Blackwell Reference Online

Extract

An outlier is an observation or measurement that is unusually large or small relative to the other values in a data set. Outliers occur for a variety of reasons. They can represent, for example, an error in measurement, data recording, or data entry, or a correct value that just happens to be extreme. Outliers can seriously affect the integrity of data and result in biased or distorted sample statistics, inflated sums of squares, distorted p values and effect sizes, and faulty conclusions. Alternatively, they can be the most interesting finding in the data. History records many scientific breakthroughs that have resulted from following up on extreme observations. There is no rigorous definition of an outlier; and no mathematical calculation can tell with certainty whether an outlier comes from the population of interest or a different population. Some outliers are obvious: a student's height of 53 feet and their IQ score of 1,200. Not all outliers are so obvious. A number of rules have been suggested for identifying obvious and not so obvious outliers. Most of the rules involve quantifying how far an outlier is from other data values. One rule identifies an outlier as any measurement or observation that falls outside of the interval given by Ȳ ± 2.5 S , where Ȳ S and S denote, respectively, the sample mean and standard deviation. Unfortunately, Ȳ S and S are greatly ... log in or subscribe to read full text