Index / Disclaimer

Disclaimer

An educational tool, not validated software — plus every convention and approximation documented.

Last updated: 20 July 2026

DatrixCalc is an educational tool, not validated statistical software. The instruments here have not been through the verification and validation process applied to R, SAS, SPSS or Stata. Use them to learn a method, to check working, and to sanity-test a figure. For anything submitted, published, or relied on in a regulated or high-stakes context, reproduce the result in software your institution or field recognises and cite that.

A calculator validates arithmetic, not reasoning

This is the most important thing on the page. Every instrument here will faithfully apply its formula to whatever you paste in. None of them can tell you whether the formula was the right one, whether your sample represents the population you care about, whether your observations are independent, or whether the conclusion you are about to draw follows from the number produced.

The errors that matter most in statistics happen before any calculation:

  • Correlation read as causation. A regression line has no opinion about which variable causes which, or whether a third variable drives both.
  • Biased sampling. No amount of correct arithmetic rescues a self-selected sample. Your confidence interval will be a precise statement about a group that does not represent your population.
  • The wrong measure or the wrong test. A mean on strongly skewed data is technically correct and substantively misleading.
  • Assumption violations. Methods here assume independence, and in places approximate normality or a minimum sample size. Applying them where those conditions fail produces a confident wrong answer.
  • Multiple comparisons. Testing many hypotheses and reporting the one that reached significance is not a finding.

Methods and conventions used

Where more than one defensible convention exists, we have picked one and documented it. If your answer differs from another tool, this table usually explains why.

QuantityMethod usedNote
Quartiles and percentilesLinear interpolation between order statisticsMatches R type 7 and Excel PERCENTILE.INC. Minitab and some textbooks differ, visibly so on small samples.
Standard deviationBoth offered; sample (n − 1) is the defaultUsing the population formula on a sample understates spread and inflates apparent significance downstream.
Normal CDFAbramowitz & Stegun 7.1.26Absolute error under 1.5 × 10−7 — tighter than any printed table.
t critical valuesExact tabulated to df = 30; linear interpolation between anchors at 30, 40, 60, 120; normal beyondInterpolation error above df = 40 is smaller than fourth-decimal rounding.
Proportion intervalsNormal (Wald) approximationRequires np̂ and n(1−p̂) both ≥ 10. The tool warns when this fails. Near the boundary use Wilson or Clopper–Pearson.
RegressionOrdinary least squares, single predictorAssumes linearity, independence, constant variance and roughly normal residuals.

Precision and rounding

Results display to four significant decimal places by default, switching to scientific notation for very large or very small magnitudes. Full double precision is retained internally throughout every calculation — only the display is rounded, so no rounding error accumulates between steps. Do not read displayed precision as measurement precision: reporting a standard deviation to four decimals from whole-number inputs implies an accuracy the data does not contain.

Sample size warnings

Several instruments flag small samples, and those flags are worth taking seriously. A slope fitted to six points, a confidence interval from n = 4, or an outlier judged by a z-score on twelve observations are all technically computable and practically unreliable. Where a tool tells you the estimate is not yet stable, it is not being cautious for form's sake.

Academic use

Using this site to understand a method, check your working, or verify a figure is ordinary study. Presenting its output as your own manual calculation, where an assessment requires you to demonstrate the working yourself, may breach your institution's academic integrity rules. That judgement is yours and your institution's. We accept no responsibility for how the site is used in assessed work.

No professional relationship

DatrixCalc is not a statistical consultancy. We do not review study designs, recommend tests for particular data, or interpret results in any field. Nothing on this site constitutes professional advice for your specific project, and using it creates no advisory relationship. Statistical results in medical, clinical, financial, legal or regulatory contexts should always be produced and reviewed by a qualified professional.

Limitation of liability

To the fullest extent permitted by law, DatrixCalc and its operators accept no liability for any loss or damage arising from use of, or reliance on, results or information on this site — including incorrect research conclusions, rejected submissions, failed assessments, or commercial decisions. Your use of the site is at your own risk. This disclaimer forms part of our terms of use and should be read alongside our privacy policy.

Report an error

If you believe a calculation here is wrong, please tell us immediately through the contact form or at corrections@datrixcalc.com, including the exact inputs you used. A statistics site with an arithmetic error is worse than no site at all, and corrections take priority over every other kind of work.