EXIN LSSA Green Belt Practical common mistakes
EXIN LSSA Green Belt Practical · Understand the traps and how to avoid them.
Treating the DMAIC phase order as a checklist to name in sequence, then answering a scenario question by jumping straight from Define to Improve without stating Measure and Analyse evidence.
Candidates memorise the acronym as a list of words rather than as a gated sequence with entry and exit criteria, so under time pressure they skip the phases that carry the data-based justification.
What to do instead
Before answering any scenario item, write the five phase names in order and next to each note the deliverable that closes it (charter, validated measurement system and baseline, verified root causes, piloted countermeasures, control plan). Reject any option whose action belongs to a later gate than the evidence described.
Confusing a CTQ (critical to quality) characteristic with an internal specification limit, and labelling an internal engineering tolerance as the voice of the customer.
Training material often presents CTQs and specification limits side by side, so candidates conflate the customer-derived requirement with the numerically translated tolerance.
What to do instead
State the rule explicitly when answering: the customer states a need in their own language; the CTQ is that need translated into a measurable characteristic; the specification limit is the numeric boundary derived from it. If an option quotes a number with no traceable customer need behind it, it is a specification, not a CTQ.
Selecting a project that has no measurable Y, no baseline, or no clear process owner, because the scenario sounds like a good improvement idea.
Candidates optimise for whether the described change sounds beneficial rather than testing the project against selection criteria such as measurable output, baseline data availability, and sponsor ownership.
What to do instead
Screen every candidate project against four tests: is there a measurable Y, is there existing baseline data, is the scope completable in a Green Belt timeframe, and is there a named process owner and sponsor. An option failing any test is the wrong answer regardless of how appealing the improvement sounds.
Describing a measurement system study as 'checking the gauge is calibrated' and omitting repeatability and reproducibility as separate components.
Calibration is the most familiar quality term, so candidates substitute it for the distinction between variation from the gauge itself and variation from operators using it.
What to do instead
When a question involves measurement variation, name both components: repeatability is equipment variation when one operator measures the same item repeatedly; reproducibility is operator or appraiser variation. State that a gauge can be calibrated and still fail a Gage R&R. Answer with the component the scenario describes, not the generic word 'calibration'.
Reporting an overall capability figure (Pp/Ppk) as if it were the short-term capability index (Cp/Cpk), or quoting a capability index for a process that has never been shown to be in statistical control.
Software outputs both sets of indices and candidates pick whichever number appears in the scenario without checking which variation source it represents or whether a control chart was established first.
What to do instead
Apply a fixed order: confirm stability on a control chart, confirm the measurement system is adequate, then compute capability. Use Cp/Cpk for within-subgroup short-term variation and Pp/Ppk for overall long-term variation. If the scenario shows an unstable chart, the correct answer is to investigate the instability, not to report a capability number.
Reading a control chart as a specification-compliance chart, concluding the process is fine because every point sits inside the customer's tolerance limits.
Control limits and specification limits look similar on a printed chart, and candidates do not distinguish limits computed from the process's own variation from limits set by the customer.
What to do instead
Before interpreting any chart, label the limits: control limits come from the process data (typically mean plus or minus three sigma of the plotted statistic); specification limits come from the customer. Then apply the out-of-control rules (a point beyond a control limit, a run of points on one side of the centre line, trends, and non-random patterns). A process can be in control and still incapable, or capable and out of control.
Proposing a countermeasure as soon as a problem is described, without evidence that the suspected cause actually drives the Y.
Practical experience makes the familiar cause feel obviously correct, and candidates skip the verification step that the Improve phase depends on.
What to do instead
Force the causal chain into the answer: state the suspected X, the data that shows X and Y move together, and the mechanism. If the scenario supplies no verification data, the correct next action is to verify the cause, for example with a stratified analysis, hypothesis test, or designed experiment, before piloting a change.
Choosing a hypothesis test by the familiarity of the test name rather than by data type, number of groups, and pairing structure, for example running a two-sample t-test on paired before-and-after readings.
Candidates memorise test names without the selection logic, and paired data from the same units looks like two independent samples.
What to do instead
Answer in three steps: identify the data type of the response (continuous or attribute), count the groups or factor levels, and check whether observations are paired or independent. Continuous, two independent groups gives a two-sample t-test; continuous, paired gives a paired t-test; three or more group means gives ANOVA; attribute proportions across categories gives a chi-square test. State the selection logic in the answer.
Interpreting a p-value as the probability that the null hypothesis is true, or as the probability the result was due to chance, and then declaring practical significance from a small p-value alone.
The p-value definition is counterintuitive and is commonly misstated in informal workplace discussion, so the wrong phrasing feels natural.
What to do instead
Use the correct sentence: the p-value is the probability of observing data at least as extreme as obtained, assuming the null hypothesis is true. Compare it to the pre-chosen alpha to decide whether to reject the null. Then separately assess practical significance using effect size and confidence intervals, and note that a large sample can make a trivial difference statistically significant.
Writing a control plan that lists only the metric and its target, omitting the monitoring method, frequency, responsibility, reaction plan, and the standardised work or mistake-proofing that sustains the gain.
Control is treated as an administrative closing document rather than as the mechanism that keeps the improved process from reverting, so candidates produce a target list.
What to do instead
Build the control plan as a table with, for each critical X and Y: the measure, the monitoring method and frequency, the owner, the control limits or target, and the specific reaction when it goes out of limits. Add the document and training updates and any error-proofing device. If the scenario shows gains decaying after the project, the correct answer is a missing or incomplete control plan, not a new Improve cycle.