Practice questions · Project Management
Lean Six Sigma Black Belt (IASSC ICBB): Practice Questions
Original concept-check questions for the IASSC Lean Six Sigma Black Belt (ICBB), following the DMAIC phases and going deeper than Green Belt into inferential statistics, multiple regression and design of experiments (DOE). Every answer is explained, including why each wrong option is wrong. Filter by phase or difficulty. These are original study questions, not real exam questions.
Answered 0 · Correct 0
-
At Black Belt level, the Define phase differs most from Green Belt in that the Black Belt is expected to:
Correct answer: D. The Black Belt's Define work centres on selecting, scoping and leading larger projects and coaching others, which is the main leadership step up from Green Belt. Avoiding the charter and skipping the voice of the customer abandon core Define activities that remain essential. Running hypothesis tests belongs to Analyze, not Define. -
A well-written project problem statement should:
Correct answer: A. A problem statement describes the gap and its business impact while staying solution-neutral, so the team can investigate causes objectively. Naming the solution up front biases the project before Analyze. Control limits come from the Control phase, and a regression equation is an Analyze/Improve output - neither belongs in the problem statement. -
Translating a broad customer need into a specific, measurable requirement is the role of a:
Correct answer: C. A CTQ converts a general customer need into a specific, measurable requirement, which is exactly this task. A control chart monitors a process over time, a Type II error is a hypothesis-testing mistake, and a design of experiments run tests input factors - none of them translate customer needs. -
A Black Belt is choosing between several candidate projects. The strongest basis for selection is:
Correct answer: B. Projects should be chosen for alignment with business priorities and a measurable, data-supported gap, so the effort delivers real value. Picking the easiest project or the one that avoids statistics ignores impact, and the largest budget is a cost, not evidence of value. -
Compared with a Green Belt, a Black Belt project charter typically reflects:
Correct answer: A. Black Belt charters tend to cover larger scope, broader stakeholders and a full-time lead, matching the bigger projects Black Belts run. A charter without goals or scope is not a charter at all. A control plan is a later Control output, and measuring the process is essential rather than banned. -
Before trusting process data, a Black Belt validates the measurement system because:
Correct answer: D. Measurement System Analysis matters because measurement error inflates variation and can mask a real effect or invent a false one, corrupting every later analysis. It does not set the budget, does not replace Analyze (which finds root causes), and validating measurement says nothing about whether the process is capable. -
A process has a Cpk of 0.7. The most accurate reading is that the process is:
Correct answer: C. A Cpk well below 1.0 indicates the process is not capable of consistently meeting its specification limits. It is therefore not comfortably capable. Cpk reflects both spread and centring, so a low value does not mean the process is perfectly centred, and no real process is entirely free of variation. -
The standard deviation of a dataset describes:
Correct answer: D. Standard deviation measures the typical spread of values around the mean, a core measure of variation. Defects per million is DPMO, the slope of a regression line is a regression coefficient, and the number of process steps comes from a process map - none of these is the standard deviation. -
Choosing a sample large enough for a study matters mainly because:
Correct answer: C. Adequate sample size protects statistical power, so a real effect is not missed (a Type II error). Larger samples usually cost more, not less, so the claim that they are always cheaper is wrong. Sample size does not remove the need for a test, and it does not guarantee the data are normally distributed. -
Discrete (attribute) data is best described as:
Correct answer: B. Discrete or attribute data consists of counts or categories such as pass/fail or defect counts. Values on a continuous scale describe continuous data instead. Discrete data is not inherently normally distributed, and it is unrelated to a regression residual, which is a modelling leftover. -
In hypothesis testing, the p-value represents:
Correct answer: A. The p-value is the probability of seeing data this extreme assuming the null hypothesis is true; a small p-value casts doubt on the null. It is not the probability that the alternative is true, and it is not the effect size - a tiny effect can still be significant with enough data. It has nothing to do with project cost. -
Using a significance level (alpha) of 0.05, you would reject the null hypothesis when the p-value is:
