Overview
ISO 3494:1976, "Statistical interpretation of data - Power of tests relating to means and variances," is an ISO technical standard that extends ISO 2854 by introducing the type II risk (β) and the concept of power for common hypothesis tests. It provides practical tools - notably sets of operating‑characteristic curves - to evaluate the probability of failing to reject a false null hypothesis and to determine required sample sizes for given alternatives and acceptable β values. The standard applies when the underlying populations are assumed normal and covers both means and variances, with cases for known and unknown population variance.
Key topics
- Type II risk (β) and power (1 − β) - formal definitions and interpretation in the context of comparison tests.
- Comparison tests covered:
- Comparison of a mean with a specified value (variance known and variance unknown).
- Comparison of two means (variance known and variance unknown).
- Comparison of a variance with a specified value.
- Comparison of two variances.
- Operating characteristic (OC) curves - sets of curves that give β as a function of the alternative parameter, for commonly used significance levels (α = 0.05 and α = 0.01) and for one‑sided and two‑sided tests.
- Sample size determination - procedures and curve sets to select sample sizes that meet specified power (or β) for a given alternative.
- Practical cautions - when variance is unknown, β can only be estimated approximately; ISO 3494 recommends accounting for uncertainty in σ (for example by using confidence limits per ISO 2854).
Applications
ISO 3494 is useful for practitioners who need to plan or interpret statistical hypothesis tests where control of both type I and type II risks is important:
- Quality engineers and production control - deciding sample sizes for acceptance testing or comparing batch means/variances.
- Laboratory and testing organizations - specifying test sensitivity and designing experiments to detect meaningful differences.
- Statisticians and data analysts - evaluating test power for clinical, industrial, and research comparisons of means and variances.
- Standards writers and auditors - ensuring statistical tests quoted in technical specifications meet agreed risk and power criteria.
Practical benefits include clearer communication of consumer/producer risk trade‑offs, better experiment/sample planning, and reproducible decision rules based on OC curves.
Related standards
- ISO 2854 - "Statistical interpretation of data - Techniques of estimation and tests relating to means and variances" (basis for test definitions and tables referenced by ISO 3494).
- ISO 3534 - Statistics vocabulary and symbols (defines β and related terminology referenced in ISO 3494).
Keywords: ISO 3494, power of tests, type II risk, β, operating characteristic curves, hypothesis testing, sample size determination, comparison of means and variances, variance known/unknown, ISO 2854.