Comparing the Means of Two Independent Groups

✍ Scribed by Prof. Rand R. Wilcox

• Book ID: 102759827
• Publisher: John Wiley and Sons
• Year: 2007
• Tongue: English
• Weight: 528 KB
• Volume: 32
• Category: Article
• ISSN: 0323-3847
• DOI: 10.1002/bimj.4710320702

✦ Synopsis

Recently, CRESSIE and WEITFORD (1986) showed that Welch's test of Ho : pi =& con be biased, under nonnormality, where p , and p z are the means of two independent treatment groups. They suggested, therefore, that a two-sample analog of Johnson's test be used instead. One goal in this paper is to examine the extent to which a two-sample analog of Johnson's test improve8 upon Welch's technique in terms of Type I errors and power when sample sizes are small or moderately large. Several alternative procedures are also considered. including an additional modification of Johnson's procedure, a procodure suggested by DUNNETT (1982) that -8 Tiku's modified maximum likelihood estimate with 10 n, ' , , trimming, two versions of Efron's bootstrap, and a test recently proposcd by WILCOX (1989). The paper also describes situation8 where Welch's procrdurc is not robust in ternis of Typo T errors. This is important because based on results ciirrcntly available, Wclch's prorctliil.c is tlioiight to hn nonrobust in terms of power, but robiist i n ternis of Type I errom.