Overview
ISO 11453:1996 - "Statistical interpretation of data - Tests and confidence intervals relating to proportions" specifies standardized statistical procedures for estimating and testing proportions. The standard defines how to compute point estimates, one- and two‑sided confidence limits, and significance tests for a single proportion or for comparisons between two proportions, and it gives guidance on sample‑size determination and operating characteristics (power, type I/II errors).
Keywords: ISO 11453, confidence intervals, proportions, statistical interpretation, significance tests, sample size determination, two-sided confidence limits.
Key topics
- Point estimator: the estimator for the population proportion is p̂ = x / n (x = number of target items in a sample of size n).
- Confidence intervals: procedures for one‑sided and two‑sided confidence limits (Forms A‑1 to A‑3). Small-sample (n ≤ 30) lookups use tabulated values; large-sample (n > 30) use asymptotic approximations with recommended corrections. Two‑sided intervals are built from one‑sided limits using α/2.
- Significance tests: procedures for testing a proportion against a specified value (Forms B‑1 to B‑3) and for comparing two proportions (Forms C‑1 to C‑3). The standard covers one‑sided and two‑sided null hypotheses and explicit decision rules (reject / do not reject).
- Sample size determination: guidance for choosing sample sizes so achieved significance and power meet specified levels. Methods include use of the Larson nomograph, Haseman tables for common cases, and approximations (including arc‑sine transformation) when tables do not apply.
- Operating characteristics: Annex A describes computation of achieved significance, power (1−β), and error probabilities for the prescribed tests.
- Forms and worked examples: standardized forms (A, B, C) and Annex B examples guide practical application and record keeping.
- Normative reference: ISO 3534‑1:1993 (statistics vocabulary and symbols) is cited for definitions and symbols.
Applications and users
ISO 11453 is practical for:
- Quality engineers and statisticians performing proportion estimates and pass/fail analysis in manufacturing and conformity assessment.
- Clinical researchers and epidemiologists comparing binary outcomes (e.g., incidence rates).
- Laboratories, certification bodies and regulatory agencies that require reproducible, standardized hypothesis tests and confidence intervals for proportions.
- Survey analysts and market researchers estimating proportions from sample data.
Practical value: ensures consistent decision rules, accounts for discrete sampling distributions (achieved significance nearest ≤ nominal), and provides small‑sample and large‑sample procedures to maintain valid inference.
Related standards
- ISO 3534‑1:1993 - Statistics: vocabulary and symbols (normative reference).
- Prepared by ISO/TC 69 (Applications of statistical methods), Subcommittee SC 3.