Overview
ISO/TR 16705:2016 - "Statistical methods for implementation of Six Sigma - Selected illustrations of contingency table analysis" provides guidance on designing, conducting and interpreting contingency table analysis for categorical data. The technical report focuses on two-dimensional (r × c) contingency tables and illustrates procedures with worked examples from diverse fields. It bridges Six Sigma practice and statistical standards by showing pragmatic workflows and software-oriented illustrations for analysing associations between nominal and ordinal variables.
Key Topics
- Scope: Analysis of relationships between two categorical variables (nominal and ordinal); only two-way contingency tables are considered.
- Structured workflow: Six basic steps - state objective, list attributes, specify null hypothesis, sampling plan, process and analyse data, conclude.
- Sampling plans: Discussion of four sampling scenarios (random counts; fixed n; fixed marginal totals for rows or columns; fixed both row and column margins) and implications for analysis.
- Statistical tests and methods:
- Chi-square (χ²) test for independence and goodness-of-fit (formula and interpretation included).
- Linear trend test for ordinal associations.
- Correspondence analysis to explore and visualise associations between categorical levels.
- Practical guidance: Emphasis on randomization, confounding factors, and how to prepare categorical (including binned quantitative) data for analysis.
- Illustrative annexes: Annexes A–D contain application examples (technical issues after product release, perception studies, customer satisfaction, nonconforming parts) demonstrating real-world use and software workflows.
Practical applications
ISO/TR 16705:2016 is intended for practitioners who need rigorous, practical guidance on categorical data analysis in quality and improvement contexts:
- Six Sigma projects (Analyse phase of DMAIC) to test relationships between factors such as defect types, shift, supplier, or inspection outcomes.
- Quality engineers & process analysts assessing attribute data and agreement between systems.
- Researchers in social sciences, medical studies, and surveys evaluating homogeneity, independence or trends in categorical responses.
- Data analysts preparing categorical contingency tables and selecting appropriate tests or multivariate visualisation like correspondence analysis.
Keywords: ISO TR 16705, contingency table analysis, Chi-square test, two-way contingency tables, categorical data analysis, Six Sigma, correspondence analysis, linear trend test.
Related standards
- ISO 3534-1 and ISO 3534-2 - referenced for statistical terminology and measurement scales.
- ISO 14468 - referenced for attribute agreement analysis methods.