What are parametric vs. nonparametric limits?
Prior to constructing limits, such as prediction limits, it is necessary to determine whether the background data set follows a normal (or bell-shaped curve) distribution, or whether the distribution is unknown. When the data set follows a normal or transformed-normal distribution, a parametric test may be constructed, where the limit is based on a pre-specified alpha level. The limit can account for the resample plan, and takes into account the background mean and standard deviation, as well as the number of available background samples and the number of future compliance samples that will be compared to the limit. When the distribution of the background data is unknown, a nonparametric limit is recommended which is based on the highest (or sometimes second highest) background value. The alpha level associated with the nonparametric tests is based on the number of background values and can account for the resample plan and number of future compliances values that will be compared to t
