Dealing With Outlying Observations.
NIOSH 1983 Aug:93 pages
Materials for a course concerned with outlying observations in experimental data are presented. The course is part of a NIOSH program of training for industrial hygienists, analysts, laboratory scientists, technicians, and others, that offers instruction in how to deal with and evaluate outlying observations in a statistically valid manner. The materials in the course provide a number of techniques for treating outliers, depending on their number and place in the array of observed data. The course consists of ten lessons. Topics include: an introduction to outlying observations; outlier rejection techniques; statistical tests available for testing a suspected outlying observation and the criteria for selecting a test; techniques for dealing with two or more outliers in a sample, handling of data sets using the k/sample slippage test of Conover and the extreme rank sums test of Youden; and rules for properly considering duplicate and triplicate readings from an instrument or laboratory results. Practice exercises and problems are also considered. The final lesson consists of a student evaluation of the course.
NIOSH-Author; Quantitative-analysis; Industrial-equipment; Industrial-environment; Analytical-methods; Laboratory-techniques; Industrial-exposures; Industrial-hazards; Analytical-instruments; Industrial-safety;
NTIS Accession No.
Division of Training and Manpower Development, NIOSH, U.S. Department of Health and Human Services, Cincinnati, Ohio, 93 pages, 44 references