INTERCOMPARING RESIDUALS TO FIND OUTLIERS IN RANDOMIZED BLOCKS

The results of experiments laid out in a randomized complete block oft en appear to contain one or more unusual values. Techniques for detect ing such exotic values, understanding of alternative descriptions of e xoticity, one or the other of which applies rather often, leading to w hat might be done when exotic values are detected, can all be of pract ical use. This study was designed to investigate such a two-way data s et: (a) by using for detection the combination of a median polish to h ighlight unusual values and an analysis on its residuals that is guide d by the order statistic medians from a unit half-Gaussian distributio n, (b) by exploring an alternate analysis on a logarithmic scale, and (c) by either Winsorizing or applying missing-value techniques to thos e values found to be exotic. It details the processes and the ways in which the results may be interpreted by application to data from an ex periment conducted by the Rubber Research Institute in Malaysia. This article provides a concise practical tool for the investigation of suc h data, leading into more appropriate statistical analysis.