Lean Six Sigma Black Belt · DMAIC · Updated 2026
Free Lean Six Sigma Black Belt Practice Exam (2026)
Full-length Black Belt mock exams with the same number of questions and time as common Black Belt certification exams. Define, Measure, Analyze, Improve and Control questions with real calculations, answers explained, and your score by phase.
1. Choose a set
2. Choose a mode
150 questions · 4 hours · same as the real exam
You have an unfinished attempt on this set.
Questions 1–10Page 1 of 15
🏆
Question 1Define
A hospital's strategic goal is higher patient satisfaction. A Black Belt's project targets emergency department wait time. How does wait time relate to the strategic goal?
Correct answer: A. It is a lower-level y that drives the big Y
Project metrics are usually lower-level y's that link up to an organizational big Y. Wait time drives satisfaction, so improving it supports the strategic goal.
Project metrics are usually lower-level y's that link up to an organizational big Y. Wait time drives satisfaction, so improving it supports the strategic goal.
Question 2Measure
A packaging machine was available 90% of planned time, ran at 80% of its ideal speed while running, and produced 95% good units. What is its overall equipment effectiveness?
Correct answer: C. 68.4%
OEE = availability x performance x quality = 0.90 x 0.80 x 0.95 = 0.684, or 68.4%. The value 88.3% is the simple average of the three factors, which overstates losses combined across all three.
OEE = availability x performance x quality = 0.90 x 0.80 x 0.95 = 0.684, or 68.4%. The value 88.3% is the simple average of the three factors, which overstates losses combined across all three.
Question 3Analyze
A team claims a new routing method reduces average delivery time below the current 45 minutes. What is the alternative hypothesis?
Correct answer: D. The mean delivery time is less than 45 minutes
The alternative hypothesis states the claim being tested, which is a decrease, so it is mean less than 45. The equals statement is the null hypothesis.
The alternative hypothesis states the claim being tested, which is a decrease, so it is mean less than 45. The equals statement is the null hypothesis.
Question 4Improve
A regression for delivery cost is Cost = 250 + 18 x Miles. What does the value 18 represent?
Correct answer: D. Each additional mile adds $18 to cost on average
The slope is the average change in the response for each one-unit increase in the predictor. The cost at zero miles is the intercept, $250.
The slope is the average change in the response for each one-unit increase in the predictor. The cost at zero miles is the intercept, $250.
Question 5Control
A hospital tracks serious medication errors, which happen only a few times per year. A standard c chart shows mostly zeros. Which approach is BEST?
Correct answer: B. Chart the time or cases between events
For rare events, charts of time or opportunities between events, such as t or g charts, are more sensitive than counts that are mostly zero. Charting yearly totals gives too few points.
For rare events, charts of time or opportunities between events, such as t or g charts, are more sensitive than counts that are mostly zero. Charting yearly totals gives too few points.
Question 6Analyze
What does a p-value of 0.04 mean?
Correct answer: B. If the null is true, results this extreme occur about 4% of the time
A p-value is the probability of seeing data at least this extreme if the null hypothesis is true. It is not the probability that either hypothesis is true.
A p-value is the probability of seeing data at least this extreme if the null hypothesis is true. It is not the probability that either hypothesis is true.
Question 7Measure
A product needs 180 seconds of total work content, and the takt time is 40 seconds. What is the minimum number of operators needed for a balanced line?
Correct answer: D. 5 operators
Operators = work content / takt = 180 / 40 = 4.5, which rounds up to 5. Rounding down to 4 would leave the line unable to meet takt.
Operators = work content / takt = 180 / 40 = 4.5, which rounds up to 5. Rounding down to 4 would leave the line unable to meet takt.
Question 8Improve
A multiple regression model is Yield = 40 + 0.5 x Temp - 2.0 x Pressure. What does the coefficient -2.0 mean?
Correct answer: C. Each unit rise in pressure lowers yield by 2, holding temp fixed
In multiple regression, each coefficient shows the change in the response per unit of that predictor with the other predictors held constant. It does not measure percent of variation explained.
In multiple regression, each coefficient shows the change in the response per unit of that predictor with the other predictors held constant. It does not measure percent of variation explained.
Question 9Define
A billing process runs at about 6,210 defects per million opportunities over the long term. Using the conventional 1.5 sigma shift, what sigma level is usually reported?
Correct answer: B. 4.0
With the 1.5 sigma shift, about 6,210 long-term DPMO corresponds to a reported 4 sigma level. A 3 sigma level corresponds to about 66,800 DPMO.
With the 1.5 sigma shift, about 6,210 long-term DPMO corresponds to a reported 4 sigma level. A 3 sigma level corresponds to about 66,800 DPMO.
Question 10Control
In an EWMA chart, what is the effect of choosing a small weighting factor such as lambda = 0.1?
Correct answer: C. More weight on past data, which helps catch small shifts
A small lambda averages over more history, which smooths noise and makes the chart sensitive to small sustained shifts. A lambda of 1 would make it the same as a Shewhart chart.
A small lambda averages over more history, which smooths noise and makes the chart sensitive to small sustained shifts. A lambda of 1 would make it the same as a Shewhart chart.
Question 11Measure
In a print shop, a job spends 4 minutes on the press but takes 3 days from order entry to delivery. Which statement is correct?
Correct answer: A. The press cycle time is 4 minutes and the lead time is 3 days
Cycle time is how long a step takes to complete one unit, while lead time is the total elapsed time from order to delivery. Takt is set by customer demand, not by the order duration.
Cycle time is how long a step takes to complete one unit, while lead time is the total elapsed time from order to delivery. Takt is set by customer demand, not by the order duration.
Question 12Analyze
A test comparing two machines returns p = 0.30. A manager says this proves the machines perform the same. What is the BEST response?
Correct answer: B. We failed to reject the null, which does not prove equality
A high p-value means there is not enough evidence of a difference, not proof that none exists. The test may lack power to detect a real difference.
A high p-value means there is not enough evidence of a difference, not proof that none exists. The test may lack power to detect a real difference.
Question 13Define
A three-step process has a rolled throughput yield of 72.9%. What is the normalized yield per step?
Correct answer: A. 90.0%
Normalized yield is the nth root of RTY: the cube root of 0.729 is 0.90, or 90%. Dividing 72.9% by 3 gives 24.3%, which is not a valid way to spread yield across steps.
Normalized yield is the nth root of RTY: the cube root of 0.729 is 0.90, or 90%. Dividing 72.9% by 3 gives 24.3%, which is not a valid way to spread yield across steps.
Question 14Improve
A regression model has an error sum of squares of 80 and a total sum of squares of 400. What is R-squared?
Correct answer: C. 80%
R-squared = 1 - SSE / SST = 1 - 80 / 400 = 1 - 0.20 = 0.80, or 80%. The value 20% is the unexplained share.
R-squared = 1 - SSE / SST = 1 - 80 / 400 = 1 - 0.20 = 0.80, or 80%. The value 20% is the unexplained share.
Question 15Control
What is the basic principle behind a CUSUM chart?
Correct answer: D. It accumulates deviations from a target over time
A CUSUM chart sums deviations from target, so even a small persistent shift builds up and becomes visible. It does not plot ranges or compare points with specs.
A CUSUM chart sums deviations from target, so even a small persistent shift builds up and becomes visible. It does not plot ranges or compare points with specs.
Question 16Measure
A value stream has three inventory points holding 2 days, 1.5 days, and 4 days of stock, and the total processing time is 20 minutes. About what is the production lead time?
Correct answer: D. About 7.5 days
Lead time on a value stream map is the sum of the inventory days plus processing time: 2 + 1.5 + 4 = 7.5 days plus 20 minutes. The 20 minutes is the value-added time, a tiny share of the total.
Lead time on a value stream map is the sum of the inventory days plus processing time: 2 + 1.5 + 4 = 7.5 days plus 20 minutes. The 20 minutes is the value-added time, a tiny share of the total.
Question 17Analyze
A Black Belt wants to know if the average fill of 25 sampled bottles differs from the 500 mL target. The population standard deviation is unknown. Which test is BEST?
Correct answer: A. 1-sample t-test
A 1-sample t-test compares a sample mean with a target when sigma is estimated from the sample. A Z-test requires a known population standard deviation.
A 1-sample t-test compares a sample mean with a target when sigma is estimated from the sample. A Z-test requires a known population standard deviation.
Question 18Define
An electronics line averages 0.10 defects per unit, and defects occur randomly. Using the Poisson approximation, what first-time yield should the team expect?
Correct answer: B. About 90.5%
With random defects, yield = e to the power of minus DPU = e^-0.10 = 0.905, or about 90.5%. Simply subtracting DPU from 1 gives 90.0%, which ignores units that carry more than one defect.
