The data for are presented as the test score means of individual
measurements. Suppose that the number of measurements used to arrive at
each mean are different for some of the groups. For example, the means for boys
performing exercise 1 were 18.8, 20.2, 27.2, and 30.3, and these means were based
on 20 measurements each. In contrast, the means for girls performing exercise 1
were 16.4, 24.1, 32.9, and 35.5, and these means were based on 10 measurements
each. Which important assumption, used by ANOVA, might be violated because of
this disparity?