useful half of the dataset, because they show the size of the gap rather than hiding it.

How often is it updated?

The files are regenerated from the live registries whenever the underlying data changes. Each edition is a dated snapshot and earlier editions are never edited in place, so a figure quoted from one edition stays valid.

2026 edition · CC BY 4.0

Calculator Formula Provenance

As of October 2026, 168 of 526 calculators (31.9%) name the published source their formula comes from, with 207 citations to 171 distinct works.

Most free online calculators give you a number and no way to check it. This dataset is our own audit of the opposite discipline: for each of our calculators, whether the page names the published formula it implements, and which body issued it.

We publish the uncited rows too. An audit that reported only the successes would be marketing, and the size of the remaining gap is the more useful figure.

526

Calculators

168

Cite a source

207

Citations

171

Distinct sources

Download CSV Download JSON summary

1. What the data shows

Of 526 calculators, 168 name at least one published source, or 31.9%. Those pages carry 207 citations drawn from 171 distinct works.

The gap is not evenly distributed, and that is the interesting part. Tools implementing a specific published equation, a clinical guideline, or a statutory rule are far more likely to cite one, because there is a single correct thing to cite. Tools that perform ordinary arithmetic, such as a percentage or a bill split, have no primary source to name, and adding one would be padding rather than provenance.

Category	Calculators	Cited	Share
finance	51	25	49.0%
health	53	44	83.0%
business	50	2	4.0%
math	57	6	10.5%
utility	50	22	44.0%
green	52	18	34.6%
tech	50	14	28.0%
property	55	9	16.4%
marketing	51	5	9.8%
hobby	57	23	40.4%

2. Where the formulas come from

Sorted by the kind of body that issued them, the citations break down as follows. Government publications lead because tax, lending, and emissions rules are set by statute and published openly; peer-reviewed papers dominate the health tools, where the equation itself was a research result.

Government publication - 69 citations
Technical standard - 46 citations
Peer-reviewed paper - 43 citations
Clinical guideline - 27 citations
Industry reference - 22 citations

The most-cited issuing bodies across the whole dataset:

National Institute of Standards and Technology - 14
Internal Revenue Service - 12
Consumer Financial Protection Bureau - 10
Internet Engineering Task Force - 9
American College of Obstetricians and Gynecologists - 6
U.S. Environmental Protection Agency - 5
US Environmental Protection Agency - 5
Mifflin MD, St Jeor ST, et al. American Journal of Clinical Nutrition, 1990;51(2):241–247 - 4
International Organization for Standardization - 4
Grepow - 4

3. How it was produced

Every row is generated directly from the site’s own registries by a build script, not compiled by hand, so the dataset cannot drift from the pages it describes. One row is emitted per calculator-source pair, and calculators with no cited source appear as a single row with the flag set to false.

A source counts only where it genuinely determines the formula the calculator evaluates. A loosely related link would inflate the headline figure and make every other citation in the dataset less trustworthy, so those are excluded rather than counted.

4. What this does not measure

It covers one site. It says nothing about how often other calculator sites cite their sources, because measuring that fairly would require auditing them under the same rule rather than sampling them.

It measures whether a source is named, not whether the implementation is faithful to it. A page can cite the right paper and still contain an arithmetic error, which is what the test suite described in our methodology exists to catch.

An uncited calculator is not necessarily unsourced. Many implement arithmetic with no primary source to name. The uncited count is therefore an upper bound on the work outstanding, not a defect count.

5. Licence and citation

This dataset is published under the Creative Commons Attribution 4.0 International licence. You may share and adapt it for any purpose, including commercially, provided you give credit and link back.

Do The Calculation (2026). Calculator Formula Provenance, 2026 edition. https://dothecalculation.com/reports/calculator-formula-provenance-2026. Licensed CC BY 4.0.

Files regenerated 2026-10-04. Earlier editions are never edited in place.