With random defects, yield = e to the power of minus DPU = e^-0.10 = 0.905, or about 90.5%. Simply subtracting DPU from 1 gives 90.0%, which ignores units that carry more than one defect.
Question 19Improve
A plot of residuals versus observation order shows a clear upward drift over time. What does this suggest?
Correct answer: D. A time-related variable is missing or runs are not independent
A pattern over time means the residuals are not independent, often because of a lurking time-related variable such as tool wear. Randomized run order helps prevent this.
A pattern over time means the residuals are not independent, often because of a lurking time-related variable such as tool wear. Randomized run order helps prevent this.
Question 20Control
When reviewing an Xbar-R chart, why should the R chart be checked before the Xbar chart?
Correct answer: C. Xbar limits depend on R-bar, so a bad R chart makes them invalid
The Xbar chart limits are computed from the average range, so if the R chart is out of control, those limits are not reliable. The R chart tracks spread, not the mean.
The Xbar chart limits are computed from the average range, so if the R chart is out of control, those limits are not reliable. The R chart tracks spread, not the mean.
Question 21Measure
A team considers recording shaft diameter only as pass or fail instead of keeping the actual readings. What is the MAIN drawback?
Correct answer: A. It loses information and needs much larger samples
Converting continuous data to attribute data discards information about how close each part is to the limits. Attribute data need far larger samples to detect the same change.
Converting continuous data to attribute data discards information about how close each part is to the limits. Attribute data need far larger samples to detect the same change.
Question 22Analyze
A sample of 16 parts has a mean of 104 and a standard deviation of 8. The target is 100. What is the 1-sample t statistic?
Correct answer: C. 2.0
t = (sample mean - target) / (s / square root of n) = (104 - 100) / (8 / 4) = 4 / 2 = 2.0. The value 0.5 divides by s instead of the standard error.
t = (sample mean - target) / (s / square root of n) = (104 - 100) / (8 / 4) = 4 / 2 = 2.0. The value 0.5 divides by s instead of the standard error.
Question 23Define
In a smartphone survey, satisfaction rises steadily as battery life increases and falls steadily as it decreases. In the Kano model, battery life is BEST classified as what?
Correct answer: A. Performance (one-dimensional) requirement
Performance requirements have a roughly linear link between how well they are met and satisfaction. A basic need gives little extra satisfaction once met.
Performance requirements have a roughly linear link between how well they are met and satisfaction. A basic need gives little extra satisfaction once met.
Question 24Improve
In a multiple regression, one predictor has a p-value of 0.62 while the others are all below 0.01. What is the BEST next step?
Correct answer: B. Consider removing it and refitting the model
A high p-value means the predictor adds little once the others are in the model, so removing it often gives a simpler, more reliable model. Adding terms always raises R-squared, which is not a reason to keep them.
A high p-value means the predictor adds little once the others are in the model, so removing it often gives a simpler, more reliable model. Adding terms always raises R-squared, which is not a reason to keep them.
Question 25Control
A process has a standard deviation of 6, and the team uses subgroups of 4. How far from the centerline are the 3 sigma limits on the Xbar chart?
Correct answer: A. 9 units
The standard deviation of the subgroup mean is 6 / square root of 4 = 3, so 3 sigma limits are 3 x 3 = 9 units from the centerline. The value 18 uses the individual sigma.
The standard deviation of the subgroup mean is 6 / square root of 4 = 3, so 3 sigma limits are 3 x 3 = 9 units from the centerline. The value 18 uses the individual sigma.
Question 26Measure
An auditor samples every 10th invoice from a system that groups invoices in batches of 10. The last invoice in each batch is always processed by the same clerk. What risk does this create?
Correct answer: C. The sample lines up with a cycle and gives biased results
Systematic sampling can be biased when the sampling interval matches a pattern in the process. Here the sample could pick only one clerk's work.
Systematic sampling can be biased when the sampling interval matches a pattern in the process. Here the sample could pick only one clerk's work.
Question 27Analyze
Error rates are 4% at Site A and 6% at Site B, with 500 transactions sampled at each site. Which test should compare the rates?
Correct answer: B. 2-proportion test
Error rates are proportions from pass or fail data, so a 2-proportion test fits. A t-test is meant for continuous measurements.
Error rates are proportions from pass or fail data, so a 2-proportion test fits. A t-test is meant for continuous measurements.
Question 28Define
A product team uses a House of Quality to translate customer needs into design targets. What does the central relationship matrix show?
Correct answer: D. How strongly each technical requirement affects each need
The central matrix links each customer need to the technical requirements that influence it, with strength ratings. The roof, not the center, shows how technical requirements interact.
The central matrix links each customer need to the technical requirements that influence it, with strength ratings. The roof, not the center, shows how technical requirements interact.
Question 29Improve
A Black Belt wants to model whether a loan will default (yes or no) using income and credit score. Which method is MOST appropriate?
Correct answer: A. Binary logistic regression
Binary logistic regression models the probability of a yes or no outcome from one or more predictors. Linear regression assumes a continuous response.
Binary logistic regression models the probability of a yes or no outcome from one or more predictors. Linear regression assumes a continuous response.
Question 30Control
Fourteen consecutive points on a control chart alternate up and down. What does this pattern MOST likely indicate?
Correct answer: B. Overadjustment or two alternating sources
A sawtooth pattern of alternating points suggests overcontrol by operators or data that alternate between two machines or shifts. A trend would show a steady rise or fall.
A sawtooth pattern of alternating points suggests overcontrol by operators or data that alternate between two machines or shifts. A trend would show a steady rise or fall.
Question 31Measure
Why can a Black Belt often apply normal-based methods to subgroup averages even when the individual values are somewhat skewed?
Correct answer: A. The central limit theorem makes averages close to normal
The central limit theorem states that sample means tend toward a normal distribution as sample size grows, even when individual data are not normal. Removing outliers does not make skewed data normal.
The central limit theorem states that sample means tend toward a normal distribution as sample size grows, even when individual data are not normal. Removing outliers does not make skewed data normal.
Question 32Analyze
A call center wants to test whether calls are evenly spread across the five weekdays. Which test fits?
Correct answer: D. Chi-square goodness of fit
A goodness of fit test compares observed counts in one categorical variable with expected counts, here equal across days. A test of independence needs two categorical variables.
A goodness of fit test compares observed counts in one categorical variable with expected counts, here equal across days. A test of independence needs two categorical variables.
Question 33Define
A Champion asks a Black Belt to strengthen the business case section of a charter. Which content BEST fits that section?
Correct answer: A. Why the project matters now and how it ties to strategy
The business case explains why the project is worth doing now, including its link to goals and the financial impact. Proposing a solution in the charter would bias the analysis.
The business case explains why the project is worth doing now, including its link to goals and the financial impact. Proposing a solution in the charter would bias the analysis.
Question 34Improve
For the same value of x in a regression, why is the prediction interval wider than the confidence interval?
Correct answer: C. It adds the scatter of single values to the uncertainty in the mean
A confidence interval covers the mean response, while a prediction interval must cover an individual new value, which adds the residual scatter. Both use the same confidence level.
A confidence interval covers the mean response, while a prediction interval must cover an individual new value, which adds the residual scatter. Both use the same confidence level.
Question 35Control
An np chart uses a constant subgroup size of 200, and the average proportion defective is 0.05. What is the upper control limit?
Correct answer: A. About 19.2
np-bar = 200 x 0.05 = 10 and sigma = square root of (10 x 0.95) = 3.08, so UCL = 10 + 3 x 3.08 = 19.2. The value 13.1 uses 1 sigma.
np-bar = 200 x 0.05 = 10 and sigma = square root of (10 x 0.95) = 3.08, so UCL = 10 + 3 x 3.08 = 19.2. The value 13.1 uses 1 sigma.
Question 36Measure
A process has a standard deviation of 12. A Black Belt takes samples of 36 units. What is the standard error of the sample mean?
Correct answer: D. 2
Standard error = sigma / square root of n = 12 / 6 = 2. The value 0.33 divides by n instead of its square root.
Standard error = sigma / square root of n = 12 / 6 = 2. The value 0.33 divides by n instead of its square root.
Question 37Analyze
A chi-square test of independence uses a table with 3 rows and 4 columns. How many degrees of freedom does the test have?
Correct answer: B. 6
df = (rows - 1) x (columns - 1) = 2 x 3 = 6. The value 12 multiplies rows by columns without subtracting one from each.
df = (rows - 1) x (columns - 1) = 2 x 3 = 6. The value 12 multiplies rows by columns without subtracting one from each.
