Overview
ISO/TR 22914:2020 - "Statistical methods for implementation of Six Sigma - Selected illustration of analysis of variance" provides a practical, example-driven guide to using analysis of variance (ANOVA) in Six Sigma projects. The Technical Report focuses on one-way and two-way ANOVA for fixed-effect models in balanced designs, illustrating how to partition observed variance to test differences among group means and guide process-improvement decisions. Note: unbalanced designs, random effects and nested designs are explicitly excluded.
Key topics and technical requirements
- Scope and model type: Fixed-effect ANOVA for balanced data (one-way and two-way classifications).
- Data planning and measurement: Guidance on data collection plans, variable descriptions and measurement-system considerations to ensure valid ANOVA inputs.
- Assumption checks: Steps to verify ANOVA assumptions - independence, normality (residuals), and homogeneity of variances (homoscedasticity) - plus outlier detection and strategies for non‑standard cases.
- ANOVA procedure: Practical workflow including stating hypotheses, graphical analysis, computing sums of squares, F-tests, p-values, and residual analysis.
- Further analysis & conclusions: Guidance on post-hoc interpretation and next steps for process improvement.
- Illustrative annexes: Multiple domain examples (e.g., bond strength; effect of script and training on income per sale; welded joint strength; water consumption in a petroleum enterprise; hub total hours) and Annex F containing ANOVA formulae.
- Normative references: Uses ISO 3534-1:2006 and ISO 3534-3:2013 for terms and symbols.
Practical applications and users
ISO/TR 22914:2020 is designed for practitioners who apply statistical methods within Six Sigma and continuous improvement programs:
- Six Sigma Black/Green Belts and project teams using ANOVA to identify factors that significantly affect process performance.
- Quality engineers, industrial engineers, and statisticians seeking standardized, example-based guidance for fixed-effect ANOVA in balanced experiments.
- Data analysts and researchers in manufacturing, services, petroleum, welding, sales and other sectors where comparing group means and allocating variance across factors informs decisions.
Practical uses include testing process changes, comparing supplier batches, evaluating training or script effects, and prioritizing improvement actions based on statistically significant factor effects.
Related standards
- ISO 3534-1:2006 - Statistics: Vocabulary and symbols (general statistical terms)
- ISO 3534-3:2013 - Statistics: Vocabulary and symbols (design of experiments)
Keywords: ISO/TR 22914:2020, ANOVA, analysis of variance, Six Sigma, one-way ANOVA, two-way ANOVA, fixed effects, balanced design, F-test, residual analysis, homoscedasticity.