TS 13216

TS 13216 defines a standard practice for calculating the sample size needed to estimate, with a specified tolerable error, the average for a characteristic of a lot or process. In practical terms, it helps determine how many units should be sampled when the goal is to estimate a mean value or the proportion of nonconforming units in a batch or manufacturing process. The standard is about planning the sampling effort, not about judging the product itself.

What the calculation is meant to support

In general statistical practice, sample size is chosen to balance precision, effort, and uncertainty. A larger sample usually improves the reliability of an estimate, but also increases time and cost. TS 13216 addresses this planning step for random samples taken from a lot or from process output, using simple methods to reach a target level of accuracy. The standard’s emphasis is on the calculation after sampling units have been defined; it does not aim to explain how materials or products should be handled or how suitable sampling units should be identified with a stated reliability.

Scope boundaries and statistical conditions

The document also makes an important statistical distinction: if the manufacturing process is not in statistical control, the calculated result should not be treated as a forecast of future production. Another boundary is variability structure. The method applies when there is a single source of variability in the samples; it does not extend to situations with multiple levels of variability. That makes the practice useful as a focused sampling-planning tool, but not a universal treatment for every production or inspection scenario.

• Mean estimation for a lot or process characteristic
• Estimation of the proportion of unsuitable units
• Sample size calculation with a specified tolerable error
• Use in random sampling contexts
• Assumption of a single source of variability