An Evaluation of Multiple Comparison Procedures for Proportions: A Monte Carlo Study
Multiple comparison procedures (MCP) for proportions are used to indentify the significant differences between pairs of groups after the Chi-square test results in a p-value of 0.05 or smaller when more than two groups are compared. In order to choose the most appropriate MCP, 7 representative MCPs were selected and evaluated in Type I error, Type II error and correct decision rate using Monte-Carlo simulation. Results indicated that if the significance level needs to be controlled within 0.05, Hochberg would be the first choice; if the significance level is allowed within 0.1, the recommended MCP would be Tukey-Kramer with a sample whose size is smaller than 20 and SNK. or FDR when a samples size bigger than 20. SNK and Tukey-Cramer would be recommended when there are more than 4 groups.
multiple comparison procedures monte carlo simulation FEW CP
SUN Hongwei XU Tianhe HAN Chunlei
Department of Health Statistics and Epidemiology, Binzhou Medical University, Yantai, P.R.China, 264 Department of Journals, Binzhou Medical University, Yantai, P.R.China, 264003
国际会议
威海
英文
634-640
2010-07-24（万方平台首次上网日期，不代表论文的发表时间）