Question 38Define
Before finalizing the charter for a packaging line project, a Black Belt spends a full shift on the floor watching operators and asking questions. What is the MAIN benefit of this?
Correct answer: C. Seeing the real process and its problems firsthand
A gemba walk lets the Black Belt see how work actually happens, which improves scope and problem definition. It is not an audit and does not replace structured baseline data collection.
A gemba walk lets the Black Belt see how work actually happens, which improves scope and problem definition. It is not an audit and does not replace structured baseline data collection.
Question 39Improve
A food scientist is testing blends of three ingredients that must always add up to 100% of the recipe. Which type of experimental design is MOST appropriate?
Correct answer: B. Mixture design
In a mixture design the factors are proportions that must sum to a constant, so they cannot be changed independently. A standard factorial assumes each factor can be set on its own.
In a mixture design the factors are proportions that must sum to a constant, so they cannot be changed independently. A standard factorial assumes each factor can be set on its own.
Question 40Control
A u chart has an average of 2.0 defects per unit, and each subgroup contains 8 units. What is the upper control limit?
Correct answer: D. 3.5
Sigma = square root of (u-bar / n) = square root of (2.0 / 8) = 0.5, so UCL = 2.0 + 3 x 0.5 = 3.5. The value 6.2 ignores the subgroup size and uses the square root of 2.
Sigma = square root of (u-bar / n) = square root of (2.0 / 8) = 0.5, so UCL = 2.0 + 3 x 0.5 = 3.5. The value 6.2 ignores the subgroup size and uses the square root of 2.
Question 41Measure
Order cycle times have a mean of 40 minutes and a standard deviation of 5 minutes. What is the z-score for an order that took 32 minutes?
Correct answer: B. -1.6
z = (32 - 40) / 5 = -1.6. The negative sign shows the value is below the mean, and -8.0 forgets to divide by the standard deviation.
z = (32 - 40) / 5 = -1.6. The negative sign shows the value is below the mean, and -8.0 forgets to divide by the standard deviation.
Question 42Analyze
Two machines have sample variances of 36 and 9. What is the F statistic for comparing their variances?
Correct answer: C. 4.0
F = larger variance / smaller variance = 36 / 9 = 4.0. The value 2.0 is the ratio of standard deviations, not variances.
F = larger variance / smaller variance = 36 / 9 = 4.0. The value 2.0 is the ratio of standard deviations, not variances.
Question 43Define
Why is overproduction often called the most serious of the Lean wastes?
Correct answer: A. It hides problems and creates other wastes
Making too much or too early creates extra inventory, handling, storage, and transport, and it hides problems such as defects and imbalance. The other options are not true of overproduction.
Making too much or too early creates extra inventory, handling, storage, and transport, and it hides problems such as defects and imbalance. The other options are not true of overproduction.
Question 44Improve
A 2^(4-1) fractional factorial uses the defining relation I = ABCD. Which effect is aliased with the AB interaction?
Correct answer: D. CD
Multiply AB by the defining word ABCD: AB x ABCD = A squared B squared CD = CD. So AB and CD cannot be separated in this design.
Multiply AB by the defining word ABCD: AB x ABCD = A squared B squared CD = CD. So AB and CD cannot be separated in this design.
Question 45Control
A supervisor draws the product specification limits on an Xbar chart. Why is this a mistake?
Correct answer: A. Specs apply to single parts, but the chart plots averages
Averages vary less than individual parts, so subgroup means can sit inside the specs while individual parts fall outside them. Control limits, not specs, belong on an Xbar chart.
Averages vary less than individual parts, so subgroup means can sit inside the specs while individual parts fall outside them. Control limits, not specs, belong on an Xbar chart.
Question 46Measure
Process A has a mean of 50 and a standard deviation of 5. Process B has a mean of 200 and a standard deviation of 10. Which statement about relative variation is correct?
Correct answer: C. Process A has more relative variation, with a CV of 10%
Coefficient of variation = sigma / mean: A is 5 / 50 = 10% and B is 10 / 200 = 5%. B has the larger absolute sigma but less variation relative to its mean.
Coefficient of variation = sigma / mean: A is 5 / 50 = 10% and B is 10 / 200 = 5%. B has the larger absolute sigma but less variation relative to its mean.
Question 47Analyze
An F-test shows that the variances of two independent groups differ significantly. How should the Black Belt compare their means?
Correct answer: D. Use a 2-sample t-test that does not assume equal variances
When variances are unequal, the unequal variance form of the 2-sample t-test adjusts the standard error and degrees of freedom. The pooled test assumes equal variances.
When variances are unequal, the unequal variance form of the 2-sample t-test adjusts the standard error and degrees of freedom. The pooled test assumes equal variances.
Question 48Define
Experienced lab technicians have ideas for faster sample preparation, but managers never ask for input and all decisions come from the top. Which waste does this describe?
Correct answer: D. Non-utilized talent
Non-utilized talent is the failure to use people's skills, knowledge, and ideas. Waiting refers to idle time between steps, which is not what is described.
Non-utilized talent is the failure to use people's skills, knowledge, and ideas. Waiting refers to idle time between steps, which is not what is described.
Question 49Improve
In a DOE on a curing process, oven temperature takes hours to change, while line speed and pressure can be changed easily. Full randomization of every run is not practical. Which design approach fits BEST?
Correct answer: B. A split-plot design with temperature as the whole-plot factor
A split-plot design groups runs by the hard-to-change factor and randomizes the easy factors within each group, and the analysis accounts for the two error terms. Running in standard order risks confusing time effects with factor effects.
A split-plot design groups runs by the hard-to-change factor and randomizes the easy factors within each group, and the analysis accounts for the two error terms. Running in standard order risks confusing time effects with factor effects.
Question 50Control
A stable process is monitored with 3 sigma limits and normal data. About what proportion of points will fall outside the limits by chance alone?
Correct answer: C. About 0.27%
About 99.73% of a normal distribution lies within plus or minus 3 sigma, so about 0.27%, or roughly 1 in 370 points, falls outside by chance. Five percent applies to 2 sigma limits.
About 99.73% of a normal distribution lies within plus or minus 3 sigma, so about 0.27%, or roughly 1 in 370 points, falls outside by chance. Five percent applies to 2 sigma limits.
Question 51Measure
A box plot of delivery times has a first quartile of 20 minutes and a third quartile of 30 minutes. Using the 1.5 x IQR rule, above what value would a point be flagged as an outlier?
Correct answer: D. 45 minutes
IQR = 30 - 20 = 10, and the upper fence = Q3 + 1.5 x IQR = 30 + 15 = 45. The value 40 adds only one IQR to Q3 instead of 1.5 x IQR.
IQR = 30 - 20 = 10, and the upper fence = Q3 + 1.5 x IQR = 30 + 15 = 45. The value 40 adds only one IQR to Q3 instead of 1.5 x IQR.
Question 52Analyze
Twelve operators were timed using both an old fixture and a new fixture. The paired differences are clearly non-normal. Which test is BEST?
Correct answer: C. Wilcoxon signed-rank test on the differences
The Wilcoxon signed-rank test is the nonparametric choice for paired data when differences are not normal. Mann-Whitney is for independent groups, not paired data.
The Wilcoxon signed-rank test is the nonparametric choice for paired data when differences are not normal. Mann-Whitney is for independent groups, not paired data.
Question 53Define
A warehouse completed a 5S program six months ago, but aisles are now cluttered again and labels are missing. Which 5S element was MOST likely neglected?
Correct answer: B. Sustain
Sustain is the discipline of keeping the system in place through audits, habits, and leadership follow-up. The earlier steps clearly worked at first, but without sustain the gains faded.
Sustain is the discipline of keeping the system in place through audits, habits, and leadership follow-up. The earlier steps clearly worked at first, but without sustain the gains faded.
Question 54Improve
A Black Belt has 11 potential factors and a tight budget, and needs to identify main effects only. Which design is MOST efficient?
Correct answer: C. A 12-run Plackett-Burman screening design
Plackett-Burman designs screen many factors for main effects in very few runs, such as 11 factors in 12 runs. A full factorial would be far too costly.
Plackett-Burman designs screen many factors for main effects in very few runs, such as 11 factors in 12 runs. A full factorial would be far too costly.
Question 55Control
Fifteen consecutive points on an Xbar chart all fall within 1 sigma of the centerline. What is the MOST likely cause?
Correct answer: A. Subgroups mix different streams, inflating the limits
Points hugging the centerline often mean the limits are too wide, for example because each subgroup mixes parts from different streams. It can also follow a real improvement, but the limits must be checked first.
Points hugging the centerline often mean the limits are too wide, for example because each subgroup mixes parts from different streams. It can also follow a real improvement, but the limits must be checked first.
