Overview
Control makes the improvement permanent. Statistical process control (SPC) covers the objectives of SPC, selecting variables to chart, rational subgrouping, choosing the right chart (X̄-R, X̄-s, individuals and moving range, p, np, c, u, and special charts such as EWMA, CUSUM and short-run charts), calculating limits, and interpreting charts with run rules to separate common-cause from special-cause variation. Other controls covers total productive maintenance (TPM) and visual controls. Maintain controls covers re-analyzing the measurement system and building a control plan that states what is measured, how often, by whom, and what to do when a signal appears. Sustain improvements covers lessons learned, documentation, training the process owners and operators, and ongoing evaluation.
Candidates must select and compute charts, read signals correctly and design controls that survive the Black Belt's departure. Control finishes the work of Domain 7 and returns ownership to the process owner described in Domain 1.
How it's tested
Expect chart-selection items, control-limit calculations, interpretation of chart patterns against run rules, and scenario questions about control plans, handoff and sustaining gains. The correct answer usually responds to special causes only, separates control limits from specifications, and places ownership with the process owner.
-
Pick the chart
A data situation is described by data type, subgroup size and whether you count defects or defectives, and you choose the chart.
-
Compute or read the limits
Summary statistics and constants are given and you compute control limits, or a pattern is described and you decide whether it signals a special cause.
-
Sustain the gain
A project is closing or gains are eroding, and you choose the control plan element, documentation or ownership step that prevents backsliding.
An operator sees one point just inside the upper control limit and wants to adjust the machine. All other points are random around the centerline and no run rule is violated. A point inside the limits with no pattern is common-cause variation, and adjusting a stable process (tampering) increases variation. The correct response is to leave the process alone and keep monitoring.
Subsections 4
Official subsections from the ASQ outline, in outline order.
8.1 Statistical process control (SPC)Objectives, variable selection, rational subgrouping, chart selection, limits and interpretation.
- What it means
- SPC uses control charts to monitor a process over time and distinguish common-cause variation, inherent to a stable process, from special-cause variation that signals an assignable change. It covers choosing which characteristics to chart, forming rational subgroups, selecting the right variables or attributes chart, computing limits, and applying run rules.
- What to understand
- Rational subgroups are formed so that variation within a subgroup is only common cause, making between-subgroup shifts visible. Control limits are calculated from process data, typically at ±3 standard errors, and are unrelated to specification limits. Variables charts (X̄-R, X̄-s, I-MR) suit measurements; attributes charts suit counts: p and np for defectives, c and u for defects. EWMA and CUSUM detect small sustained shifts faster than Shewhart charts.
- How it shows up
- A data situation is described and you select the chart, compute limits, or decide whether a pattern violates a run rule and what action to take.
- Common confusion
- Placing specification limits on an X̄ chart, or treating a point outside specification as the same thing as a point outside control limits.
8.2 Other controlsTotal productive maintenance and visual controls.
- What it means
- This covers controls that are not charts: total productive maintenance (TPM), which keeps equipment reliable through autonomous maintenance by operators, planned maintenance and OEE tracking, and visual controls such as andon lights, shadow boards, floor markings, kanban boards and visual work instructions that make abnormal conditions obvious.
- What to understand
- TPM targets the major equipment losses (breakdowns, setup, minor stops, reduced speed, defects and startup losses) and shifts routine care to operators. Visual controls let anyone see at a glance whether the process is in its standard condition, which speeds reaction and reduces reliance on memory.
- How it shows up
- A scenario describes recurring equipment problems or hard-to-see deviations and you choose TPM or a visual control as the sustaining measure.
- Common confusion
- Treating TPM as maintenance department work only, when autonomous operator maintenance is central to it.
8.3 Maintain controlsMeasurement system re-analysis and the control plan.
- What it means
- This covers confirming the measurement system remains adequate as the process improves, since a gage acceptable for the old variation may be too coarse for the new, and building a control plan that lists each controlled characteristic, specification, measurement method, sample size and frequency, control method, owner and reaction plan.
