Subdivision techniques in applying the bartlett-kendall test for equal variances

In applying the Bartlett-Kendall test for equal variances, an arbitrary decision must be made as to how samples should be divided into subsamples. Since this decision is not always clear-cut, the purpose of this study was to compare nine different procedures which the researcher might employ in accomplishing the aforementioned purpose. Since previous research studies have answered this question for equal sample sizes, this study dealt only with unequal sample sizes. Six different sample size conditions were considered in the investigation, three of which involved large absolute differences among sample sizes and three of which involved small absolute differences. Each of the nine procedures was first applied in the circumstance in which variances were homogeneous as a means of determining each procedure's ability to control Type I error rates. Then four conditions under which the variances were heterogeneous to different degrees were included so as to provide estimates of each procedure's power at a given nominal alpha level. Four of the nine alternatives proved to be consistently superior to the other five alternatives and relatively equal in performance when compared with one another.