Question 56Measure
A gage R&R study used 10 parts all taken from one hour of production, and percent study variation came out at 45%. What is the MOST likely issue with the study?
Correct answer: B. The parts did not span the process range, which inflates %GRR
Percent study variation compares measurement variation with total variation, so parts with very little spread make the gauge look worse. Parts should represent the full range of normal process variation.
Percent study variation compares measurement variation with total variation, so parts with very little spread make the gauge look worse. Parts should represent the full range of normal process variation.
Question 57Analyze
A hotel group collects ordinal 1-to-5 satisfaction ratings from four locations. The distributions have similar shapes and no extreme outliers. Which test BEST compares the locations?
Correct answer: A. Kruskal-Wallis test
Kruskal-Wallis compares three or more groups using ranks and suits ordinal data. ANOVA assumes a continuous, roughly normal response.
Kruskal-Wallis compares three or more groups using ranks and suits ordinal data. ANOVA assumes a continuous, roughly normal response.
Question 58Define
A pharmaceutical packer must, by regulation, perform a documented inspection of every lot. How should this step be classified in value stream analysis?
Correct answer: C. Business non-value-added, so minimize but keep it
Required steps that do not change the product are business non-value-added and should be made as lean as possible but not removed. Inspection does not transform the product, so it is not value-added.
Required steps that do not change the product are business non-value-added and should be made as lean as possible but not removed. Inspection does not transform the product, so it is not value-added.
Question 59Improve
In a 2^2 experiment, the responses are: A low and B low = 20, A high and B low = 30, A low and B high = 25, A high and B high = 45. What is the AB interaction effect?
Correct answer: D. 5
AB effect = average of runs where A and B match minus average where they differ = (20 + 45) / 2 - (30 + 25) / 2 = 32.5 - 27.5 = 5. The value 15 is the main effect of A.
AB effect = average of runs where A and B match minus average where they differ = (20 + 45) / 2 - (30 + 25) / 2 = 32.5 - 27.5 = 5. The value 15 is the main effect of A.
Question 60Control
How should a team decide how often to take subgroups for a control chart?
Correct answer: D. Often enough to catch shifts before much bad output is made
Sampling frequency balances the cost of sampling with the risk of producing bad output before a shift is detected. Processes that change quickly need more frequent sampling.
Sampling frequency balances the cost of sampling with the risk of producing bad output before a shift is detected. Processes that change quickly need more frequent sampling.
Question 61Measure
A measurement system has a standard deviation of 0.5, and the characteristic has a total tolerance of 20. What is the precision-to-tolerance ratio using 6 sigma?
Correct answer: A. 15%
P/T = 6 x sigma of measurement / tolerance = 6 x 0.5 / 20 = 3 / 20 = 15%. The value 2.5% forgets the factor of 6.
P/T = 6 x sigma of measurement / tolerance = 6 x 0.5 / 20 = 3 / 20 = 15%. The value 2.5% forgets the factor of 6.
Question 62Analyze
A team analyzed before-and-after data on the same 15 machines with a 2-sample t-test and found no significant difference. What is the MOST likely issue?
Correct answer: B. The test ignored the pairing and lost power
Treating paired data as independent leaves machine-to-machine variation in the error term, which hides real changes. A paired t-test removes that variation and is more sensitive.
Treating paired data as independent leaves machine-to-machine variation in the error term, which hides real changes. A paired t-test removes that variation and is more sensitive.
Question 63Define
A software company wants to uncover user needs that users cannot easily put into words. Which voice of the customer method is MOST effective?
Correct answer: D. Observing users as they work in their own setting
Observation in context reveals unspoken and latent needs that people cannot describe in a survey. Complaint logs only capture problems customers already know about.
Observation in context reveals unspoken and latent needs that people cannot describe in a survey. Complaint logs only capture problems customers already know about.
Question 64Improve
What is the MAIN purpose of replicating runs in a designed experiment?
Correct answer: C. Estimating experimental error to judge significance
Replication gives an estimate of pure error, which is needed to judge whether effects are larger than noise. Replication does not change resolution or add levels.
Replication gives an estimate of pure error, which is needed to judge whether effects are larger than noise. Replication does not change resolution or add levels.
Question 65Control
Six months after project closure, a new supplier and a new gauge are introduced on the line. What should happen to the control plan?
Correct answer: B. Update it to reflect the new conditions
A control plan is a living document that should be updated whenever the process, materials, or measurement methods change. Leaving it unchanged makes it inaccurate.
A control plan is a living document that should be updated whenever the process, materials, or measurement methods change. Leaving it unchanged makes it inaccurate.
Question 66Measure
A Black Belt must run a measurement system analysis on a tensile test that destroys each sample. Which approach is MOST appropriate?
Correct answer: C. A nested gage R&R using samples from uniform batches
When parts cannot be remeasured, a nested design uses samples from homogeneous batches so each batch acts like one part. A crossed design needs every operator to measure the same parts more than once.
When parts cannot be remeasured, a nested design uses samples from homogeneous batches so each batch acts like one part. A crossed design needs every operator to measure the same parts more than once.
Question 67Analyze
Data show that cities with more fire trucks also have more fire damage each year. What BEST explains this pattern?
Correct answer: A. A lurking variable, city size, drives both
A lurking variable can create correlation between two variables that do not cause each other. Larger cities have both more trucks and more fires.
A lurking variable can create correlation between two variables that do not cause each other. Larger cities have both more trucks and more fires.
Question 68Define
A project team agrees quickly on every decision, and members later admit they had doubts they never voiced. What should the Black Belt do to reduce this problem?
Correct answer: B. Assign a devil's advocate and invite anonymous input
This is groupthink, and structured dissent and anonymous input help bring hidden concerns forward. Letting a senior member decide would make the problem worse.
This is groupthink, and structured dissent and anonymous input help bring hidden concerns forward. Letting a senior member decide would make the problem worse.
Question 69Improve
A team wants a response surface design for three factors but cannot run any combination where all factors are at their extreme levels at once due to safety limits. Which design fits BEST?
Correct answer: A. Box-Behnken design
A Box-Behnken design places points at the midpoints of edges and the center, so it avoids running all factors at extremes together. A full factorial includes every corner combination.
A Box-Behnken design places points at the midpoints of edges and the center, so it avoids running all factors at extremes together. A full factorial includes every corner combination.
Question 70Control
An automated sensor records temperature every second, and an individuals chart on these readings shows frequent false alarms. What is the MOST likely cause?
Correct answer: C. Autocorrelation between consecutive readings
Readings taken very close in time are often correlated, which makes the moving range too small and the limits too tight. Sampling less often or using a time series model can help.
Readings taken very close in time are often correlated, which makes the moving range too small and the limits too tight. Sampling less often or using a time series model can help.
Question 71Measure
In an attribute study, each inspector agrees with their own earlier ratings 95% of the time but agrees with the expert standard only 70% of the time. What does this MOST likely indicate?
Correct answer: B. Inspectors are consistent but apply the wrong criteria
High within-appraiser agreement shows consistency, while low agreement with the standard shows a shared bias, often fixed by training and clearer criteria. Random guessing would also lower within-appraiser agreement.
High within-appraiser agreement shows consistency, while low agreement with the standard shows a shared bias, often fixed by training and clearer criteria. Random guessing would also lower within-appraiser agreement.
Question 72Analyze
A correlation of r = 0.70 between two variables drops to r = 0.15 when one data point is removed. What does this show?
Correct answer: D. One influential point drives the correlation
A single extreme point can create or hide a correlation, which is why data should be plotted before r is trusted. The point should be investigated, not simply kept or deleted.
A single extreme point can create or hide a correlation, which is why data should be plotted before r is trusted. The point should be investigated, not simply kept or deleted.
Question 73Define
A Black Belt wants to list and weigh the forces that support and oppose a planned change to shift schedules. Which tool fits BEST?
Correct answer: A. Force field analysis
Force field analysis lists driving and restraining forces so the team can strengthen supports and reduce barriers. A Pareto chart ranks problem categories, not change forces.
Force field analysis lists driving and restraining forces so the team can strengthen supports and reduce barriers. A Pareto chart ranks problem categories, not change forces.
Question 74Improve
A Black Belt is choosing low and high levels for factors in a DOE. Which guideline is BEST?
Correct answer: B. Set levels far enough apart to show an effect but within safe limits
Levels that are too close may hide real effects in noise, while levels outside safe limits risk damage or harm. Product specification limits apply to outputs, not to input factor settings.
Levels that are too close may hide real effects in noise, while levels outside safe limits risk damage or harm. Product specification limits apply to outputs, not to input factor settings.
Question 75Control
New hires at a call center keep making errors even though a written procedure exists. Which control action is MOST effective?