Correct answer: D. You reject the null hypothesis when the p-value is less than the chosen alpha (here 0.05), signalling a statistically significant result. A p-value greater than 0.05 means you fail to reject, a p-value of exactly 1.0 gives no evidence against the null, and you never reject regardless of the p-value - that would defeat the test. -
A Type I error occurs when you:
Correct answer: A. A Type I error is rejecting a true null hypothesis - a false alarm, concluding there is an effect when there is none. Failing to reject a false null is the opposite mistake, a Type II error. A small sample and the wrong control chart are study-design problems, not the definition of a Type I error. -
You want to compare the mean cycle time across four different machines at once. The most appropriate test is:
Correct answer: C. ANOVA compares the means of three or more groups in one test, which fits four machines exactly and avoids the inflated error of many two-sample tests. A Pareto chart ranks categories by frequency rather than comparing means, a capability study compares a process to its specs, and a SIPOC is a high-level process map - none compares group means. -
Your data fail a normality test and cannot reasonably be assumed normal. The best next step for a comparison is to:
Correct answer: B. For clearly non-normal data, a non-parametric test or a suitable transformation gives valid results, which is core Black Belt practice. Using a standard t-test anyway can violate its assumptions and mislead. Stopping the project is unnecessary, and declaring significance without any test abandons the statistics entirely. -
A correlation coefficient of -0.9 between two variables indicates:
Correct answer: D. A coefficient near -0.9 shows a strong negative linear relationship: as one variable rises, the other tends to fall. A value near zero would mean little or no linear relationship, so 'no relationship' and 'weak positive' are both wrong. Correlation never proves causation, however strong it is. -
In a simple linear regression output, the slope coefficient tells you:
Correct answer: C. The slope coefficient estimates how much the output is expected to change for each one-unit change in the input, the core of a regression model. It is not a defect count, has nothing to do with an exam pass mark, and does not set a control plan's monitoring frequency. -
A high R-squared value in a regression model means:
Correct answer: A. R-squared is the share of variation in the output explained by the model, so a high value means the inputs account for much of that variation. A good fit still does not prove causation. The sample size is a separate question, and R-squared says nothing directly about whether it was too small. Residuals must always be checked, not ignored. -
The Improve phase at Black Belt level adds, beyond Green Belt tools, the deliberate use of:
Correct answer: B. Black Belt Improve work uses design of experiments and multiple regression to study and optimise several factors at once, which is the main step up from Green Belt. Relying only on brainstorming ignores that depth, skipping the pilot raises rollout risk, and removing measurement would make it impossible to prove the improvement. -
The core advantage of design of experiments (DOE) over changing one factor at a time is that it:
Correct answer: D. DOE varies several factors together in a planned way, so you learn their individual effects and interactions far more efficiently than one-factor-at-a-time testing. It still requires data and a clear question, so 'avoids data' and 'removes the need for a hypothesis' are wrong, and no method can guarantee zero defects. -
In a designed experiment, an interaction effect means that:
Correct answer: A. An interaction means the effect of one factor changes depending on the level of another, which is exactly what DOE is designed to reveal. It does not mean the factors never matter, nor that the experiment failed - finding interactions is a useful result. The presence of an interaction has nothing to do with whether the data are discrete. -
FMEA (Failure Mode and Effects Analysis) prioritises potential failures using the:
Correct answer: B. FMEA ranks failures by a Risk Priority Number that multiplies severity, occurrence and detection ratings, so the worst risks are tackled first. The customer's address and a budget code are administrative details, and the raw number of process steps does not measure risk - it is the severity/occurrence/detection logic that does. -
Piloting a solution before full rollout is still important at Black Belt level because it:
Correct answer: C. A pilot tests the change on a small scale first, limiting the cost and disruption if it does not work as expected. It does not replace the control plan, which sustains the gain afterwards. A pilot alone does not prove causation, and no change eliminates all variation. -
Poka-yoke (mistake-proofing) improves a process by:
Correct answer: B. Poka-yoke designs the step so the error cannot easily happen in the first place, preventing defects at the source. Adding more end inspection only catches errors later, larger batches do not prevent mistakes, and removing the control plan would weaken the Control phase rather than improve the process. -
When choosing among solutions that all address the verified root cause, the Black Belt should weigh:
Correct answer: D. A sound choice balances effectiveness, cost, risk and how easily the gain can be sustained, because the best solution must also last. Looking only at cost ignores effectiveness and risk, changing the most at once raises risk without adding value, and avoiding a pilot removes a key way to de-risk the rollout. -
On a control chart, a point falling beyond the control limits signals:
Correct answer: A. A point beyond the control limits flags possible special-cause variation - something unusual that warrants investigation. It is not a measure of customer satisfaction, and it is the opposite of common-cause-only behaviour. It certainly does not mean the process is perfect; it means something may be wrong. -
The main purpose of a control plan in the Control phase is to:
Correct answer: C. A control plan documents what to monitor, the limits, and how to respond, so the improved process keeps performing over time. It does not re-open Define, does not replace the regression model from Analyze, and is far more than a budget record - its job is to sustain the gain. -
Distinguishing common-cause from special-cause variation matters in Control because:
Correct answer: A. The distinction drives the response: common-cause variation calls for changing the system, while special-cause variation calls for finding and addressing a specific event. They are not identical, special-cause variation clearly exists, and recognising the difference is a reason to keep monitoring, not to stop. -
Standardising the improved process (standard work) in Control helps mainly because it:
Correct answer: B. Standard work captures the improved method as the documented, repeatable norm, which is how a gain is sustained after the team moves on. It is meant to make the change visible, not hide it; it does not restart DMAIC; and it complements statistical process control rather than removing the need for it. -
A Black Belt formally hands the improved process back to the process owner at the end of Control mainly to:
Correct answer: D. Handover transfers day-to-day ownership and the control plan to the people who run the process, so the gain is sustained after the project team leaves. Avoiding documentation would undermine that handover, re-opening Measure would restart finished work, and cancelling the SPC charts would remove the very monitoring the handover relies on. -
A financial benefits estimate signed off with the finance function strengthens a Black Belt project mainly because it:
Correct answer: C. A finance-validated benefits estimate gives the project credible value that leadership trusts, which matters for the larger projects Black Belts lead. It does not replace measurement, guarantee success, or set control limits. -
A project scope that is too broad for a single Black Belt project is best handled by:
Correct answer: B. An over-broad scope should be narrowed or split into linked, bounded projects so each is achievable and measurable. Ignoring scope, removing the business case, or skipping Measure would all undermine the work. -
When a Black Belt coaches Green Belts on their projects, the Black Belt's role is best described as:
Correct answer: A. Black Belts mentor Green Belts, guiding method and statistical choices while the Green Belts lead their own projects. They do not do all the analysis, replace the sponsor, or act merely as paperwork auditors. -
A Pareto analysis at the Define or Measure stage helps a Black Belt to:
Correct answer: D. Pareto analysis ranks categories so the project targets the vital few driving most of the problem. It is not a designed experiment, a regression slope, or a way to set alpha. -
The Central Limit Theorem is important in Six Sigma because it states that:
Correct answer: C. The Central Limit Theorem says the sampling distribution of the mean approaches normal as sample size increases, regardless of the population shape, which underpins many tests. It does not claim all data are normal, comment on cost, or relate to causation. -
A process is described as 'stable but not capable'. This most likely means it is:
Correct answer: B. Stable but not capable means the process is in statistical control (predictable) yet its spread or centring still cannot meet the specification. It is not out of control, free of variation, or necessarily centred. -
For a Gage R&R study, a total measurement variation consuming a very high percentage of the total variation indicates the measurement system is:
Correct answer: A. If measurement variation consumes a large share of total variation, the system is inadequate and can hide real process signals. That is the opposite of excellent or perfectly accurate, and it is highly relevant to data quality. -
