# Statistical Advisor, Sampling Plans

POWER ANALYSIS. This chapter discusses when to perform sample size estimation (and power analysis) for a variety of research designs, and statistical tests.

PROCESS ANALYSIS. This chapter discusses techniques to generate and analyze fixed and sequential sampling plans for normal means, binomial frequencies, and Poisson frequencies.

In general, sampling plans are devised in order to detect quality problems in a production process or supplier with a particular degree of certainty. After specifying the extent of acceptable deviation from specifications, the amount of random (acceptable) variability, and the probability or certainty with which one desires to detect shifts in the product quality, one can compute the necessary size of the sample that one needs to take from the production process.

Fixed sampling plans are those where single samples of a fixed size are taken from the production process or from product lots shipped by the supplier. Sequential sampling plans are those where individual items (parts, products) are inspected, one by one. The cumulative deviation of those items from specification is then plotted. The program will establish critical limits (lines) in that plot that will allow the engineer to decide whether or not the process or supplier complies with specifications. Sequential sampling plans are usually more efficient than fixed sampling plans.