Section IV: Analyze Phase 18 questions
Analyze carries 18 of the 100 scored questions on the 2022 CSSGB exam (up from 15 in the 2014 BoK). It is where a Green Belt uses data to confirm root causes rather than just brainstorm them: exploratory data analysis, hypothesis testing, and root cause and gap analysis.
The emphasis is on choosing the right tool or test and interpreting the result correctly.
Key topics
| Topic | What to know |
|---|---|
| Exploratory data analysis | Multi-vari studies (positional, cyclical and temporal variation); correlation coefficient r; simple linear regression and r². |
| Hypothesis testing basics | H₀/Hₐ, α and β risk, power, p-values, statistical vs. practical significance. |
| Tests for means, variances and proportions | z- and t-tests (one-sample, two-sample, paired), F-test for two variances, tests for proportions. |
| ANOVA and chi-square | One-way (single-factor) ANOVA for three or more means; chi-square tests for goodness of fit and contingency tables. |
| Additional analysis methods | Gap analysis (current vs. desired state) and root cause analysis: cause-and-effect diagrams, relational matrices, 5 Whys, fault tree analysis. |
Test selection
| Situation | Test |
|---|---|
| One mean vs. a target (continuous) | One-sample t-test (z-test if σ is known) |
| Two independent group means | Two-sample t-test |
| Same units measured twice (before/after) | Paired t-test |
| Three or more group means | One-way ANOVA |
| Two variances | F-test (Levene’s test is a common alternative for non-normal data) |
| Two proportions | Two-proportion z-test (or chi-square) |
| Two categorical variables, or observed vs. expected counts | Chi-square test |
| Relationship between two continuous variables | Correlation / simple linear regression |
p-values and errors
Type I error (α): rejecting a true H₀ (false alarm). Type II error (β): failing to reject a false H₀ (miss). Power = 1 − β.
What p does not mean: p = 0.03 is not "a 3% chance H₀ is true." It is the probability of data at least this extreme if H₀ were true. Failing to reject H₀ does not prove it.
Correlation and regression
r runs from −1 to +1 and measures the strength of a linear relationship. r² is the share of the variation in Y explained by the linear model: r = 0.8 gives r² = 0.64, so 64% of the variation is explained. A strong correlation is not proof of causation.
Root cause tools
| Tool | Best for |
|---|---|
| Cause-and-effect (fishbone) diagram | Organizing possible causes by category (e.g., the 6Ms) as hypotheses to test |
| 5 Whys | Drilling down a single cause chain from symptom to a controllable root cause |
| Fault tree analysis | Top-down logic (AND/OR gates) for an undesired event with combined causes |
| Pareto chart | Finding the vital few categories to focus on |
Common exam traps
- Paired vs. two-sample. If the same units are measured under both conditions, the data are paired.
- Significant ≠ important. With large samples, tiny differences can be statistically significant but practically irrelevant.
- Correlation ≠ causation. A lurking variable can drive both X and Y.
- Brainstorming tools do not prove causes. Fishbones and 5 Whys produce hypotheses; data confirms them.
Learn pages for this section
The free 20-question CSSGB diagnostic covers all six sections and shows your score by section.
Take the free diagnostic →Section names and question counts follow the 2022 CSSGB Body of Knowledge (110 questions, 100 scored). Topic summaries are our paraphrase; check the current BoK on asq.org for the official wording.
CSSGB Quest by Bizarre Act. Original study material; not affiliated with ASQ. CSSGB is a certification of ASQ. Terms · Privacy