🎓 CSSBB Quest › Section VI

Section VI — Analyze Phase ~15%

Analyze carries 22 of the 150 scored questions. It tests the statistical toolkit the BB uses to move from symptom to root cause. Questions blend concept (what does a p-value mean?) with application (which test fits this scenario?). The FMEA and root-cause tools also appear here despite being qualitative.

Analyze subtopics

Subtopic	Key exam content
A. Modeling relationships between variables	Correlation coefficient; correlation vs. causation; multi-vari charts (positional, cyclical, temporal variation)
B. Hypothesis testing	H₀/H_a structure; Type I (α) and Type II (β) errors; power; p-value interpretation; one- vs. two-tail; selecting the right test
B. Tests for means and variances	1-sample t, 2-sample t, paired t, one-way ANOVA; F-test and Levene’s test for variances; sample size; point and interval estimates
B. Proportions, counts and nonparametric tests	1-proportion z, 2-proportion z, chi-square goodness-of-fit, contingency tables (chi-square test of independence), nonparametric tests (e.g., Mann-Whitney, Kruskal-Wallis)
A. Regression	Simple linear regression; R² and adjusted R²; residual analysis; multiple regression (interpretation)
A. Multivariate tools	Factor analysis, discriminant analysis, multivariate analysis of variance (MANOVA)
C. Failure mode and effects analysis (FMEA)	Severity, occurrence and detection ratings; RPN = Severity × Occurrence × Detection; recalculating after actions
D. Additional analysis methods	Gap analysis; root cause analysis (fishbone/Ishikawa diagram, 5 Whys, fault tree analysis); waste analysis

Selecting the right hypothesis test

What are you comparing?	Data type	Test
One group mean vs. a target	Continuous	1-sample t-test
Two independent group means	Continuous	2-sample t-test
Before-after on the same subjects	Continuous	Paired t-test
Three or more group means	Continuous	One-way ANOVA
Two group variances	Continuous	F-test (equal variance)
One proportion vs. a target	Attribute	1-proportion z
Two proportions	Attribute	2-proportion z
Relationship between two categorical variables	Attribute	Chi-square test

FMEA RPN calculation

RPN = Severity (1–10) × Occurrence (1–10) × Detection (1–10)

Detection scale is INVERTED:
  1 = almost certain to detect before reaching customer
 10 = almost impossible to detect

High RPN = highest priority for corrective action.
After action: recalculate RPN to verify improvement.

Exam trap: A low Detection score (e.g., 1) is good — it means the defect is almost certain to be detected. Many candidates reverse this and think low Detection = bad. Also: RPN alone is not the only priority; a severity-9 or 10 failure mode still warrants attention even if its RPN is low, because the consequence is severe.

Design of experiments (DOE) is in Section VII Improve in the 2022 BoK, not Analyze.