Preliminary analysis of designs with one independent variable with more than two levels

When there are more than two levels of an IV it is not considered appropriate to use a series of tests of the difference between pairs of those conditions. Instead, as a preliminary stage, it is advisable to see whether there is an overall difference between the conditions. Such analysis can be conducted using the appropriate form of the ANOVA, or equivalent for nominal data. However, if homogeneity of variance is not present but the other requirements of a parametric between-subjects ANOVA are fulfilled, then use Welch’s F-test. If there is an overall difference between the conditions, then its precise source can be sought using techniques given in Chapter 18.