Published under: data analysis, Data analytics, Statgraphics, tolerance limits, non-normal data

Constructing Statistical Tolerance Limits for Non-Normal Data

This webinar describes the process of constructing statistical tolerance limits given data from a population which is not characterized by a normal distribution. Statistical tolerance limits are widely used to demonstrate that the underlying process which generated the data conforms to a set of requirements or specifications. A step-by-step process is described which includes tests for normality, power transformations, selection of alternative distributions such as the Weibull or extreme value distribution, and nonparametric tolerance limits. The problem of sample size selection is also considered.