The difference between accuracy and precision in a measurement system is that accuracy concerns:
Correct answer: C. Accuracy is how close measurements are to the true value (bias), while precision is the consistency of repeated readings (spread). Neither relates to process steps, budget, or control-plan frequency. -
A Cp of 2.0 but a Cpk of 1.0 tells a Black Belt that the process:
Correct answer: D. When Cp is high but Cpk is notably lower, the spread is tight enough but the process is off-centre within the specification. It is therefore not perfectly centred, and the gap shows centring can be improved; all real processes have variation. -
Converting a continuous defect rate to a sigma level allows a Black Belt to:
Correct answer: B. A sigma level puts different processes on a common scale so their performance can be compared. It does not avoid data collection, prove causation, or set prices. -
A confidence interval for a process mean expresses:
Correct answer: A. A confidence interval gives a range likely to contain the true population mean at a stated confidence level, reflecting sampling uncertainty. It is not the value of each data point, a schedule, or a defect count. -
Increasing the sample size in a hypothesis test, all else equal, generally:
Correct answer: D. A larger sample increases power, the probability of detecting a real effect (reducing Type II error risk). It does not decrease power, leave it unchanged, or guarantee a false positive. -
The power of a test is defined as:
Correct answer: C. Power is the probability of correctly rejecting a false null hypothesis, equal to one minus beta (the Type II error rate). It is not the Type I error rate, the p-value, or a specification width. -
In ANOVA, a statistically significant result indicates that:
Correct answer: D. A significant ANOVA result indicates at least one group mean differs from the rest, though it does not say which. It is the opposite of all means being equal, does not require discrete data, and variance is central to the test. -
To compare the means of two paired measurements (before and after on the same units), the appropriate test is the:
Correct answer: B. A paired t-test compares two related measurements on the same units, accounting for the pairing. An independent-sample t-test assumes unrelated groups, chi-square tests categorical association, and one-way ANOVA compares three or more independent groups. -
When comparing the variances (spread) of two groups rather than their means, a Black Belt would use a test such as:
Correct answer: C. Tests such as the F-test or Levene's test compare variances between groups. A Pareto chart, a SIPOC, and a capability index do not test for differences in variance. -
A 95% confidence level corresponds to an alpha (significance level) of:
Correct answer: A. A 95% confidence level corresponds to alpha = 0.05 (the 5% chance of a Type I error). 0.95 is the confidence level itself, while 0.50 and 0.01 correspond to other confidence levels. -
In multiple regression, multicollinearity refers to:
Correct answer: B. Multicollinearity is when predictors are highly correlated with one another, which inflates uncertainty and can distort the individual coefficient estimates. It is not about the response's normality, the sample size, or a perfect fit. -
Checking the residuals of a regression model matters because:
Correct answer: D. Residual analysis checks whether assumptions such as constant variance, independence and normality hold, validating the model. Residuals do not set control limits, prove causation, or equal R-squared. -
A statistically significant result with a trivially small effect size should be interpreted as:
Correct answer: A. Statistical significance with a tiny effect size means the effect is real but may not matter in practice, especially with large samples. It does not imply a large practical effect, prove causation, or mean the test failed. -
A normal probability plot is used mainly to:
Correct answer: C. A normal probability plot assesses whether data follow a normal distribution, with points near a straight line suggesting normality. It does not rank causes, map flow, or compute an FMEA risk priority number. -
Reducing alpha from 0.05 to 0.01, holding sample size constant, will:
Correct answer: B. A stricter alpha reduces Type I error risk but, with the same sample size, increases the chance of a Type II error (missing a real effect). It cannot lower both at once for a fixed sample, leave both unchanged, or guarantee significance. -
A correlation coefficient of 0.0 between two variables means:
Correct answer: D. An r of 0 indicates no linear relationship, but the variables could still be related non-linearly. It does not mean a strong relationship, causation, or invalid data. -
A full factorial design with three factors each at two levels requires how many runs (excluding replicates)?
