McNemar's Test Calculator

Background.

McNemar's Test Calculator evaluates exact two-sided McNemar p value from the two discordant cell counts of paired binary observations. The page keeps every model input visible and uses the relationship p = 2 × P[Binomial(b+c, 0.5) ≤ min(b,c)], capped at 1. It is designed for a transparent calculation where the quantities have already been measured or selected from an appropriate source. It does not choose a material, operating condition, reference state, or empirical coefficient on the user's behalf.

Enter discordant count b, discordant count c in the units printed beside the fields. These are coherent SI quantities, so the displayed equation can be followed without a hidden unit factor. A result is only comparable with another source when the same quantity definitions, reference conditions, and sign or magnitude convention are used. Record those conditions whenever the number supports engineering, laboratory, or coursework decisions.

The calculator performs arithmetic with Decimal.js and rounds once at the output boundary to twelve significant digits. That protects very small and very large scientific results from early decimal-place rounding. The tests do more than pin one example: they check the dimensional scaling implied by each variable, finite and positive domain guards, several orders of magnitude, and the formula-engine registration used by the live page.

Exact equal-discordance test for paired binary observations; it ignores concordant cells by design and does not establish pairing quality, independence between pairs, effect size, confidence interval, multiplicity, or causal interpretation. The scope statement appears beside the numerical result because it changes how the answer may be used. A neat number does not remove uncertainty in measurements, material properties, geometry, calibration, or the assumptions used to reduce a real system to one equation.

Use scaling as a quick reasonableness check. If an input appears in the numerator, increasing it should move the result in the same direction; a denominator should move it in the opposite direction; a square-root term changes more slowly. If the page behaves differently from the displayed relationship, stop and review the units. The calculator rejects zero, negative, infinite, and nonnumeric quantities where the equation requires a positive magnitude.

What is mcnemar's test calculator?

McNemar's Test Calculator is a transparent implementation of p = 2 × P[Binomial(b+c, 0.5) ≤ min(b,c)], capped at 1 for exact two-sided McNemar p value from the two discordant cell counts of paired binary observations.

How to use this calculator.

1. 01Confirm that the displayed quantity equation matches the model you intend to use.
2. 02Convert every measurement to the SI unit printed beside its field.
3. 03Enter sourced magnitudes and keep their reference conditions with the result.
4. 04Read the numeric result together with the model-scope output.
5. 05Round the reported value to the uncertainty supported by the inputs.

The formula.

The implementation evaluates p = 2 × P[Binomial(b+c, 0.5) ≤ min(b,c)], capped at 1 with Decimal.js. Inputs are required to be finite and positive because this page treats them as magnitudes. Arithmetic is not rounded between operations; each numeric output is rounded once to twelve significant digits. The scaling tests independently verify the power of every input in the equation.

A worked example.

Frequently asked questions.

References& sources.

1. [1]McNemar Q. Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika. 1947.
2. [2]NIST/SEMATECH e-Handbook, McNemar Test.
3. [3]BIPM, The International System of Units (SI Brochure), 9th ed., version 3.01, coherent derived units and quantity equations.