Correct answer: D. Train with demonstration and observed practice
Hands-on training with demonstration and verified practice builds the skill to follow standard work. Repeating the written document has already failed.
Hands-on training with demonstration and verified practice builds the skill to follow standard work. Repeating the written document has already failed.
Question 76Measure
All gauges at a plant are within their calibration dates, yet a gage R&R study shows poor results. How is this possible?
Correct answer: D. Calibration checks accuracy, not repeat and operator variation
Calibration confirms a gauge reads correctly against a standard, but it does not capture variation from repeated readings or different operators. Gage R&R measures that precision.
Calibration confirms a gauge reads correctly against a standard, but it does not capture variation from repeated readings or different operators. Gage R&R measures that precision.
Question 77Analyze
A multi-vari chart for machined rings shows large differences between readings at different points around each ring, but small differences between rings and over time. Where should the team focus?
Correct answer: A. Positional causes such as fixturing or spindle alignment
Large within-piece variation is positional and points to causes such as fixturing, clamping, or machine alignment. Shift or lot causes would show up as time-to-time variation.
Large within-piece variation is positional and points to causes such as fixturing, clamping, or machine alignment. Shift or lot causes would show up as time-to-time variation.
Question 78Define
A stakeholder commitment chart shows the IT manager is currently letting it happen, but the project needs the manager to help it happen. What does this gap tell the Black Belt?
Correct answer: C. Plan targeted actions to raise the manager's involvement
A commitment chart compares current and required support, and any gap needs a plan to close it. Lack of opposition is not enough when active help is required.
A commitment chart compares current and required support, and any gap needs a plan to close it. Lack of opposition is not enough when active help is required.
Question 79Improve
A plant wants to improve a running process by making small, planned changes to settings during normal production without making scrap. Which approach fits BEST?
Correct answer: A. Evolutionary operation (EVOP)
EVOP makes small, structured changes during routine production and analyzes results over many cycles. Large screening experiments use wide factor ranges that may produce defective output.
EVOP makes small, structured changes during routine production and analyzes results over many cycles. Large screening experiments use wide factor ranges that may produce defective output.
Question 80Control
A plant paints green, yellow, and red zones on its pressure gauges so anyone can tell at a glance whether readings are normal. What kind of control is this?
Correct answer: B. Visual management
Visual management makes normal and abnormal conditions obvious at a glance, so problems are caught quickly. SPC uses control charts and statistical rules.
Visual management makes normal and abnormal conditions obvious at a glance, so problems are caught quickly. SPC uses control charts and statistical rules.
Question 81Measure
A process has a mean of 52, LSL of 40, USL of 60, and an overall standard deviation of 2.5. What is Ppk?
Correct answer: B. 1.07
Ppu = (60 - 52) / 7.5 = 1.07 and Ppl = (52 - 40) / 7.5 = 1.60, so Ppk is the smaller value, 1.07. The value 1.33 is Pp, which ignores centering.
Ppu = (60 - 52) / 7.5 = 1.07 and Ppl = (52 - 40) / 7.5 = 1.60, so Ppk is the smaller value, 1.07. The value 1.33 is Pp, which ignores centering.
Question 82Analyze
A defect appears on Line 2 but never on Line 1, and only since a supplier change in March. Which tool BEST structures this comparison to narrow the possible causes?
Correct answer: C. Is/Is-not analysis
Is/Is-not analysis compares where, when, and what the problem is with where it could be but is not, which narrows likely causes. Kano analysis classifies customer needs, not causes.
Is/Is-not analysis compares where, when, and what the problem is with where it could be but is not, which narrows likely causes. Kano analysis classifies customer needs, not causes.
Question 83Define
A team ranks possible projects on a two-by-two grid of effort versus impact. Which quadrant should usually be pursued FIRST?
Correct answer: B. Low effort, high impact
Low effort, high impact items deliver the most value for the resources used, so they are often called quick wins. High effort, high impact items are worthwhile but need more planning.
Low effort, high impact items deliver the most value for the resources used, so they are often called quick wins. High effort, high impact items are worthwhile but need more planning.
Question 84Improve
A pilot of a new emergency triage process is being planned. Which pilot design is BEST?
Correct answer: D. A typical unit, a set period, and clear success measures
A pilot should run where conditions are typical, with defined measures, so results predict full-scale performance. Choosing the best unit gives an overly positive result.
A pilot should run where conditions are typical, with defined measures, so results predict full-scale performance. Choosing the best unit gives an overly positive result.
Question 85Control
Which inspection approach is MOST effective at preventing defects?
Correct answer: A. Source inspection that checks conditions before work
Source inspection checks for the conditions that cause errors before the work is done, so defects are not created. The other methods find defects after they occur.
Source inspection checks for the conditions that cause errors before the work is done, so defects are not created. The other methods find defects after they occur.
Question 86Measure
A stable, normal process has Cpk = 1.0 on the upper side, and the lower limit is very far from the mean. About how many parts per million exceed the upper limit?
Correct answer: C. About 1,350
Cpk = 1.0 means the upper limit is 3 sigma from the mean, and the area beyond z = 3 is about 0.135%, or 1,350 ppm. The value 2,700 counts both tails, but the lower tail is negligible here.
Cpk = 1.0 means the upper limit is 3 sigma from the mean, and the area beyond z = 3 is about 0.135%, or 1,350 ppm. The value 2,700 counts both tails, but the lower tail is negligible here.
Question 87Analyze
In a fault tree, a top event occurs only if both the main pump fails and the backup alarm fails. Which gate connects these two events?
Correct answer: A. AND gate
An AND gate means the output event occurs only when all input events occur. An OR gate would mean either failure alone causes the top event.
An AND gate means the output event occurs only when all input events occur. An OR gate would mean either failure alone causes the top event.
Question 88Define
A project needs a $60,000 investment and is expected to save $20,000 per year. What is the simple payback period?
Correct answer: D. 3 years
Simple payback = investment / annual savings = $60,000 / $20,000 = 3 years. The value 0.33 years inverts the ratio.
Simple payback = investment / annual savings = $60,000 / $20,000 = 3 years. The value 0.33 years inverts the ratio.
Question 89Improve
After a successful pilot, what should the full-scale implementation plan include?
Correct answer: A. Tasks, owners, timeline, training, and communication
An implementation plan defines who does what and when, along with training and communication, to make rollout smooth. Pilot statistics support the plan but do not replace it.
An implementation plan defines who does what and when, along with training and communication, to make rollout smooth. Pilot statistics support the plan but do not replace it.
Question 90Control
How should a Black Belt confirm in the Control phase that a process improvement is real?
Correct answer: C. Compare before and after data with a test and a stable chart
A hypothesis test shows the change is statistically significant, and a stable control chart shows the new level is sustained. One good week or a model prediction does not prove lasting improvement.
A hypothesis test shows the change is statistically significant, and a stable control chart shows the new level is sustained. One good week or a model prediction does not prove lasting improvement.
Question 91Measure
A coating must be at least 25 microns thick, and there is no upper limit. Which capability index is appropriate?
Correct answer: D. Cpl, the lower one-sided index
With only a lower limit, capability is judged by Cpl = (mean - LSL) / 3 sigma. Cp needs both limits and cannot be computed.
With only a lower limit, capability is judged by Cpl = (mean - LSL) / 3 sigma. Cp needs both limits and cannot be computed.
Question 92Analyze
A charter goal is to cut warranty cost. A Pareto by frequency shows minor label errors on top, but a Pareto by cost shows rare part failures cost the most. Which should drive project focus?
Correct answer: D. The cost Pareto, since it matches the goal
When the goal is to reduce cost, the Pareto should be weighted by cost so effort goes where the money is. Frequency alone can point to many small, cheap problems.
When the goal is to reduce cost, the Pareto should be weighted by cost so effort goes where the money is. Frequency alone can point to many small, cheap problems.
Question 93Define
A food plant pays for lab testing of incoming ingredients and routine sampling of finished product. In a cost of quality model, these costs are classified as what?
Correct answer: B. Appraisal costs
Appraisal costs cover inspection, testing, and audits to find defects. Prevention costs are spent to stop defects from happening in the first place, such as training or design reviews.
Appraisal costs cover inspection, testing, and audits to find defects. Prevention costs are spent to stop defects from happening in the first place, such as training or design reviews.
Question 94Improve
A batch of 50 units moves through 3 steps, and each step takes 1 minute per unit. The whole batch must finish a step before moving on. If the line switches to one-piece flow, about how long until all 50 units are done?
Correct answer: C. 52 minutes
With full batches, lead time is 3 x 50 = 150 minutes. With one-piece flow, the first unit exits after 3 minutes and one unit exits every minute after, so the total is 50 + 2 = 52 minutes.
