Variability Analysis for Three or More Samples

Abstract

In some clinical studies, the spread of the data may be more relevant than the average of the data. E.g., when we assess how a drug reaches various organs, variability of drug concentrations is important, as in some cases too little and in other cases dangerously high levels get through. Also, variabilities in drug response may be important. For example, the spread of glucose levels of a slow-release-insulin is important. In Chap. 21, the chi-square test for one sample and the F-test for two samples have been explained. In this chapter we will explain the Bartlett’s test which is suitable for comparing three or more samples. An alternative to the Bartlett’s test is the Levene’s test. Levene’s test requires a lot of arithmetic, and is usually performed using statistical software. Assessing significance of differences between 3 or more variances does not answer which of the samples produced the best outcome. Just like with analysis of variance (Chap. 19), separate post hoc one by one analyses are required.