- What to understand
- When process variation falls, gage variation becomes a larger share of the total, so %GRR measured against process variation can rise and must be re-checked. The reaction plan is the most important control plan element: it tells the operator exactly what to do, including containment, when a signal or nonconformance occurs.
- How it shows up
- A project is closing and you identify the missing control plan element, or improved data look noisier than expected and you choose to re-analyze the measurement system.
- Common confusion
- Assuming the original gage study remains valid after variation has been substantially reduced.
8.4 Sustain improvementsLessons learned, documentation, training and ongoing evaluation.
- What it means
- This covers capturing lessons learned, updating procedures, work instructions and FMEAs, training operators and process owners on the new methods, transferring ownership formally, replicating the solution elsewhere, and auditing results over time to confirm the gain holds.
- What to understand
- A project closes only when the process owner accepts the control plan and can run it independently, results are verified over a sustained period, and financial benefits are validated. Replication to similar processes multiplies the value of the project.
- How it shows up
- Gains erode after closure, or a project is ready to close, and you choose the step that sustains or completes the improvement.
- Common confusion
- Treating project closure as the moment results first improve rather than after sustained, owned performance.
High-yield concepts
The terms, rules and formulas this domain turns into questions most often.
-
Control chart selection
Measurements: X̄-R for subgroups of about 2 to 9, X̄-s for larger subgroups, I-MR for individual values. Counts: np (number defective, constant n), p (fraction defective, n may vary), c (defects per unit, constant area), u (defects per unit, varying area).
- Remember
- Defectives (units that fail) use p or np; defects (flaws, several possible per unit) use c or u.
- Tested as
- The most common Control question type.
- Trap
- Using a p chart for scratches per panel, which are defects and need a c or u chart.
-
X̄-R chart limits
X̄ chart: UCL/LCL = X̿ ± A₂R̄. R chart: UCL = D₄R̄, LCL = D₃R̄. For n = 5, A₂ = 0.577, D₃ = 0, D₄ = 2.114.
- Remember
- Read the R chart first; if ranges are out of control, X̄ limits are unreliable.
- Tested as
- Limit calculations from summary data and constants tables.
- Trap
- Using the overall standard deviation of all data instead of R̄ to set limits.
-
Individuals and moving range limits
I chart: X̄ ± 2.66 × MR̄; MR chart: UCL = 3.267 × MR̄, where MR̄ is the average moving range of consecutive points.
- Remember
- I-MR suits slow, expensive or one-at-a-time data such as batch chemistry.
- Tested as
- Chart selection and calculation for individual data.
- Trap
- Applying I-MR to strongly non-normal data without considering transformation.
-
p and np chart limits
p chart: p̄ ± 3√[p̄(1 − p̄) / n]; np chart: np̄ ± 3√[np̄(1 − p̄)]; negative LCLs are set to zero.
- Remember
- With varying n, p chart limits vary by subgroup.
- Tested as
- Attribute-limit calculations.
- Trap
- Using an np chart when sample sizes vary.
-
c and u chart limits
c chart: c̄ ± 3√c̄; u chart: ū ± 3√(ū / n), where n is the number of inspection units.
- Remember
- These charts follow the Poisson distribution, so variance equals the mean.
- Tested as
- Defect-count chart calculations.
- Trap
- Forgetting the square root, or using c when inspection area varies.
-
Run rules (Western Electric)
Common signals: one point beyond 3σ; two of three consecutive points beyond 2σ on the same side; four of five beyond 1σ on the same side; eight consecutive points on one side of the centerline. Nelson rules add trends, such as six points steadily increasing or decreasing.
- Remember
- Run rules increase sensitivity but also the false-alarm rate when many are applied.
- Tested as
- Pattern-interpretation items.
- Trap
- Reacting to a single point inside the limits with no pattern.
-
Rational subgrouping
Form subgroups from units produced under nearly identical conditions, close together in time, so within-subgroup variation reflects common cause only.
- Remember
- Mixing shifts or machines within a subgroup inflates R̄ and widens limits, hiding special causes.
- Tested as
- Asked about chart design and why a chart fails to detect known shifts.
- Trap
- Sampling across several machines into one subgroup for convenience.