Correct answer: A. A two-level full factorial with three factors needs 2 to the power of 3, which is 8 runs, to cover every combination. 6, 3 and 16 do not match 2^3. -
The main reason to use a fractional factorial design instead of a full factorial is to:
Correct answer: C. A fractional factorial studies many factors with fewer runs, accepting some confounding (aliasing) of effects as the trade-off. It cannot guarantee zero defects, avoid analysis, or remove real interactions. -
In a designed experiment, 'confounding' (aliasing) means that:
Correct answer: B. Confounding (aliasing) occurs in fractional designs when the effects of two or more terms cannot be separated. It is not the absence of factors, a statement about normality, or a guarantee of independence. -
A main effect in a designed experiment is:
Correct answer: A. A main effect is the average change in the response as a single factor moves from low to high. An effect that depends on another factor's level is an interaction, not a main effect; runs and residuals are different concepts. -
Randomising the run order in a designed experiment is done to:
Correct answer: C. Randomising run order spreads the effect of unknown time-related factors (such as tool wear or warm-up) so they do not bias the results. It is not about speed, replacing replication, or forcing significance. -
Replication in a designed experiment (running each combination more than once) primarily allows you to:
Correct answer: D. Replication provides an estimate of pure experimental error and sharpens the precision of effect estimates. It does not reduce factors, replace randomisation, or eliminate interactions. -
A center point added to a two-level factorial design helps to:
Correct answer: C. Center points let you check for curvature, signalling whether a linear model is adequate or a higher-order design is needed. They are not used to add confounding, remove factors, or set control limits. -
Response surface methodology (RSM) extends DOE mainly to:
Correct answer: A. RSM models curvature in the response and helps locate optimal factor settings, going beyond simple two-level screening. It does not rank causes, validate measurement, or draw a SIPOC. -
The key advantage of DOE over changing one factor at a time (OFAT) is that DOE:
Correct answer: D. DOE varies factors together, revealing interactions OFAT misses and using runs efficiently. It is generally more efficient than OFAT, can use replication, and certainly detects main effects. -
A screening design (such as a resolution III or a Plackett-Burman design) is typically used early to:
Correct answer: B. Screening designs efficiently sift many candidate factors down to the vital few with few runs, before deeper optimisation. They are not for final optimisation, replacing the control plan, or proving causation without data. -
When a DOE reveals a strong two-factor interaction, the correct interpretation is that:
Correct answer: D. A strong interaction means the optimal level of one factor depends on the other's level, so they must be set together. It does not mean independent optimisation, an invalid experiment, or no effect. -
Before running a designed experiment, defining the factors, levels and the response in advance is important because it:
Correct answer: B. Planning factors, levels and the response up front ensures the design answers the question and yields analysable data. It does not guarantee significance, replace randomisation, or remove measurement error. -
An X-bar and R chart is most appropriate for monitoring:
Correct answer: A. X-bar and R charts track the subgroup mean and range of continuous data over time. Attribute counts use p, np, c or u charts; Pareto categories and DOE factors are unrelated chart purposes. -
To monitor the proportion of defective units across varying subgroup sizes, the right control chart is the:
Correct answer: C. A p chart monitors the proportion defective and handles varying subgroup sizes. X-bar and R and I-MR charts are for continuous data, and a scatter plot shows relationships, not control over time. -
An individuals and moving range (I-MR) chart is used when:
Correct answer: D. I-MR charts suit continuous data collected as individual observations, where rational subgroups are not available. Large subgroups suit X-bar and R, counts suit attribute charts, and a DOE is a different activity. -
Western Electric (run) rules on a control chart are used to:
Correct answer: B. Run rules flag non-random patterns (such as trends or too many points on one side) that signal special causes even when no point is outside the limits. They do not set specifications, calculate Cpk, or select projects. -
Tampering with a stable process by adjusting it in response to common-cause variation typically:
Correct answer: C. Reacting to common-cause noise as if it were a signal (overadjustment) usually adds variation, a point often illustrated by the funnel experiment. It does not reduce variation, leave it unchanged, or remove special causes. -
The purpose of a control plan reaction (response) plan is to specify:
