Bias Affiliated With Two Variants Of Cohen's d When Determining U1 As A Measure Of The Percent Of Non-Overlap

David A. Walker
2005 Journal of Modern Applied Statistical Methods  
Variants of Cohen's d, in this instance d t and d adj , has the largest influence on U 1 measures used with smaller sample sizes, specifically when n 1 and n 2 = 10. This study indicated that bias for variants of d, which influence U 1 measures, tends to subside and become more manageable, in terms of precision of estimation, around 1% to 2% when n 1 and n 2 = 20. Thus, depending on the direction of the influence, both d t and d adj are likely to manage bias in the U 1 measure quite well for smaller to moderate sample sizes.
doi:10.22237/jmasm/1114906260 fatcat:34bqsg6hvjed7kmlfqk23ofw5u