-
Control limits vs. specification limits
Control limits are the voice of the process, calculated from data; specification limits are the voice of the customer, set by requirements.
- Remember
- A process can be in control and still produce out-of-spec output, or out of control while within spec.
- Tested as
- Concept and interpretation items.
- Trap
- Drawing specification limits on an X̄ chart, where averages hide individual values outside specification.
-
EWMA and CUSUM
EWMA charts a weighted average of current and past values; CUSUM charts the cumulative sum of deviations from target. Both detect small sustained shifts (around 0.5 to 1.5σ) faster than Shewhart charts.
- Remember
- Shewhart charts are best for large, sudden shifts.
- Tested as
- Chart selection when the concern is small drifts.
- Trap
- Choosing an X̄ chart to detect a small gradual shift.
-
Tampering (over-adjustment)
Adjusting a stable process in response to common-cause variation; Deming's funnel experiment shows it increases variation.
- Remember
- Act on special causes; reduce common-cause variation only by changing the system.
- Tested as
- Scenario items where an operator wants to adjust after a random point.
- Trap
- Viewing every adjustment as responsible process management.
-
Control plan
A document listing each controlled characteristic, specification, measurement method, sample size and frequency, control method, responsible person and reaction plan.
- Remember
- The reaction plan specifies containment and correction when a signal or nonconformance occurs.
- Tested as
- Asked which element is missing or most important at handoff.
- Trap
- Choosing a control plan with charts but no reaction plan.
-
Total productive maintenance (TPM)
A program to maximize equipment effectiveness through autonomous maintenance by operators, planned maintenance, focused improvement and OEE measurement.
- Remember
- TPM targets breakdowns, setup losses, minor stops, speed losses, defects and startup losses.
- Tested as
- Chosen when equipment reliability threatens the improved process.
- Trap
- Assigning all TPM activity to the maintenance department.
Side by side
Concepts candidates mix up, compared on the point that decides the answer.
p vs. np vs. c vs. u charts
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| p chart | Fraction defective per subgroup | Defectives; sample size can vary | Used for defect counts |
| np chart | Number defective per subgroup | Defectives; constant sample size | Used when n varies |
| c chart | Number of defects per inspection unit | Defects; constant area of opportunity | Used when unit size varies |
| u chart | Defects per unit | Defects; area of opportunity can vary | Confused with p chart |
X̄-R vs. X̄-s vs. I-MR vs. EWMA
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| X̄-R | Subgroup means and ranges | Small subgroups, about 2 to 9 | Used with large subgroups where range is inefficient |
| X̄-s | Subgroup means and standard deviations | Subgroups of about 10 or more | Overlooked when n is large |
| I-MR | Individual values and moving ranges | One measurement per period | Forced subgrouping of slow data |
| EWMA | Exponentially weighted moving average | Detects small sustained shifts | Expected to catch a single large spike quickly |
Common cause vs. special cause
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| Common cause | Inherent, random variation of a stable system | Reduced only by changing the process | Treated with operator adjustments (tampering) |
| Special cause | Assignable change such as a new lot or worn tool | Shown by points beyond limits or run-rule patterns | Ignored because output is still within specification |
Exam traps
Each trap is an answer that looks right. Learn the clue that should change your mind.
-
Defects on a p chart
- What candidates pick
- A p chart to track the number of blemishes per car door.
- Why it looks right
- p charts are the best-known attributes chart.
- Why it's wrong
- Blemishes are defects, and one door can have several; defect counts follow Poisson and need a c chart (constant area) or u chart (varying area).
Clue to watch for The data count flaws per unit rather than units that fail.
-
Spec limits as control limits
- What candidates pick
- Draw the customer's specification limits on the X̄ chart.
- Why it looks right
- Specifications are what the customer cares about.
- Why it's wrong
- Control limits come from process data; specifications on an averages chart are misleading because averages vary less than individuals.
Clue to watch for The question asks how to set or interpret control limits.
-
Tampering
- What candidates pick
- Adjust the process after a single point near the upper limit.
- Why it looks right
- It feels proactive and responsible.
- Why it's wrong
- Without a run-rule violation the point is common-cause variation; adjusting increases variation.
