ABSTRACT
Parametric one-or two-sample analyses of continuous data using test and estimation procedures based on normality are modified for three or more samples. Emphasis shifts from the t-distribution and related test and estimation procedures to the analysis of variance (ANOVA) and tests based on the F -distribution. The t-distribution, however, is related to a particular case of the F -distribution. The concept of experimental design becomes more important. Readers familiar with the analysis of variance in a parametric context, or who are not particularly concerned with that aspect, need only skim through this section briefly, or may prefer to proceed directly to Section 7.2.