Correct answer: A. A reaction plan defines what to do when a monitored characteristic goes out of control, so issues are corrected quickly and consistently. It is not a marketing strategy, a DOE run order, or a sponsor choice. -
A capability study run during Control, after improvements, is used to:
Correct answer: C. A post-improvement capability study confirms the process now meets specification consistently, verifying the gain. Identifying root causes is Analyze, selecting the project is Define, and a capability study complements rather than replaces a control chart. -
Statistical process control limits should be recalculated when:
Correct answer: B. Control limits are recalculated when the process has genuinely and verifiably changed, such as after a validated improvement, so the chart reflects current behaviour. Recalculating every shift, on a stakeholder's whim, or refusing ever to update would all undermine monitoring. -
Sustaining the financial benefits of a Black Belt project over time depends most on:
Correct answer: D. Lasting benefits depend on effective control plans and clear ownership so the improvement holds after the team leaves. A one-off celebration, removing monitoring, or needlessly re-running Analyze do not sustain the gain. -
A key reason Black Belt projects emphasise verifying root causes with data before improving is that:
Correct answer: A. Verifying the root cause with data avoids investing in a 'fix' that does not address the real driver. Skipping verification is risky, data alone does not prove causation, and a pilot is still valuable afterwards. -
A hypothesis test comparing a process before and after a change is an example of using statistics to:
Correct answer: C. Such a test judges whether the before-after difference is statistically real rather than chance variation, validating the improvement. It does not define customers, draw a SIPOC, or set budgets. -
Practical significance differs from statistical significance in that practical significance asks:
Correct answer: D. Practical significance asks whether the effect is big enough to matter in practice, whereas statistical significance only asks whether it is unlikely to be chance. Sample randomness and chart control are separate questions. -
When non-normal continuous data must be compared and transformation is undesirable, a Black Belt can use a:
Correct answer: B. Non-parametric tests such as Mann-Whitney (two groups) or Kruskal-Wallis (several groups) compare non-normal data without assuming normality. Forcing a standard t-test can violate assumptions, and a Pareto chart or capability index does not compare groups this way. -
A control chart that signals out-of-control during the Control phase should prompt the team to:
Correct answer: A. An out-of-control signal should trigger investigation of the special cause and the documented reaction plan. Widening limits to silence the signal, ignoring it, or restarting Define would all defeat the purpose of control. -
Hypothesis testing controls the risk of a false positive primarily through the choice of:
Correct answer: C. Alpha sets the acceptable probability of a Type I error (false positive). The regression slope, the count of process steps, and the control-plan owner do not control this risk. -
In a two-level factorial experiment, coding factor levels as -1 and +1 is done to:
Correct answer: D. Coding the low and high levels as -1 and +1 standardises the scale, simplifying analysis and making effects comparable. It does not increase runs, force confounding, or remove the response. -
The hierarchy of effects in most designed experiments assumes that:
Correct answer: B. Effect sparsity and hierarchy assume main effects are usually largest, two-factor interactions next, and high-order interactions small, which justifies fractional designs. It is not the case that high-order interactions dominate, that interactions never occur, or that main effects can be ignored. -
A Black Belt should treat a measurement system's bias by:
Correct answer: A. Bias (a consistent offset from the true value) is addressed by calibrating against a known reference standard. Ignoring it, widening the specification, or changing batch size does not correct measurement bias.
Practice questions FAQ
- Are these real ICBB exam questions?
- No. These are original study questions written to test understanding. They are not real exam questions, exam dumps, or copied from any provider.
- How should I use these practice questions?
- Answer each one, read the explanation (including why the wrong options are wrong), and use the per-domain score below to focus your revision on weak areas. Revisit before exam day.
- How many questions should I do before the exam?
- Enough to score consistently across every domain, alongside full-length practice from official or reputable providers. Understanding why each answer is right matters more than raw volume.
- What score means I am ready?
- A good signal is consistently scoring around 80% or higher across all domains on questions you have not seen before, and being able to explain why the wrong options are wrong.
- Should I use exam dumps?
- No. Dumps (real or leaked questions) breach provider policy, can void your certification, and do not build the understanding the exam actually tests.