Lean Six Sigma Green Belt Practice Exam — All Questions

A designed experiment (DOE) that studies two factors, each at two levels, in every combination is called a:

• a.One-factor-at-a-time (OFAT) study
• b.Fractional factorial design
• c.Full factorial 2^2 design✓
• d.Simple linear regression

Testing every combination of 2 factors at 2 levels each is a full factorial 2^2 design (four runs). Changing one factor at a time is OFAT, and a fractional factorial deliberately runs only a subset of the combinations.

Poka-yoke, applied in the Improve phase, refers to:

• a.A statistical sampling plan
• b.A supplier scorecard
• c.Mistake-proofing that prevents or immediately detects errors✓
• d.A capacity-planning calculation

Poka-yoke is error- or mistake-proofing that makes a defect impossible or obvious, such as a connector that only fits one way. It is a prevention technique, not a sampling, capacity, or supplier-rating method.

A process has available production time of 480 minutes per shift and customer demand of 240 units per shift. What is the takt time?

• a.480 minutes per unit
• b.2 minutes per unit✓
• c.0.5 minutes per unit
• d.240 minutes per unit

Takt time = available time / customer demand = 480 / 240 = 2 minutes per unit, the pace needed to meet demand. Inverting the ratio or ignoring demand gives the wrong result.

The overall purpose of the Improve phase in DMAIC is to:

• a.Develop, test, and implement solutions that address the verified root causes✓
• b.Validate that the measurement system is accurate and repeatable
• c.Rank the observed defects from the most to the least frequent
• d.Define the overall project scope and draft the initial team charter document

Improve is where the team generates, selects, pilots, and implements solutions that attack the root causes confirmed in Analyze. Chartering is Define, measurement validation is Measure, and Pareto ranking is an Analyze activity.

A full factorial experiment with 3 factors, each at 2 levels, requires how many runs for a single replicate?

• a.6, from multiplying the 3 factors by the 2 levels
• b.9, from squaring the number of factors in the design being studied
• c.3, from counting one run for each factor studied
• d.8, from two levels raised to the power of three factors✓

A 2-level full factorial needs 2^k runs; with k = 3 factors that is 2^3 = 8 runs to cover every combination. Multiplying or adding the factor and level counts (6 or 9) does not give the number of unique combinations.

The primary advantage of a designed experiment (DOE) over one-factor-at-a-time (OFAT) testing is that DOE:

• a.Can estimate interactions between factors and is more run-efficient✓
• b.Never needs any replication to give valid conclusions
• c.Requires changing only one single factor between each run
• d.Removes the need to define a measured response variable

DOE varies factors together in a structured plan, allowing it to detect interactions (where the effect of one factor depends on another) that OFAT cannot see, and it does so with fewer total runs. OFAT changes one factor at a time and misses interactions.

In DOE, an interaction effect occurs when:

• a.A factor turns out to have no effect at all on the response
• b.The effect of one factor on the response depends on the level of another factor✓
• c.Two different factors simply happen to be measured with the same gage or instrument
• d.Every run in the experiment happens to produce identical results

An interaction means the influence of one factor changes depending on the setting of another, shown by non-parallel lines on an interaction plot. Interactions are a key reason to run factorial experiments rather than isolated tests.

A fractional factorial design is chosen instead of a full factorial primarily to:

• a.Guarantee that no effects at all end up confounded with each other
• b.Avoid the need to analyze any interaction effects whatsoever
• c.Reduce the number of runs when there are many factors, accepting some confounding✓
• d.Deliberately increase the total number of experimental runs required

Fractional factorials run only a carefully chosen subset of combinations to screen many factors economically, at the cost of aliasing (confounding) some higher-order effects together. They trade some resolution for far fewer runs.

In a DOE, 'replication' means:

• a.Running the whole experiment through only one single time with no repeats at all being carried out
• b.Deleting any outlier runs that disagree with expectations
• c.Repeating the entire set of experimental runs to estimate pure experimental error✓
• d.Holding every factor constant throughout the experiment

Replication repeats the full set of runs under the same conditions so the team can estimate the natural experimental error and test effects against it. More replicates improve the precision and power of the analysis.

Randomizing the run order in a designed experiment protects mainly against:

• a.The requirement to include a separate control group
• b.Bias from unknown, time-related lurking variables like tool wear or warm-up✓
• c.The problem of having too few factors in the study
• d.Genuine interaction effects arising between two of the studied factors themselves

Randomization spreads the influence of uncontrolled, time-ordered nuisance variables across all treatment combinations so they do not systematically bias any one factor. It is a core DOE principle alongside replication and blocking.

