Is SQC identical to SPC?

Usage varies. Here SQC is broader, including product and lot evaluation as well as process monitoring.

Does an accepted lot contain no defects?

No. Sampling plans allow some chance of accepting lots at different defect levels.

Can SQC apply to services?

Yes, when requirements and measurable outcomes are defined appropriately.

AI Glossary / Statistics and probability / Intermediate

Statistical quality control

Definition

Statistical quality control, or SQC, applies statistical methods to measure, monitor, evaluate, and improve quality in relation to defined requirements. It includes approaches such as process monitoring, capability analysis, and acceptance sampling. The exact boundaries of the term vary, and it is sometimes used interchangeably with statistical process control. In the broader usage adopted here, SQC includes both understanding process performance and making evidence-based decisions about products or lots. It does not guarantee that every accepted item is defect-free.

What it means

Quality decisions are made with incomplete information. Testing every item may be expensive, slow, or destructive, while process performance naturally varies. Statistical methods help quantify what can be learned from samples and how much uncertainty remains.

Different SQC tools answer different questions. A control chart asks whether process behavior has changed. A capability analysis compares stable process variation with requirements. An acceptance-sampling plan supports a decision about a particular lot under defined risks. Confusing these purposes can lead to inappropriate conclusions even when each calculation is performed correctly.

How it works

Define the quality characteristic, requirement, unit of inspection, and decision to be made. Establish a measurement method that is sufficiently accurate and consistent for that decision. Choose a statistical approach appropriate to the objective and data structure.

For process monitoring, collect ordered data and investigate signals against an established baseline. For capability, evaluate whether the stable process distribution is compatible with specifications. For acceptance sampling, specify how units are selected, how many are inspected, and the rule for accepting or rejecting the lot. Examine the plan's operating characteristic to understand its acceptance probabilities at different quality levels.

Connect results to a documented response and improvement process. Sampling and measurement assumptions, dependence, supplier differences, and changes over time must be considered. Track the consequences of decisions and revise methods when they no longer match the operation. Statistical quality control complements engineering and operational knowledge; it cannot compensate for an undefined requirement or a flawed measurement system.

Three practical examples

1. A manufacturer monitors product dimensions over time and investigates changes before they become widespread quality problems.

2. A receiving department applies a documented random-sampling plan to decide whether to accept a shipment.

3. A service organization studies recurring reporting errors, quantifies variation, and evaluates whether a redesigned workflow meets its accuracy requirements.

Think of it this way

Think of quality control as several instruments on a dashboard. One shows whether behavior is changing, another whether performance meets a target, and another supports a particular acceptance decision. Reading one instrument cannot answer every question about the vehicle.

Worked example

Consider an original simplified acceptance plan that inspects 20 independent items and accepts when at most one is defective. Under a binomial model with defect probability p, the probability of acceptance is P(X ≤ 1) = (1 - p) to the power 20 + 20p(1 - p) to the power 19.

If p = 0.10, the acceptance probability is 0.9 to the power 20 plus 20 times 0.10 times 0.9 to the power 19, approximately 0.391747. Thus this plan still accepts about 39.17% of samples generated at a 10% defect probability.

If p = 0.02, the acceptance probability is approximately 0.940101, or 94.01%. The corresponding rejection probability is about 5.99%. These two points illustrate how a sampling plan's decision behavior depends on the underlying quality level.

The plan is not recommended here; it is chosen to make the arithmetic clear. A real plan should be selected using justified quality levels and acceptable decision risks. For sampling without replacement from a finite lot, a hypergeometric model may be appropriate instead of the independent binomial model. Accepting a lot under the rule does not establish that its defect fraction is at most one twentieth or that every untested item conforms.

Illustrative case study

In a fictional appliance business, incoming lots often pass a small acceptance sample, but field failures remain costly. Managers initially assume that passing the sample certifies the entire shipment.

The quality team explains the plan's operating characteristic and reviews whether its risks match the business consequences. It also investigates supplier processes and field failure mechanisms rather than only increasing inspection. The revised program combines appropriate acceptance decisions with process improvement and better measurement, recognizing that inspection alone does not create quality in the product.

Common misconceptions and mistakes

Do not treat acceptance as proof of perfect quality or a direct estimate of the lot's defect rate. Acceptance sampling is a decision procedure with known or modeled error risks. Its sample size and acceptance number must be considered together.

An in-control process can still be incapable of meeting specifications. A high capability estimate can be misleading if the process is unstable or its distributional assumptions are unsuitable. More inspection does not necessarily address the cause of defects, and inspecting only convenient items can invalidate the sampling model. SQC also depends on consistent definitions of defects and reliable measurement; changing either can make apparent improvement difficult to interpret.

Business use

SQC helps businesses make quality decisions transparently and allocate effort between prevention, monitoring, and inspection. It supports supplier management, process improvement, and customer assurance when methods are matched to consequences. Document sampling rules, measurement performance, acceptance criteria, and follow-up actions. Report uncertainty and limitations so stakeholders understand what a passing result establishes and what it leaves unresolved.

Check your understanding

A lot passes a plan that inspects 20 items and accepts at most one defect. Does this prove the whole lot has no more than 5% defects?

Answer: No. The observed sample and acceptance rule do not establish that population bound. The plan's operating characteristic and an appropriate estimation procedure answer different questions.

Related terms

Statistical process control
Acceptance sampling
operating characteristic
process capability
specification
producer's risk
consumer's risk
measurement system

Learn next

Compare monitoring, capability assessment, and lot acceptance as separate tasks. Learn operating-characteristic curves before selecting a sampling plan or interpreting an accepted shipment.

Sources

NIST/SEMATECH, glossary, Statistical quality control: https://www.itl.nist.gov/div898/handbook/glossary.htm

NIST/SEMATECH, What is Acceptance Sampling?: https://www.itl.nist.gov/div898/handbook/pmc/section2/pmc21.htm

The primary handbook terminology and acceptance-sampling explanation were read. The binomial plan and fictional case are original illustrations, not a recommended inspection specification.

Previous: Statistical process control Next: Statistical relational learning