Non-Parametric Comparison of Single Parameter Histograms

Abstract

A number of methods have been developed to compare single parameter histograms. Some perform a channel-by-channel analysis and others give a single statistic about how the histograms may or may not differ. If they do differ, then the significance of the difference or confidence limit is usually provided. The specific location(s) for the greatest deviations may also be given. Some are more effective at resolving severely overlapping populations and others work poorly when there is any significant overlap. Each method makes certain assumptions about the data. It is important to understand the assumptions being made and to understand the limitations of each method. It is essential to know how to identify when a comparison method will work for a given set of histograms. This unit explores the different methods, and provides a guide for the reader to choose the most appropriate method(s) to use for a specific data set(s). © 2018 by John Wiley & Sons, Inc.