Blocking in a designed experiment is used to:

• a.Guarantee that the experiment will produce a significant result
• b.Increase the total number of factors that can be studied at once in the design
• c.Remove the measured response variable from the analysis
• d.Account for a known nuisance source of variation (like batches or days) so factor effects stay clear✓

Blocking groups runs by a known nuisance variable, such as raw-material lot or shift, so its variation is separated out and the factor effects can be seen more clearly. It controls a recognized source of variability rather than adding factors.

During brainstorming for solutions, the most important ground rule is to:

• a.Immediately evaluate and criticize each idea the moment it is offered out loud
• b.Generate a large quantity of ideas first and defer judgment until later✓
• c.Strictly limit the entire session to no more than two ideas
• d.Allow only the team leader to contribute ideas to the list

Effective brainstorming separates idea generation from evaluation: the team aims for quantity and builds on ideas freely, then screens them afterward. Early criticism suppresses the creative flow that produces breakthrough solutions.

A Pugh matrix (solution selection matrix) helps the Improve team by:

• a.Calculating the defects-per-million-opportunities of the current process
• b.Testing statistically whether two group means are truly equal
• c.Scoring each candidate solution against a baseline using weighted criteria to pick the best✓
• d.Mapping the suppliers and inputs that feed into the process

A Pugh matrix compares alternative solutions against a reference (datum) across weighted selection criteria, producing a structured, objective ranking of the options. It supports the decision but is not a statistical or process-mapping tool.

Running a small-scale pilot of a solution before full rollout is valuable because it:

• a.Confirms the solution works and surfaces unexpected problems while risk and cost stay low✓
• b.Guarantees in advance that the solution is certain to fail
• c.Completely replaces the need to build any control plan later
• d.Removes the need to measure or track the results of the change at all afterward

A pilot tests the solution on a limited scale so the team can verify effectiveness, refine the approach, and catch side effects before committing to a costly full-scale implementation. It reduces the risk of deploying a flawed solution broadly.

Poka-yoke devices are classified into two main functions: control and warning. A 'control' poka-yoke:

• a.Sounds an alarm to alert the operator but lets the process keep running uninterrupted
• b.Increases the overall production rate of the line it is fitted to
• c.Physically prevents the error or stops the process until it is corrected✓
• d.Simply records each defect that occurs for later analysis and review

A control-type poka-yoke stops the process or makes the incorrect action impossible, for example a fixture that will not accept a part loaded backward. A warning-type only alerts the operator, relying on a human response.

A kaizen event (kaizen blitz) is best described as:

• a.A focused, short-duration team workshop to rapidly improve a specific process✓
• b.A year-long, slow-moving statistical study of a single process over time and all of its data
• c.A one-person audit of the department's financial records
• d.A permanent increase in the staffing of a work area

A kaizen event is an intense, typically 3-to-5-day cross-functional workshop that rapidly analyzes and improves a targeted process, delivering quick, tangible gains. 'Kaizen' means continuous improvement, and the event format concentrates that effort in a short burst.

The 5S step 'Set in order' (Seiton) focuses on:

• a.Removing all of the unneeded items from the immediate work area entirely at the very start of the day
• b.Auditing the area regularly to sustain the earlier gains
• c.Arranging needed items so they are easy to find, use, and return to their place✓
• d.Cleaning and scrubbing the workplace thoroughly and often

Set in order arranges the items kept after sorting into logical, labeled, easy-to-access locations so that anything can be found and returned quickly. Removing unneeded items is Sort, cleaning is Shine, and auditing is Sustain.

In a Lean pull system using kanban, production of a part is triggered by:

• a.A downstream signal indicating actual consumption or real demand✓
• b.The maximum output capacity currently available on the machine
• c.A monthly production forecast pushed down from central planning each month
• d.The shift supervisor's daily judgment about what to build next

A pull system makes parts only when a downstream kanban signals that stock has been consumed, tying production to real demand and limiting inventory. Building to a forecast or to available capacity is a push approach that tends to overproduce.

SMED (Single-Minute Exchange of Die) is a technique aimed at:

• a.Drastically reducing equipment changeover and setup time✓
• b.Eliminating the need to document any standard work
• c.Increasing the batch size run between each changeover on the line
• d.Adding several extra inspection steps to the process

SMED reduces changeover time (ideally to under ten minutes, 'single-minute') by converting internal setup steps to external ones done while the machine still runs. Shorter setups enable smaller batches and more flexible flow, the opposite of enlarging batches.

In SMED, converting 'internal' setup activities to 'external' ones means:

• a.Outsourcing the entire setup operation to an external supplier
• b.Doing all of the setup work only after the machine has been completely stopped
• c.Performing setup steps while the machine is still running, so less is done during downtime✓
• d.Eliminating the product changeover from the process altogether

Internal activities must be done while the equipment is stopped, whereas external activities can be prepared while it still runs; shifting work from internal to external shortens the actual downtime. This conversion is the core lever of SMED changeover reduction.

