tolerance-interval · research · also tolerance interval, statistical tolerance limits

Statistical tolerance interval

A statistical tolerance interval uses sample data to bound at least a specified fraction of a population with a specified confidence level.

Related Observatory article: Read on hesela.com

Report two separate percentages: the target population fraction and confidence in covering at least that fraction. For example, 95% confidence for at least 99% population coverage describes the sampling procedure, not certainty that every realized interval covers 99%. It differs from a confidence interval for a mean and a prediction interval for one future observation. [1] [2]

In lifetime work, a one-sided lower tolerance limit can bound a value exceeded by a specified population fraction. For a continuous lifetime distribution, a lower confidence bound for the (1-p) quantile can serve as a lower tolerance limit with at least fraction p above it. Distributional assumptions and censoring must match the method; a normal-data formula is not automatically valid for device lifetimes. Statistical tolerance limits are inferred from a population model or sampling design, not prescribed engineering acceptance limits. [1] [3]

Relations

related
Confidence interval, Prediction interval, Quantile, Survival function

Citations

  1. NIST Dataplot: Tolerance limits, population coverage and confidence (accessed 2026-11-19)
  2. NIST e-Handbook 4.1.3.2: Prediction and interval targets (accessed 2026-11-19)
  3. NIST e-Handbook 7.2.6.3: Population content, confidence and engineering tolerances (accessed 2026-11-19)

JSON: /data/terms/tolerance-interval.json

Published by Hesela team · Updated

Report a correction · Editorial standards · Reuse guidance

Cite this page
Hesela team. Statistical tolerance interval. Hesela. 2026-11-19. https://hesela.dev/terms/tolerance-interval/