With full batches, lead time is 3 x 50 = 150 minutes. With one-piece flow, the first unit exits after 3 minutes and one unit exits every minute after, so the total is 50 + 2 = 52 minutes.
Question 95Control
Who should sign off on the financial benefits claimed by a Six Sigma project?
Correct answer: B. A finance representative who validates the savings
An independent finance partner confirms that savings are real and measured correctly, which protects credibility. The Black Belt has a natural bias toward reporting larger savings.
An independent finance partner confirms that savings are real and measured correctly, which protects credibility. The Black Belt has a natural bias toward reporting larger savings.
Question 96Measure
Cycle time data from a stable process are strongly right-skewed and fail a normality test. What is the BEST way to assess capability?
Correct answer: C. Fit a suitable distribution or transform the data
Normal-based indices give misleading defect rates for skewed data, so the analyst should fit a distribution such as Weibull or use a transformation. Deleting long times hides real performance.
Normal-based indices give misleading defect rates for skewed data, so the analyst should fit a distribution such as Weibull or use a transformation. Deleting long times hides real performance.
Question 97Analyze
To cut the margin of error of a mean estimate in half, how must the sample size change?
Correct answer: B. It must be quadrupled
Margin of error is proportional to 1 / square root of n, so halving it needs four times the sample size. Doubling n only reduces the margin by about 29%.
Margin of error is proportional to 1 / square root of n, so halving it needs four times the sample size. Doubling n only reduces the margin by about 29%.
Question 98Define
Which spending is a prevention cost of quality?
Correct answer: C. Training operators on a new procedure before launch
Prevention costs are spent before defects occur, such as training, planning, and error-proofing. Inspecting finished goods is an appraisal cost.
Prevention costs are spent before defects occur, such as training, planning, and error-proofing. Inspecting finished goods is an appraisal cost.
Question 99Improve
Machines for one product family are spread across separate departments, causing long moves and queues between steps. Which layout change BEST supports flow?
Correct answer: A. Place the machines in sequence in a U-shaped cell
A cell arranges equipment in process order so parts flow with little transport or waiting. More forklifts or larger batches treat symptoms and add waste.
A cell arranges equipment in process order so parts flow with little transport or waiting. More forklifts or larger batches treat symptoms and add waste.
Question 100Control
A project cut defects sharply at one plant, and three sister plants run the same process. What is the BEST next step to spread the benefit?
Correct answer: D. Replicate the solution after checking local differences
Replication spreads proven gains, but local differences in equipment or materials should be checked before transfer. Copying settings blindly can fail where conditions differ.
Replication spreads proven gains, but local differences in equipment or materials should be checked before transfer. Copying settings blindly can fail where conditions differ.
Question 101Measure
A process shows Cpk of 1.6 but Ppk of only 0.9. What does this gap MOST likely indicate?
Correct answer: A. Large shifts occur between subgroups over time
Cpk uses within-subgroup variation, while Ppk uses overall variation, so a big gap means extra variation between subgroups, such as shifts or drift. The process is likely not stable over time.
Cpk uses within-subgroup variation, while Ppk uses overall variation, so a big gap means extra variation between subgroups, such as shifts or drift. The process is likely not stable over time.
Question 102Analyze
Which set of inputs is needed to calculate the sample size for a 2-sample t-test?
Correct answer: C. Difference to detect, standard deviation, alpha, and power
Sample size depends on the size of difference that matters, process variation, alpha, and desired power. The p-value is a result of the test, not an input.
Sample size depends on the size of difference that matters, process variation, alpha, and desired power. The p-value is a result of the test, not an input.
Question 103Define
An inspector finds 3 scratches on one painted panel and no issues on the other 9 panels in the sample. How many defects and defectives are in this sample?
Correct answer: D. 3 defects, 1 defective
A defect is each nonconformity, while a defective is a unit with one or more defects. One panel has three scratches, so there are 3 defects and 1 defective unit.
A defect is each nonconformity, while a defective is a unit with one or more defects. One panel has three scratches, so there are 3 defects and 1 defective unit.
Question 104Improve
A plant builds all of Model A on Monday and all of Model B on Tuesday, which causes uneven demand on suppliers and long waits for customers. Which Lean method addresses this?
Correct answer: D. Heijunka, leveling mix and volume
Heijunka levels production by mixing models in small, repeating batches, which smooths demand on upstream processes. SMED supports heijunka by making frequent changeovers possible, but it is not the leveling method itself.
Heijunka levels production by mixing models in small, repeating batches, which smooths demand on upstream processes. SMED supports heijunka by making frequent changeovers possible, but it is not the leveling method itself.
Question 105Control
Which practice BEST helps sustain project gains over the long term?
Correct answer: B. Include key metrics in regular management reviews
When key metrics are reviewed regularly by management, problems get attention and gains are sustained. Stopping tracking after closure invites backsliding.
When key metrics are reviewed regularly by management, problems get attention and gains are sustained. Stopping tracking after closure invites backsliding.
Question 106Measure
A Black Belt wants to estimate long-term capability for a bottling process. Which data collection approach is BEST?
Correct answer: B. Sample over weeks to include lots, shifts, and operators
Long-term capability must include all normal sources of variation, such as material lots, shifts, and operators. Consecutive samples from one morning reflect short-term variation only.
Long-term capability must include all normal sources of variation, such as material lots, shifts, and operators. Consecutive samples from one morning reflect short-term variation only.
Question 107Analyze
A Black Belt has 200 observations per group, and both groups are mildly non-normal. Can a 2-sample t-test be used to compare means?
Correct answer: D. Yes, with large samples the t-test is robust to mild non-normality
With large samples, the sampling distribution of the mean is close to normal, so t-tests handle mild non-normality well. Nonparametric tests are not required in this case.
With large samples, the sampling distribution of the mean is close to normal, so t-tests handle mild non-normality well. Nonparametric tests are not required in this case.
Question 108Define
A team counts 200 opportunities per invoice, including many fields that never go wrong, when calculating DPMO. What is the MAIN problem with this approach?
Correct answer: C. It inflates the denominator and makes sigma look better
Counting opportunities that cannot realistically fail enlarges the denominator, which lowers DPMO and inflates the sigma level. Opportunities should be meaningful and defined the same way over time.
Counting opportunities that cannot realistically fail enlarges the denominator, which lowers DPMO and inflates the sigma level. Opportunities should be meaningful and defined the same way over time.
Question 109Improve
A hospital supply room keeps two bins of each item. When the first bin empties, it is sent for refill while staff use the second bin. What is this system called?
Correct answer: B. Two-bin kanban pull system
A two-bin system is a simple kanban, where an empty bin is the signal to replenish. It is driven by actual use, not by a forecast.
A two-bin system is a simple kanban, where an empty bin is the signal to replenish. It is driven by actual use, not by a forecast.
Question 110Control
In pre-control, the green zone typically covers the middle half of the tolerance. What is pre-control MOST often used for?
Correct answer: C. Quick setup approval and simple monitoring by operators
Pre-control uses zones based on the specification to give operators simple go, caution, and stop rules, often during setup. It is less sensitive than SPC and is not used for capability studies.
Pre-control uses zones based on the specification to give operators simple go, caution, and stop rules, often during setup. It is less sensitive than SPC and is not used for capability studies.
Question 111Measure
An FMEA team wants to lower the risk of a high-severity failure mode. Which action is generally MOST effective?
Correct answer: A. Change the design or process to prevent the cause
Preventing the cause reduces occurrence at the source, and only a design change can truly lower severity. Added inspection only improves detection and leaves the failure in place.
Preventing the cause reduces occurrence at the source, and only a design change can truly lower severity. Added inspection only improves detection and leaves the failure in place.
Question 112Analyze
Which graphical tool is BEST for visually checking whether a data set follows a normal distribution?
Correct answer: B. Normal probability plot
On a normal probability plot, normal data fall close to a straight line, making departures easy to see. A run chart shows time order, not distribution shape.
On a normal probability plot, normal data fall close to a straight line, making departures easy to see. A run chart shows time order, not distribution shape.
Question 113Define
What is the MAIN purpose of a tollgate review at the end of each DMAIC phase?
Correct answer: A. Confirm phase deliverables are complete before moving on
Tollgate reviews let the Champion and Black Belt confirm that each phase is done well and decide whether to proceed, adjust, or stop. Final savings are validated at project closure, not at every tollgate.
Tollgate reviews let the Champion and Black Belt confirm that each phase is done well and decide whether to proceed, adjust, or stop. Final savings are validated at project closure, not at every tollgate.
