How should outliers in background data sets be handled?
Outliers in background may provide limits that are not regulatory conservative. When outlier values are present, they should not automatically be removed, but rather researched to determine whether the value is an error and can either be corrected, or flagged as an outlier and removed prior to construction of statistical limits. Careful consideration should be given when the values cannot be determined to be in error, as these values could represent natural variation. Removing the values could result in limits that will increase exceedances in future compliance points when no contamination is present.
