SAIC

Description: It has been said that you can prove anything with statistics. However, the “proof is in the pudding” and valid proofs depend on valid application of statistical principles and assumptions. 

The use of MARSSIM (Multi-Agency Radiation Survey and Site Investigation Manual) guidance and its supplement MARSAME (Multi-Agency Radiation Survey and Assessment of Material and Equipment Manual) have provided a statistical framework for sampling and analysis of characterization data for both site decommissioning and waste management projects based on a standard data life cycle and on meeting developed data quality objectives. However, statistical assumptions are often not verified to be applicable to the material in question. Furthermore, misunderstanding of how to apply statistical principles and methods to radiological data can lead to erroneous conclusions. 

Some examples of misapplication of statistics include out of scope application of guidance, misapplication of statistics across multiple populations for statistical sampling based on poor or missing historical process knowledge or scoping data; poor or unverified assumptions underlying the evaluation method or statistical tests; improper basis for statistical sampling design by applying area based to methods to discrete populations or by applying decisions based on means of populations without evaluation of hot spots; and, improper application of statistical methods for “undetected” versus “detected” data. Some available statistical methods and software packages lend themselves to misapplication in these ways, especially for the unwary and uninitiated. Examples from some actual site decommissioning and waste management projects using various statistical methods and available statistical software packages will demonstrate the misapplication and proper application of statistical principles.