Question 114Improve
A weaving loom automatically stops when a thread breaks so that defective cloth is not produced. Which Lean principle does this illustrate?
Correct answer: C. Jidoka (automation with a human touch)
Jidoka builds quality in by stopping the process at the first abnormality so defects are not passed on. Heijunka is about leveling the mix and volume of production.
Jidoka builds quality in by stopping the process at the first abnormality so defects are not passed on. Heijunka is about leveling the mix and volume of production.
Question 115Control
A plant manager follows a written daily routine to check visual boards, review key metrics, and coach supervisors on the floor. What is this practice called?
Correct answer: D. Leader standard work
Leader standard work defines the routine checks and coaching that leaders do to sustain standards. Value stream mapping is an analysis tool, not a daily routine.
Leader standard work defines the routine checks and coaching that leaders do to sustain standards. Value stream mapping is an analysis tool, not a daily routine.
Question 116Measure
A team is analyzing how mistakes made while assembling a pump on the production line could cause leaks. Which type of FMEA fits this work?
Correct answer: D. Process FMEA
A process FMEA examines how manufacturing or service steps can fail. A design FMEA looks at weaknesses in the product design itself.
A process FMEA examines how manufacturing or service steps can fail. A design FMEA looks at weaknesses in the product design itself.
Question 117Analyze
Sixteen parts were each measured once with Gauge A and once with Gauge B. The mean of the paired differences is 3 and the standard deviation of the differences is 4. What is the paired t statistic?
Correct answer: A. 3.0
t = mean difference / (sd of differences / square root of n) = 3 / (4 / 4) = 3.0. The value 0.75 divides by the standard deviation instead of the standard error.
t = mean difference / (sd of differences / square root of n) = 3 / (4 / 4) = 3.0. The value 0.75 divides by the standard deviation instead of the standard error.
Question 118Define
In a hospital, the lab produces test results that emergency physicians use to make decisions. In SIPOC terms, how are the physicians best described?
Correct answer: B. Internal customers of the lab process
Anyone inside the organization who receives the output of a process is an internal customer. The physicians receive the lab's output, so they are customers, not suppliers.
Anyone inside the organization who receives the output of a process is an internal customer. The physicians receive the lab's output, so they are customers, not suppliers.
Question 119Improve
A Black Belt cuts processing time at a non-bottleneck step by 30%. What is the MOST likely effect on overall throughput?
Correct answer: D. Little or no change, since the bottleneck still limits output
System output is set by the constraint, so improving a non-bottleneck adds capacity that cannot be used. Effort should focus on the bottleneck first.
System output is set by the constraint, so improving a non-bottleneck adds capacity that cannot be used. Effort should focus on the bottleneck first.
Question 120Control
Why should a control plan include periodic re-verification of the measurement system?
Correct answer: C. Gauges can drift or wear and give false readings
Measurement systems change over time, so periodic checks ensure control chart signals reflect the process and not the gauge. MSA does not replace process monitoring.
Measurement systems change over time, so periodic checks ensure control chart signals reflect the process and not the gauge. MSA does not replace process monitoring.
Question 121Measure
A Black Belt can analyze a crossed gage R&R study with either the ANOVA method or the Xbar and R method. What is the main advantage of the ANOVA method?
Correct answer: A. It can estimate the operator-by-part interaction
The ANOVA method separates an operator-by-part interaction, which shows when operators measure some parts differently, and it uses the data more efficiently. The Xbar and R method cannot estimate that interaction.
The ANOVA method separates an operator-by-part interaction, which shows when operators measure some parts differently, and it uses the data more efficiently. The Xbar and R method cannot estimate that interaction.
Question 122Analyze
A survey of 400 customers finds that 20% report a billing problem. What is the approximate 95% confidence interval for the true proportion?
Correct answer: C. 16.1% to 23.9%
Standard error = square root of (0.2 x 0.8 / 400) = 0.02, and 1.96 x 0.02 = 0.039, giving 16.1% to 23.9%. The range 18% to 22% uses one standard error instead of 1.96.
Standard error = square root of (0.2 x 0.8 / 400) = 0.02, and 1.96 x 0.02 = 0.039, giving 16.1% to 23.9%. The range 18% to 22% uses one standard error instead of 1.96.
Question 123Define
A Black Belt wants a metric that signals problems before customer complaints arrive. Which is the BEST leading indicator for an order fulfillment process?
Correct answer: A. Percent of orders picked late each day
A leading indicator moves before the final outcome, so daily late picks warn of delivery problems early. Satisfaction scores and lost revenue are lagging indicators.
A leading indicator moves before the final outcome, so daily late picks warn of delivery problems early. Satisfaction scores and lost revenue are lagging indicators.
Question 124Improve
A plant wants operators to take on routine cleaning, lubrication, and inspection of their own machines to reduce breakdowns. Which practice does this describe?
Correct answer: B. Autonomous maintenance within TPM
Autonomous maintenance is a pillar of total productive maintenance in which operators handle basic care so problems are found early. SMED focuses on shortening changeovers, not on routine machine care.
Autonomous maintenance is a pillar of total productive maintenance in which operators handle basic care so problems are found early. SMED focuses on shortening changeovers, not on routine machine care.
Question 125Control
A job shop runs many different part numbers with different targets in small batches on one machine. Which SPC approach is MOST practical?
Correct answer: A. Plot deviation from each part's nominal on one chart
A deviation-from-nominal chart subtracts each part's target, so different parts can share one chart if their variation is similar. Separate charts would never have enough data.
A deviation-from-nominal chart subtracts each part's target, so different parts can share one chart if their variation is similar. Separate charts would never have enough data.
Question 126Measure
A Black Belt wants to show the physical path a nurse walks while preparing medications on a hospital unit. Which tool is BEST?
Correct answer: C. Spaghetti diagram
A spaghetti diagram traces the actual movement of people or materials on a layout. A swimlane map shows responsibilities and handoffs, not physical walking paths.
A spaghetti diagram traces the actual movement of people or materials on a layout. A swimlane map shows responsibilities and handoffs, not physical walking paths.
Question 127Analyze
A Black Belt runs three independent hypothesis tests, each at alpha = 0.05. What is the chance of at least one false positive if all null hypotheses are true?
Correct answer: B. About 14.3%
The chance of no false positives is 0.95 cubed = 0.857, so at least one is 1 - 0.857 = 14.3%. This is why multiple comparisons need adjustment.
The chance of no false positives is 0.95 cubed = 0.857, so at least one is 1 - 0.857 = 14.3%. This is why multiple comparisons need adjustment.
Question 128Define
In the Measure phase, a Black Belt finds that the problem in the charter is much smaller than estimated, and savings would be under 10% of the target. What is the BEST action?
Correct answer: D. Show the data to the Champion and discuss rescoping or closing
When data show a project is not worth the effort, the Black Belt should bring it to the Champion so resources can be used well. Continuing blindly wastes effort, and adjusting data is unethical.
When data show a project is not worth the effort, the Black Belt should bring it to the Champion so resources can be used well. Continuing blindly wastes effort, and adjusting data is unethical.
Question 129Improve
When is a rapid improvement (kaizen) event MOST appropriate instead of a full DMAIC project?
Correct answer: A. Scope is narrow, causes are fairly clear, and fixes take days
Kaizen events suit focused problems with fairly clear causes that a team can fix in a few days. Complex problems with unknown causes need the full DMAIC approach.
Kaizen events suit focused problems with fairly clear causes that a team can fix in a few days. Complex problems with unknown causes need the full DMAIC approach.
Question 130Control
Why should a control plan focus on monitoring key input variables, not only the output Y?
Correct answer: B. Controlling inputs prevents defects before they occur
Monitoring the critical x's found in Analyze lets the team act before the output goes bad. Watching only the Y detects problems after defects are made.
Monitoring the critical x's found in Analyze lets the team act before the output goes bad. Watching only the Y detects problems after defects are made.
Question 131Measure
Why must a Black Belt establish baseline performance in the Measure phase?
Correct answer: A. To compare later results and verify improvement
A baseline gives a reference point so the team can later prove that changes made a real difference. Choosing solutions before analysis would skip the Analyze phase.
A baseline gives a reference point so the team can later prove that changes made a real difference. Choosing solutions before analysis would skip the Analyze phase.
Question 132Analyze
The overall defect rate looks stable, but the Black Belt suspects one of three suppliers is worse than the others. What should be done FIRST with the existing data?
Correct answer: D. Stratify the defect data by supplier
Stratification splits data by a factor such as supplier, which can reveal differences hidden in the combined data. A single chart of all data would mask supplier differences.
Stratification splits data by a factor such as supplier, which can reveal differences hidden in the combined data. A single chart of all data would mask supplier differences.