Clue to watch for No points are beyond limits and no pattern is described.
-
Original gage still fine
- What candidates pick
- Keep using the original gage study after variation fell by half.
- Why it looks right
- The gage passed MSA at the start of the project.
- Why it's wrong
- Gage variation is now a larger share of the smaller process variation; the measurement system must be re-analyzed.
Clue to watch for Improved process data look noisy or %GRR is compared with the new process spread.
-
Close on first good month
- What candidates pick
- Close the project as soon as one month of data meets the goal.
- Why it looks right
- The goal has been achieved.
- Why it's wrong
- Closure requires sustained, stable performance, an accepted control plan, trained owners and validated savings.
Clue to watch for Handoff, documentation or training has not been confirmed.
-
Charts without reaction
- What candidates pick
- A control plan with charts and sampling frequency but no instructions for out-of-control signals.
- Why it looks right
- Monitoring seems to be the essence of control.
- Why it's wrong
- A signal without a defined reaction plan leads to inconsistent responses and lost gains.
Clue to watch for The question asks which control plan element is missing or most important.
Answer framework
Use this sequence for chart selection, interpretation and sustainment questions.
- Classify the data as measurement or count, and for counts decide between defectives and defects, then note subgroup size.
- Choose the chart: X̄-R, X̄-s or I-MR for measurements, p or np for defectives, c or u for defects, and EWMA or CUSUM for small shifts.
- Compute limits from process data with the correct constants or formula, set negative lower limits to zero, and never substitute specification limits.
- Interpret using run rules: act on special causes with the reaction plan and leave common-cause variation alone to avoid tampering.
- Before closing, confirm the measurement system, document the control plan with reaction steps, train the owners and verify sustained results.
Practice question
Want more like this? The CSSBB practice test has questions for every domain.
Checklist
0 of 14
Tick each one you can do without notes. Your ticks stay in this browser.
Study strategy
Begin with chart selection, because it drives most Control questions. Make a decision tree that branches on measurement versus count, defects versus defectives, and subgroup size, ending at each chart. Drill it with scenarios until you can choose in seconds.
Next, practice limit calculations: X̄-R with a constants table, I-MR with 2.66 and 3.267, and the attributes formulas for p, np, c and u. Work several by hand, including cases where the lower limit is negative and set to zero. Then practice interpreting patterns with Western Electric rules and the difference between special causes and tampering.
The commonly underestimated area is sustainment: control plans, measurement re-analysis, TPM, visual controls and handoff. These are scenario questions about what is missing or what to do next, and the expected answer builds control into the process and places ownership with the process owner.
Tab a constants table and chart-selection tree in your open-book reference. Afterwards, practice mixed Control sets, and revisit Measure, since rational subgrouping, MSA and capability all reappear here.
FAQ
How many questions come from Control?
ASQ assigns 15 of the 150 scored questions to Control, about 10% of the exam.
How do I tell defects from defectives?
A defective is a unit that fails; a defect is a flaw, and one unit can have several. Defectives use p or np charts; defects use c or u charts.
Should specification limits ever appear on a control chart?
Not on an X̄ chart. Control limits come from process data. Specifications belong on histograms or capability analyses, where individual values are compared with them.
When should I use EWMA or CUSUM instead of a Shewhart chart?
When the concern is a small, sustained shift of roughly half to one and a half standard deviations, which Shewhart charts detect slowly.
What is the most important part of a control plan?
The reaction plan, because it tells operators exactly what to do, including containment, when a signal or nonconformance appears.
When is a six sigma project finished?
When results are sustained over time, the control plan is accepted and run by the process owner, documentation and training are complete, and financial benefits are validated.
How does Control connect to Measure?
Control reuses rational subgrouping, measurement system analysis and capability from Measure to confirm and monitor the improved process.
Other Quality Management certifications
Where the same subject shows up in other exams. Cards marked "Same subject" open the matching domain.
Test your Domain 8 knowledge
You have covered SPC chart selection and limits, run rules, TPM and visual controls, control plans and sustaining gains. Test whether you can choose and read charts accurately under time pressure.