Read a lot acceptance curve without certifying the accepted lot

Define the decision before reviewing its behavior

NIST’s single-sample rule accepts when the nonconforming count is at most c in n inspected items. Its binomial approximation assumes random sampling from a much larger lot.

For the fictional brochure exercise, fix n = 50 and c = 2. Count each inspected brochure once as conforming or nonconforming under a previously specified print criterion. This is an invented teaching plan, not a recommendation for a purchase.

Write the actual inspection question in ordinary production language. An unreadable required line, a missing panel and an unexpected paper finish may belong to different decisions. Have the responsible quality owner define the criterion before anyone sees the inspection result. A coordinator should not change that criterion to make a late delivery easier to approve.

Identify the delivered lot, its boundaries and the person responsible for selecting inspection items. Preserve how the sample was obtained. Selecting the neatest cartons or asking a supplier to submit its preferred examples does not provide the selection evidence required by the stated exercise. This worksheet cannot recover a missing sampling record by adding a polished chart afterward.

Compute the stated rule rather than a favorable substitute

The operating characteristic is P(accept | stipulated fraction nonconforming p), not P(good lot | accepted). For this plan, acceptance probabilities at p = 0.01, 0.05 and 0.10 are approximately 0.986183, 0.540533 and 0.111729.

The executed calculation uses SciPy 1.17 binom.cdf(2, 50, p), checked against a separate sum of the binomial terms for counts zero through two. The three p values are teaching assumptions, not estimates from the sample.

Place the plan inputs beside the outputs rather than sending a screenshot containing only a curve. Record the software version and the exact function call so the technical owner can check what was calculated. Keep the invented values visibly separate from any real supplier record. A number copied into a production slide should not lose the label that explains where it came from.

Require an independent arithmetic route for this compact example. A second screenshot of the same application is not an independent calculation. If the two routes disagree, pause the interpretation and inspect the input convention and returned quantity before discussing whether the result looks plausible. Do not select the more reassuring output or average incompatible calculations.

Name whose risk is being described

Using 1% and 10% as illustrative acceptable and unacceptable quality points, the producer’s rejection risk at the former is about 1.3817%; the consumer’s acceptance risk at the latter is about 11.1729%. These are different conditional operating risks, not guarantees about one accepted lot.

Ask the commercial and quality owners whether the proposed teaching points resemble the actual decision they need to make. Do not promote them into a standard because the example produces convenient percentages. Their selection belongs with the approved use, consequence of nonconformance and governing agreement, not with the person preparing the chart.

Keep serious unresolved findings visible outside the numerical summary. This exercise does not authorize accepting a known consequential problem simply because a sample count satisfies a generic rule. The relevant owner must decide which procedure applies and whether the stated assumptions describe the actual inspection. A favorable calculation cannot grant authority that the reviewer does not hold.

Use the interpretation sheet to expose the decision that still needs ownership. Who selects the plan, who conducts the inspection, and who can approve the final disposition? Those roles may be different. Avoid wording that makes the analyst’s calculation appear to constitute a supplier acceptance notice or a promise to pay.

Close the evidence gap before the operational decision

Prepare a short handoff containing the criterion, lot identification, selection record, stated plan, reproducible calculation and unanswered production questions. Attach the underlying observation record when an actual authorized review occurs. The arithmetic shown here supplies no observations about a delivered print run, so the fictional result must remain separate from real acceptance paperwork.

Have the technical reviewer confirm the scope of the model and the authorized quality owner decide whether it is appropriate. If the actual lot or inspection method falls outside the stated conditions, do not silently retain the result and soften the wording. Record the mismatch and request the applicable analysis before using the chart in a consequential decision.

Review the final sentence with the same care as the calculation. It should explain the checked procedure and its limits without implying that every unit was examined or that a contractual obligation has been discharged. Preserve the unresolved question rather than substituting a broad quality badge for the evidence that a reader would need.

Retain the worksheet when a later delivery uses a different plan. Compare the approved inputs and decision context before reusing any old graphic. A previous calculation may be reproducible yet answer the wrong current question. This lesson finishes with an interpretation record; it neither accepts a delivery nor changes a supplier agreement.

Make the reasoning concrete.

A fictional brochure coordinator receives a proposal to inspect a small selection and wants to put a quality-approved badge on the whole delivery. No actual inspection observations are available in this exercise.

The coordinator withholds the unsupported badge and routes the specific plan questions to the responsible owner. No supplier communication or acceptance action is taken.

Take the questions into your next brief.

Use these prompts in your own notes. Nothing is submitted from this page.

1. Criterion

   What approved observation makes an item nonconforming?

2. Selection

   How was the inspected set selected from the identified lot?

3. Calculation

   Which exact inputs and independent arithmetic route support the output?

4. Interpretation

   What does the reported probability condition on?

5. Disposition

   Who decides the actual delivery outcome and what evidence is missing?

Watch for the wrong conclusion.

• Choosing the inspection criterion after viewing the result.
• Replacing a missing selection record with an attractive chart.
• Treating an analytical worksheet as commercial acceptance authority.

Sources & scope

Original fictional print-review workflow and numerical fixture. Named sources support the statistical distinctions. Local arithmetic only was executed; no actual goods were inspected, no supplier quality was established and no acceptance action is authorized.

• NIST: Single sampling plans and their operating characteristics →Primary statistical engineering handbook · Supports: Supports the conditional operating characteristic and producer/consumer risk distinction.
• SciPy: scipy.stats.binom →Primary scientific software documentation · Supports: Supports the cumulative binomial calculation.