ABSTRACT
This chapter looks at the variation of the scores within any group or treatment. The variation of scores within a treatment is due to chance variations in the population, because of differences between people and uncontrolled experimental variables. They can be thought of as "experimental error". The chapter compares the differences between treatments with the differences within treatments (experimental error). The range of treatment means is larger than the range of scores within a treatment (experimental error). This is the basic logic of analysis of variance. The chapter also compares the range of scores between and within treatments, but the range is not really the best measure because the odd "wildcat" extreme value may give an exaggerated appearance of overlap of the scores. It is worth considering formally the assumptions for use of the one-factor completely randomized analysis of variance (often abbreviated ANOVA). It is called "one-factor" because we are only varying one set of treatments.
