IASSC Black Belt Analyze Phase Practice Questions: 9 Worked Questions with Explanations

Analyze is the Black Belt exam’s center of gravity at 27% under the official syllabus: patterns of variation, inferential statistics, and the full hypothesis-testing family — normal-data tests (t-tests, ANOVA, variance tests, sample-size calculation) and non-normal methods (Mann-Whitney, Kruskal-Wallis, Friedman, sign and Wilcoxon tests, proportion tests, chi-square). Our bank holds 405 Analyze questions; these nine worked questions are drawn from that bank, including computed items with the full calculation shown, with the correct answer and a full explanation for each.

Worked questions

1. A safety valve must not open late; opening early wastes product but is tolerable, while opening late risks equipment damage. Engineers test a redesign against the current valve. Why is a one-sided formulation defensible here?

A. One-sided tests are defensible whenever engineers feel strongly about an outcome, regardless of when the direction was selected

B. The formulation is defensible because one-sided tests need smaller samples, and valve timing studies are expensive to run

C. Only lateness carries consequence, so the direction of concern was fixed by the hazard analysis before timing data existed

D. Early opening is physically impossible to observe, so the unused tail of the distribution can be discarded without consequence

Correct: C. Legitimate one-sided testing rests on a decision that cares about one direction only, declared in advance — here the hazard analysis singles out late opening before data collection begins.

2. A one-sample t analysis studies fill volume against a 500 mL target: n = 25, sample mean 503 mL, sample standard deviation 10 mL, giving t = 1.50 with 24 degrees of freedom and a two-sided p-value of 0.147. What conclusion matches the output?

A. The process mean is proven to equal 500 mL exactly, because a p-value above 0.05 certifies the target value as correct

B. The target is rejected, because a sample mean of 503 mL differs from 500 mL by more than the rounding tolerance allowed

C. The data do not establish a departure from 500 mL at the 0.05 level; the mean sits just 1.5 standard errors high

D. The analysis is invalid because degrees of freedom must equal the sample size of 25 rather than one less than that count

Correct: C. Standard error = 10 / sqrt(25) = 2, so t = (503 – 500) / 2 = 1.50 and p = 0.147 exceeds 0.05; the evidence is insufficient, which is not proof the mean equals 500.

3. A Levene test across two production lines returns p = 0.92 for a study whose group spreads in the sample are s = 3.8 and s = 6.9. A report writer wants to state 'the test confirms the spreads are equal.' What is wrong with leaning on that line?

A. Nothing is wrong, because Levene p-values above 0.90 are defined as confirmations of equality in the procedure's original formulation

B. The error is citing Levene at all when the spreads differ, since the procedure refuses to compute when sample variances are unequal

C. The writer should cite the means instead, because a Levene value of 0.92 primarily certifies that the two line averages coincide

D. The spreads differ substantially, and a low-powered variance pre-test cannot erase a near twofold gap with one bland result

Correct: D. Pre-tests for variance are low-powered; p = 0.92 alongside s values of 3.8 versus 6.9 is better read as an insensitive check than as confirmation, and Welch remains the prudent primary analysis.

4. An ANOVA table for a viscosity study shows a within-groups sum of squares of 450 on 45 degrees of freedom and a between-groups mean square of 70. What F value follows?

A. F is 0.14, from dividing the within mean square of 10 by the between mean square of 70 in this viscosity comparison

B. F is 7.0, from a within mean square of 10 under a between mean square of 70 for the viscosity groups being compared

C. F is 45, from dividing the within sum of squares by its degrees of freedom and reporting that quotient as the F value

D. F is 6.4, from dividing 450 by 70 after treating the within sum of squares as though it were already a mean square

Correct: B. MSW = 450 / 45 = 10, so F = MSB / MSW = 70 / 10 = 7.0. The within sum of squares must be averaged by its degrees of freedom before it can serve as the denominator.

5. All nine welds tested after a settings change broke at higher loads than their matched controls. What is the two-sided sign-test p-value?

