AI Glossary / Statistics and probability / Intermediate
Variables sampling plan
Definition
A variables sampling plan is an acceptance-sampling procedure that uses quantitative measurements from sampled items to decide whether to accept a lot. It uses information such as a sample mean and standard deviation rather than only counting conforming and nonconforming items. The plan specifies sample size, acceptance criteria and relevant distributional assumptions. Variables refers to measured characteristics, not to changing the sample fraction or selecting variables for a machine-learning model.
What it means
A measurement tells more than whether an item passed a limit: it indicates how far the item lies from that limit. A variables plan uses this additional information to assess lot quality. This can reduce the required sample size for some risk targets compared with an attributes plan, provided its assumptions are appropriate. It does not mean the accepted lot contains no nonconforming items. Acceptance is a decision made under a stated sampling rule with nonzero producer and consumer risks.
How it works
Define the quality characteristic, specification limits, lot and sampling procedure. Choose a plan whose sample size and acceptance constant provide the intended operating-characteristic behavior under the assumed model. For a simple one-sided upper-limit rule with unknown standard deviation, an illustrative form accepts when (U minus sample mean)/sample standard deviation is at least k. Other plans use different formulas, known variability, two-sided specifications or nonnormal models. The plan must match the measurement distribution and inspection context. Measurement-system quality and representative sampling are essential to meaningful performance.
Three practical examples
Example 1: A manufacturer measures component diameter and uses its distance from an upper specification limit to assess a lot. Example 2: A laboratory evaluates a quantitative material property against a lower limit. Example 3: A service-quality audit considers a measurement-based acceptance rule for a batch of deliverables, after checking whether the model assumptions fit. These examples concern lot decisions; they do not replace ongoing process monitoring or establish that a process remains stable over time.
Think of it this way
Think of judging a batch by both how close its measurements are to a boundary and how much they vary. Two batches can have the same average while one is much more spread out and therefore more likely to cross a specification limit. A variables plan uses that measured distance and spread rather than reducing every observation immediately to pass or fail.
Worked example
For a purely illustrative upper-limit rule, let the specification limit U be ten units and the acceptance constant k be 1.5. Suppose a prespecified sample of 25 items has mean 9.2 and sample standard deviation 0.4. The standardized distance is (10 minus 9.2)/0.4 = 2.0, which exceeds 1.5, so this rule accepts the lot. Equivalently, 9.2 + 1.5 times 0.4 = 9.8 is below the limit. If another sample has mean 9.5 with the same standard deviation, its distance is 1.25 and its upper expression is 10.1, so the rule rejects. The chosen k is illustrative, not a certified plan or a claimed protection level. Establishing producer and consumer risks requires calibrating the full plan and its assumptions.
Illustrative case study
In this fictional case study, Cedar Components considers replacing a pass/fail inspection with measured-diameter sampling. Analysts check the gauge system, lot construction and distributional shape before selecting a plan. They compare operating-characteristic curves rather than choosing a small sample size merely because measurements are available. Staff retain the exact rule and distinguish an accepted lot from a perfectly conforming lot. The company also continues process monitoring because lot acceptance and process stability answer different questions.
Common misconceptions and mistakes
Common mistakes include assuming every quantitative dataset is suitable for a normal-based plan, choosing an acceptance constant without risk calibration, and interpreting acceptance as proof of zero defects. Another mistake is using specification limits as though they were control limits. Specifications describe requirements; control limits describe expected process behavior under a monitoring model. Heavy tails, mixtures, dependence or measurement error can change a plan's actual acceptance probabilities and should not be ignored simply because the sample mean looks acceptable.
Business use
Businesses can use variables plans when reliable measurements are available and their additional information justifies the inspection cost. Select a plan appropriate to the required quality levels and risks, and document its assumptions and lot-sampling procedure. Check whether variability is known or estimated and whether one or both specification limits matter. Use an applicable validated standard or a properly designed custom plan rather than copying an illustrative formula. Reassess performance when the process, measurement system or product mix changes.
People also ask
Is variables sampling the same as inspecting a variable number of items?
No.
Does it use quantitative measurements?
Yes.
Must every variables plan assume normality?
No, but many common plans do, and each plan has its own assumptions.
Does acceptance guarantee every item conforms?
No.
Can it sometimes use fewer observations than an attributes plan?
Yes, for comparable risks under appropriate conditions.
Check your understanding
What is the first sample's standardized distance from the upper limit?
Answer: Two.
Does the illustrative rule accept it?
Answer: Yes, because two exceeds 1.5.
Why is the second sample rejected?
Answer: Its distance is only 1.25.
Can a consumer-risk level be inferred from these arithmetic checks alone?
Answer: No; the complete plan must be calibrated under a specified model.
Related terms
- Attribute sampling
- Specification limits
- Related terms include acceptance sampling
- attributes sampling
- operating characteristic curve
- specification limit
- producer's risk
- consumer's risk and tolerance limit
Learn next
Learn next how an operating-characteristic curve maps lot quality to acceptance probability. Compare one-sided and two-sided variables plans and investigate the effect of estimating variability. Then examine measurement-system analysis and departures from the assumed distribution before using a plan operationally.
More in Statistics and probability
- Upper confidence bound - An upper confidence bound is an upper limit intended to cover an unknown quantity with a specified confidence under stated assumptions.
- Van der Waerden test - The Van der Waerden test is a rank-based test that compares independent groups by converting pooled ranks into standard normal quantiles, called normal scores.
- Variable sampling fraction - A variable sampling fraction means that different proportions of units are selected from different parts of a population, commonly from different strata.
- Variance - Variance measures spread by averaging squared deviations from a mean.
- Variance components - Variance components are parameters that represent distinct sources of random variation in a statistical model.
- Variance estimation - Variance estimation is the task of estimating the variability of a statistical estimator across repeated samples or repeated realizations of its...
Sources
Sources: NIST, Acceptance sampling by variables, https://nvlpubs.nist.gov/nistpubs/jres/39/jresv39n3p271_A1b.pdf; Variables Sampling Plans Based on the Normal Distribution, Technometrics, https://doi.org/10.1080/00401706.1967.10490485. The NIST PDF could not be retrieved during review; Owen's 1967 publisher abstract was available. Also consulted: D. B. Owen, Factors for One-Sided Tolerance Limits and for Variables Sampling Plans (1963), section 1.3, https://digital.library.unt.edu/ark:/67531/metadc1036120/m1/12/ocr/, for the measurement-based rule and assumptions. The two arithmetic decisions are original and independently checked. The acceptance constant is illustrative, and the manufacturing case is fictional.
Previous: Variable sampling fraction Next: Variance