The analysis of one-way between effects in fluency research

In fluency research, the potential for extreme variance heterogeneity exists when comparing stutterers to other disordered or control populations. There is now a growing concern on the part of statisticians that the F statistic is not nearly so robust to violations of the homogeneity of variance assumption as was once thought. Together, these two circumstances represent potential for compromise on the statistical conclusion validity of fluency research. In this first of two papers, alternatives to the usual analysis of variance are described for the research design involving a single between effect. A second paper will be a similar treatment of the two-way between effects design.