Total Productive Maintenance (TPM) aims to improve equipment effectiveness primarily by:

• a.Waiting until equipment actually fails and only then repairing it
• b.Running the machines well beyond their rated design capacity
• c.Engaging operators in routine care and preventing breakdowns before they occur✓
• d.Removing the maintenance staff to cut overhead costs

TPM builds proactive and preventive maintenance into daily operations, with operators performing autonomous upkeep to maximize uptime and reduce breakdowns. Run-to-failure ('reactive') maintenance is exactly what TPM seeks to eliminate.

Overall Equipment Effectiveness (OEE) is calculated as the product of:

• a.Severity, Occurrence, and Detection
• b.Cp, Cpk, and the process Sigma level
• c.Speed, Cost, and total Volume produced
• d.Availability, Performance, and Quality✓

OEE = Availability x Performance x Quality, combining uptime, speed, and first-pass quality into one measure of how effectively equipment is used. Severity x Occurrence x Detection is the FMEA RPN, an unrelated calculation.

A machine has Availability = 0.90, Performance = 0.95, and Quality = 0.98. Its OEE is approximately:

• a.0.90, taking only the availability factor by itself
• b.0.98, taking only the highest of the three factors
• c.2.83, from adding the three factors together
• d.0.84, from multiplying all three factors together✓

OEE = 0.90 x 0.95 x 0.98 = 0.8379, or about 84%. Adding the three factors (2.83) or taking any single one does not reflect the multiplicative nature of OEE.

Standard work (standardized work) documents the current best method so that:

• a.The process becomes fixed and can never be improved again
• b.Inspection of the finished product becomes entirely unnecessary
• c.Each operator is free to improvise their own steps and sequence however they wish
• d.The task is done the same safe, efficient way every time, forming the baseline for improvement✓

Standard work records the agreed best sequence, timing, and content of a task, ensuring consistency and providing the stable baseline against which future kaizen is measured. Without a standard there is nothing to improve upon reliably.

In a solution-selection matrix, weighting the criteria before scoring options ensures that:

• a.The cheapest available option automatically wins every time
• b.The most important criteria have proportionally more influence on the final decision✓
• c.No single option can ever score higher than any other one
• d.Every criterion in the selection matrix counts for exactly the same fixed amount every single time

Assigning weights reflects that some criteria (for example, safety or cost) matter more than others, so the weighted scores steer the choice toward what the organization values most. Unweighted scoring would treat trivial and critical criteria as equally important.

A response surface method (RSM) experiment is typically used when the team wants to:

• a.Screen a very large number of candidate factors as quickly as possible
• b.Simply count the number of defects produced per unit
• c.Organize brainstormed improvement ideas into related groups
• d.Optimize a response by modeling curvature and finding the best factor settings✓

RSM adds center and axial points to fit a curved (quadratic) model, allowing the team to locate the optimum settings of the vital few factors. Simple 2-level screening designs, by contrast, only estimate linear main effects and interactions.

Continuous flow (one-piece flow) in Lean seeks to:

• a.Inspect every finished unit at least twice before shipping
• b.Move products one unit at a time with minimal waiting or work-in-process✓
• c.Build large buffers of inventory between successive steps
• d.Maximize the size of each individual production batch that is run on the line

One-piece flow passes each unit directly to the next step as it is completed, slashing work-in-process, lead time, and the waste of waiting. Large batches and buffers are what continuous flow is designed to eliminate.

A spaghetti diagram supports Improve by:

• a.Displaying the Risk Priority Number of each failure mode
• b.Calculating the process capability index of the operation
• c.Statistically testing whether the variances of two different samples are equal
• d.Mapping the physical movement of people or material to expose motion and transport waste✓

A spaghetti diagram traces the actual travel paths of an operator or product across a layout, making motion and transportation waste visible so the layout can be improved. It is a Lean visualization tool, not a statistical calculation.

Heijunka (production leveling) improves flow by:

• a.Smoothing the type and quantity of production over a period to reduce peaks and troughs✓
• b.Increasing the changeover time required between product types
• c.Producing the entire month's demand together in one very large batch
• d.Eliminating the need for any documented standard work at all

Heijunka levels the production schedule by volume and mix so the workload is even, which lowers inventory, stabilizes demand on upstream processes, and supports pull. Large, uneven batches create the very peaks that leveling removes.

A future-state value stream map, built during Improve, is used to:

• a.Rank the competing suppliers strictly by their unit cost
• b.List out the project's income statement and financial reports
• c.Design the improved process flow the team intends to implement, with less waste✓
• d.Record only the current wasteful process exactly as it exists today

The future-state map depicts the target design after waste is removed, showing where flow, pull, and leveling will be applied as the blueprint for implementation. The current-state map, in contrast, captures today's process as it exists.

