Questioning the Defense of Different Numbers

Algozzine and Ysseldyke's (this issue) criticisms of our examination of prevalence rate variability for the special education categories using the coefficient of variation (Hallahan, Keller, & Ball, 1986) rest on three points: (a) Inferential statistics and sampling distributions should be used for comparing measures of prevalence rate variability, (b) different conclusions can be drawn when such statistics and sampling distributions are used, and (c) the coefficient of variation seems to be a biased measure of variability. A consideration of the nature of prevalence rate data, the correct use of inferential statistics, and the coefficient of variation itself suggests that their arguments are not justified.