A. It is 0.018, because each of the nine welds contributes 0.002 to the total under the exact binomial calculation used here

B. It is 0.001, because nine straight successes is a one-in-a-thousand event under any reasonable no-effect model in testing

C. It is 0.50, because the sign test splits its probability evenly between the two directions regardless of how results fall out

D. It is about 0.004, because one direction gives (0.5)^9 and the two-sided value doubles that tail to roughly four in a thousand

Correct: D. The chance of nine positives in nine fair trials is (0.5)^9 ≈ 0.00195. Doubling for the two-sided test gives ≈ 0.0039, about 0.004 — strong evidence from direction alone, with magnitudes never consulted.

6. A survey plan promises a margin of error of plus or minus 4 points for a proportion, and a sponsor demands plus or minus 2 points instead at the same confidence. What happens to the required sample?

A. It roughly doubles, because halving the margin is bought by doubling the interviews in a simple proportional exchange here

B. It roughly quadruples, because margin shrinks with the square root of the sample size, so halving it costs four times the data

C. It stays the same, because margin of error is set by the confidence level chosen and not by how many interviews are gathered

D. It falls by half, because a tighter margin means each interview is used more efficiently in the sponsor's preferred design

Correct: B. Margin ≈ z√(p(1−p)/n); the 1/√n dependence means half the margin requires (2)² = 4 times the sample. This square-root law governs proportions just as it governs means.

7. A multi-vari chart shows large swings from hour to hour but tiny differences between cavities and consecutive pieces. Which family dominates?

A. Cyclical variation dominates, because hour-to-hour movement is defined as the cycle family in multi-vari analysis

B. Positional variation dominates, because the chart's largest swings always belong to the positional family by convention

C. Temporal variation dominates, because change unfolding over longer time spans dwarfs the within-moment differences

D. No family dominates, because a multi-vari chart cannot distinguish time-based movement from position-based movement

Correct: C. Hour-to-hour swings are time-scale effects – the temporal family – while cavity (positional) and consecutive-piece (cyclical) differences are small by comparison.

8. Inference language check: a report says the population statistic is 42 based on a sample parameter of 40. What error is embedded?

A. The arithmetic is wrong, because statistics must always exceed parameters by at least the margin of error allowed

B. No error exists, because modern usage treats parameter and statistic as interchangeable labels for numerical summaries

C. The terms are reversed: populations have parameters and samples have statistics, so the sentence mislabels both values

D. The error is reporting two different numbers, because a valid statistic must exactly equal its population parameter

Correct: C. A statistic summarizes a sample; a parameter describes a population. Swapping the words signals conceptual confusion even if the numbers themselves are fine.

9. A filling operation reports a 95% confidence interval for average fill weight of 498 g to 502 g, built from 60 bottles. The line supervisor concludes that roughly 95% of individual bottles weigh between 498 g and 502 g. How should a Black Belt correct that reading?

A. The interval addresses uncertainty about the long-run process average, not the spread of individual bottle weights

B. The reading holds whenever the sample exceeds 30 units, because large samples make averages and individuals interchangeable

C. The interval actually covers the middle 95% of bottle weights, provided the 60 bottles were drawn across several shifts

D. The supervisor understated the claim, since the interval guarantees every future bottle will land between those two bounds

Correct: A. A confidence interval quantifies uncertainty about a population mean, not the spread of individual units; individual bottles vary far more widely than the average does.

A note on format. The real IASSC exam mixes multiple-choice with true/false questions, while PassProfessor’s bank trains you with four-option multiple-choice questions throughout — the same knowledge, in a format that forces you to discriminate between close alternatives. No worked set on this page reproduces the real exam’s item-type mix.

These questions are drawn from the same bank as the full program. Start with the 20 free sample questions (also presented readably on the free sample questions page), read the IASSC Black Belt exam guide, then see the full IASSC Lean Six Sigma Black Belt Exam Prep program — 1,500 original questions built at the official syllabus weights (Define 13% · Measure 21% · Analyze 27% · Improve 17% · Control 22%), with an explanation for every answer, for $59 with 6 months of access.

Disclaimer

PassProfessor is an independent exam-preparation provider. This program is exam preparation only: it is not the real exam, it provides no education credit, and completing it does not by itself qualify you for any license or certification. The real exam is run by its own exam body — you register for it, pay its fees, and sit it through that body, not through PassProfessor. All practice questions in this program are original; they are not actual exam questions.