A pilot study shows the new method reduces cycle time, but only under ideal staffing. Before full rollout the team should:

• a.Abandon the solution entirely and return to the old method
• b.Roll the solution out immediately regardless of the staffing concern that was raised
• c.Increase the production batch size to compensate for staffing
• d.Address the staffing dependency and validate under realistic conditions✓

A pilot's job is to reveal conditions that could undermine success; a dependency on ideal staffing must be resolved and the solution re-validated under normal conditions before scaling. Rolling out unverified assumptions risks failure at full scale.

In DOE, a 'center point' run (all factors set at their mid-level) is added mainly to:

• a.Replace and remove the need for run-order randomization
• b.Increase the number of factors being studied in the design
• c.Remove all interaction terms from the fitted model
• d.Detect curvature (nonlinearity) in the response between the levels✓

Center points let the team check whether the response bends between the low and high settings; a significant difference from the factorial average signals curvature that a linear model would miss. They also provide an estimate of pure error.

A 2^4 full factorial design (4 factors at 2 levels) requires how many runs for a single replicate?

• a.24, from multiplying the number of factors by six
• b.4, from counting one run for each factor in the design that is performed just once
• c.8, from multiplying the 4 factors by the 2 levels
• d.16, from two levels raised to the power of four factors✓

The number of runs in a 2-level full factorial is 2^k = 2^4 = 16 for four factors. Multiplying 4 x 2 (8) or 4 x 6 (24) does not represent the count of unique factor-level combinations.

The main risk of implementing a solution without piloting it first is that:

• a.Unforeseen problems appear at full scale where they are costly and hard to reverse✓
• b.The team unexpectedly ends up saving far too much money on the whole project budget somehow
• c.The verified root cause of the problem is automatically fixed
• d.The control plan for the process becomes entirely unnecessary

Skipping the pilot means any hidden flaws, side effects, or false assumptions surface only after full deployment, when correcting them is expensive and disruptive. A pilot contains that risk to a small, reversible scale.

The 5S step 'Sustain' (Shitsuke) is concerned with:

• a.Arranging the tools neatly on a labeled shadow board
• b.Building the discipline, audits, and habits that keep the other 4S in place over time✓
• c.Scrubbing and cleaning the floors and equipment every day
• d.Throwing away tools and parts that are broken or that are no longer needed

Sustain embeds the gains through training, standard audits, and management support so the workplace does not slip back to disorder. Discarding items is Sort, cleaning is Shine, and shadow boards support Set in order.

A 'gemba walk' during the Improve phase means:

• a.Running a computer simulation of the process without observing it
• b.Going to the actual place where work is done to observe the process firsthand✓
• c.Reviewing financial spreadsheets from within a closed conference room
• d.Interviewing the end customers about their needs by telephone

'Gemba' means 'the real place'; a gemba walk sends leaders and the team to observe the actual work where value is created, gathering direct facts rather than relying on reports. Direct observation often reveals waste and improvement ideas that data alone miss.

A cost-benefit analysis in the Improve phase supports solution selection by:

• a.Mapping the full value stream from supplier to customer
• b.Testing statistically whether two defect proportions are equal
• c.Comparing the expected savings or gains of each solution against its implementation cost✓
• d.Measuring the long-term process sigma level of the operation being studied

Cost-benefit analysis weighs each solution's projected benefits against the resources required to implement it, helping the team choose options with the strongest return. It informs the decision but does not compute sigma or test proportions.

A force field analysis helps the Improve team by:

• a.Ranking the observed defects from highest to lowest cost
• b.Calculating the takt time that sets the required pace of the production line
• c.Identifying the driving forces supporting a change and the restraining forces resisting it✓
• d.Estimating the variance of the underlying process population

Force field analysis lists the forces pushing for and against a proposed change so the team can strengthen drivers and reduce barriers to improve adoption. It is a change-management planning tool, not a statistical or Lean-timing calculation.

In DOE terminology, a 'factor' is:

• a.The random experimental error left in the collected results
• b.An input variable deliberately changed to study its effect on the response✓
• c.The total number of replicates performed in the design
• d.The measured outcome or final result that is recorded from each experiment

A factor is a controllable input (an X) the experimenter sets at defined levels to see how it influences the response (the Y). The measured outcome is the response, not a factor.

The 'levels' of a factor in a designed experiment are:

• a.The number of separate response outcomes that are measured in each single run of the design matrix used
• b.The randomized order in which the runs are performed
• c.The team members assigned to carry out the experiment
• d.The specific settings or values at which the factor is tested, such as low and high✓

Levels are the discrete settings a factor takes during the experiment, such as low and high temperature, that let the team compare the response across conditions. The choice of levels defines the range over which effects are estimated.