Question 133Define
Which activity is a typical responsibility of a Black Belt in a Six Sigma deployment?
Correct answer: A. Leading complex projects and coaching Green Belts
Black Belts lead complex, cross-functional projects and mentor Green Belts. Setting strategy and approving budgets belong to executives and Champions.
Black Belts lead complex, cross-functional projects and mentor Green Belts. Setting strategy and approving budgets belong to executives and Champions.
Question 134Improve
A hospital team studies how a top-performing airline turns aircraft around quickly to get ideas for faster operating room turnover. What technique is this?
Correct answer: C. Benchmarking
Benchmarking learns from best practices in other organizations, even in different industries. Kano analysis classifies customer requirements.
Benchmarking learns from best practices in other organizations, even in different industries. Kano analysis classifies customer requirements.
Question 135Control
An Xbar-S chart uses subgroups of 9. The grand average is 100.0 and the average standard deviation is 2.0. With A3 = 1.032, what is the upper control limit for the Xbar chart?
Correct answer: A. 102.06
UCL = grand average + A3 x S-bar = 100.0 + 1.032 x 2.0 = 100.0 + 2.06 = 102.06. The value 106.00 uses 3 x S-bar without adjusting for subgroup size.
UCL = grand average + A3 x S-bar = 100.0 + 1.032 x 2.0 = 100.0 + 2.06 = 102.06. The value 106.00 uses 3 x S-bar without adjusting for subgroup size.
Question 136Measure
A gauge gives readings that are tightly clustered, but all of them are about 0.5 units above the true value. How is this measurement system BEST described?
Correct answer: D. Precise but not accurate
Tight clustering shows good precision, while the consistent offset shows bias, which means poor accuracy. Accuracy refers to closeness to the true value.
Tight clustering shows good precision, while the consistent offset shows bias, which means poor accuracy. Accuracy refers to closeness to the true value.
Question 137Analyze
A team wants to study how oven temperature, a continuous variable, relates to bread moisture, also continuous. Which tool fits BEST?
Correct answer: B. Scatter plot with regression analysis
When both the input and output are continuous, a scatter plot with correlation or regression describes the relationship. Chi-square tests are for categorical variables.
When both the input and output are continuous, a scatter plot with correlation or regression describes the relationship. Chi-square tests are for categorical variables.
Question 138Define
A three-step loan process can handle 50, 30, and 45 applications per day at each step, in that order. What is the maximum daily output of the whole process?
Correct answer: C. 30 applications
A process can produce no faster than its bottleneck, which is the 30-per-day step. Averaging the steps gives 41.7, which ignores the constraint.
A process can produce no faster than its bottleneck, which is the 30-per-day step. Averaging the steps gives 41.7, which ignores the constraint.
Question 139Improve
In a weighted criteria matrix, Solution A scores 4, 3, and 5 on three criteria with weights of 5, 3, and 2. What is Solution A's total weighted score?
Correct answer: B. 39
Weighted score = 4 x 5 + 3 x 3 + 5 x 2 = 20 + 9 + 10 = 39. The value 12 is the unweighted sum of the scores.
Weighted score = 4 x 5 + 3 x 3 + 5 x 2 = 20 + 9 + 10 = 39. The value 12 is the unweighted sum of the scores.
Question 140Control
The calculated lower control limit for a c chart comes out to -2.1. What should be done?
Correct answer: D. Set the lower control limit to zero
Counts cannot be negative, so a negative lower limit is set to zero, meaning there is no lower limit in practice. The chart is still valid.
Counts cannot be negative, so a negative lower limit is set to zero, meaning there is no lower limit in practice. The chart is still valid.
Question 141Measure
A Black Belt wants to estimate a mean within plus or minus 2 units at 95% confidence. The estimated standard deviation is 10. What is the minimum sample size?
Correct answer: B. 97
n = (1.96 x 10 / 2) squared = 9.8 squared = 96.04, which rounds up to 97. Sample size must always be rounded up to keep the required precision.
n = (1.96 x 10 / 2) squared = 9.8 squared = 96.04, which rounds up to 97. Sample size must always be rounded up to keep the required precision.
Question 142Analyze
A data set of cycle times includes one value ten times larger than the rest. What should the Black Belt do FIRST?
Correct answer: C. Investigate its cause before keeping or excluding it
An outlier may be a data entry error or a real event that reveals an important cause. Deleting it without investigation can hide valuable information.
An outlier may be a data entry error or a real event that reveals an important cause. Deleting it without investigation can hide valuable information.
Question 143Define
A molding cell started 500 parts. Of these, 40 were scrapped and 60 were reworked and then passed. What is the first pass yield?
Correct answer: A. 80%
First pass yield counts only units that passed the first time: (500 - 40 - 60) / 500 = 400 / 500 = 80%. The value 92% is the final yield, which counts reworked units as good.
First pass yield counts only units that passed the first time: (500 - 40 - 60) / 500 = 400 / 500 = 80%. The value 92% is the final yield, which counts reworked units as good.
Question 144Improve
A Black Belt wants a product to work consistently even though customer homes vary widely in humidity, which cannot be controlled. Which approach fits BEST?
Correct answer: D. Robust design that reduces sensitivity to noise
Robust parameter design finds settings of controllable factors that make performance insensitive to uncontrollable noise factors such as humidity. Inspection cannot fix variation that occurs in the customer's home.
Robust parameter design finds settings of controllable factors that make performance insensitive to uncontrollable noise factors such as humidity. Inspection cannot fix variation that occurs in the customer's home.
Question 145Control
A team updates a work instruction after a project. What is the MOST important step to make sure the change sticks?
Correct answer: A. Control the revision so only the current version is in use
Document control ensures old versions are removed and everyone works from the current standard. Leaving old copies in use leads people back to the old method.
Document control ensures old versions are removed and everyone works from the current standard. Leaving old copies in use leads people back to the old method.
Question 146Measure
A Black Belt has 6 months of daily defect rates. Before building a histogram, what should be plotted to check for shifts or trends?
Correct answer: C. A run chart or time series plot
A histogram hides time order, so a run chart should come first to reveal shifts, trends, or cycles. A histogram of data from a shifting process can be misleading.
A histogram hides time order, so a run chart should come first to reveal shifts, trends, or cycles. A histogram of data from a shifting process can be misleading.
Question 147Analyze
Two judges each rank 10 product designs from best to worst. Which measure BEST assesses how closely their rankings agree?
Correct answer: D. Spearman rank correlation
Spearman rank correlation measures association between two sets of ranks and does not assume a linear relationship or normal data. Pearson correlation is meant for continuous measurements.
Spearman rank correlation measures association between two sets of ranks and does not assume a linear relationship or normal data. Pearson correlation is meant for continuous measurements.
Question 148Define
Two team members disagree strongly about which data to collect, and the argument is slowing the project. What is the BEST approach for the Black Belt?
Correct answer: D. Facilitate a discussion based on data needs and goals
Focusing the discussion on the project goals and what the data must answer turns a personal conflict into a shared problem. Siding with seniority can damage trust and lead to poor choices.
Focusing the discussion on the project goals and what the data must answer turns a personal conflict into a shared problem. Siding with seniority can damage trust and lead to poor choices.
Question 149Improve
A DOE model predicts the best settings will give 92% yield. What should the team do before full implementation?
Correct answer: B. Run confirmation trials at those settings
Confirmation runs check that the predicted result actually occurs under real conditions. Rolling out on all lines without confirmation risks acting on a flawed model.
Confirmation runs check that the predicted result actually occurs under real conditions. Rolling out on all lines without confirmation risks acting on a flawed model.
Question 150Control
When is a project ready to be closed and handed over to the process owner?
Correct answer: C. When the improved process is stable for an agreed period
Closure should wait until data show the improvement is sustained and the owner can run the control plan. Closing at installation risks losing gains before they are proven.
Closure should wait until data show the improvement is sustained and the owner can run the control plan. Closing at installation risks losing gains before they are proven.
The real Lean Six Sigma Black Belt exam at a glance
150questions
4htime limit
MCQmultiple choice
75%+aim for here
Score 75% or more on two different sets before you book the real exam. Need a plan for your weak topics? See how we help you pass, or get a full refund.
More free Six Sigma & ITIL practice tests
Lean Six Sigma Green Belt Practice Test100 questions · DMAICITIL 4 Foundation Practice Test40 questions · ITIL 4
See all 36 free practice tests →
Original practice questions written by CertOcean in the style of the Lean Six Sigma Black Belt exam. They are not real exam questions. Lean Six Sigma Black Belt and Lean Six Sigma are trademarks of their owners. Real exam format: 150 questions · 4 hours · about 30 questions per DMAIC phase · common pass mark around 70%.