A quality result is only worth signing if you can trust how it was calculated. Every tool on this site is checked against published standards, worked examples, hand-checked cases and specialised statistical software, and the checks run again on every change.
2,916automated checks
15tools verified
4 November 2026last full run
How we verify
1
Published worked examples
We run the same data as the standard or the textbook, and our result must equal the printed answer at the precision it is printed.
2
An independent second calculation
The same datasets go through specialised statistical software and independent statistical libraries, and every value is compared one by one.
3
Run again on every change
All checks run automatically on every update to the site, and a failing check is flagged straight away.
What is checked, tool by tool
Check counts come from the latest run, not from a document.
SPC Engine and control charts (I-MR, X̄-R, X̄-S)
221 checks
AIAG SPCMontgomery
Variable control charts: control limits, tests for special causes, Cp, Cpk, Pp, Ppk, Cpm (including a target off the midpoint), PPM and the normality test, on 12 runs. Attribute charts (p, np, c, u): centre line, control limits (stepped when subgroup sizes differ) and the tests for special causes, on 6 datasets. All compared value by value with specialised statistical software. The SPC Engine and the three chart pages are checked to give identical numbers.
Both methods (Average & Range and ANOVA) against the manual’s worked example, and 7 studies (the manual’s example among them) compared with specialised statistical software: variance components, % study variation, % tolerance, ndc, F and P.
AIAG MSA worked example (ANOVA)%GRR 27.9% · ndc 4Our result%GRR 27.9% · ndc 4
Every value of the graphical summary — quartiles, skewness, kurtosis, normality test and the confidence intervals for the mean, median and standard deviation — on 7 datasets from 3 to 100 readings, compared with specialised statistical software and an independent statistical library.
Readings 1 to 10 · 95% CI for the median2.6577 – 8.3423Our result2.6577 – 8.3423
Coefficients, standard errors, ANOVA and R² against the certified values of the U.S. National Institute of Standards and Technology, and p-values, intervals, VIF and residual diagnostics against an independent statistical library.
Effects, sums of squares, F and p with replicates, Lenth’s method without replicates, R², R²(pred) and the equation in your units, on the textbook’s worked examples, compared with published results and an independent statistical library.
Montgomery Example 6.2 (filtration rate)effect A 21.625 · PSE 2.625Our resulteffect A 21.625 · PSE 2.625
The sample-size code letters and every cell of the normal, tightened and reduced master tables that the tool uses (AQL 0.010 to 6.5), read from the 2026 edition, plus the standard’s Annex A examples.
Tightened inspection · letter J · AQL 0.40L · n 200 · Ac 1 · Re 2Our resultL · n 200 · Ac 1 · Re 2
Model choice between batches, the one-sided confidence bound and the shelf life from the data, on published potency data, and the extrapolation limits ICH Q1E allows, checked against the guideline text.
LeBlond et al. (2011) potency data26.22 monthsOur result26.22 months