医学统计学（扬州大学）统计基础和多重比较过程的一致性.pdfVIP

Basic Statistics and the Inconsistency of Multiple Comparison Procedures DAVID J. SAVILLE, AgResearch, New Zealand Abstract This paper has two main themes. First, the sponding to selected treatments. This leads to a rec- various statistical measures used in this journal are ommendation that planned and unplanned compar- summarized and their interrelationships described by isons between treatment means should be made using way of a flow chart. These are the pooled standard the same statistical tool (e.g., a 5% level test per com- deviation, the pooled variance or mean square error parison), without any adjustment for multiplicity. The (MSE), the standard error of each treatment mean (SEM) proviso is that experimenters should clearly distinguish and of the difference between two treatment means between the formulation of new hypotheses involving (SED), and the least difference between two means such comparisons (that require confirmation in subse- which is significant at (e.g.) the 5% level of significance quent experimentation), and the testing of pre-existing, (LSD(5%)). The last three measures can be displayed as a priori hypotheses that are being confirmed in the cur- vertical bars in graphs, and the relationship between rent experiment. the lengths of these bars is graphically illustrated. It is suggested that the LSD is the most useful of these three Basic Statistical Measures measures. Second, when the experimenter has no prior hypotheses to be tested using analysis of variance As an agricultural research statistician attempting to “contrasts,” a multiple comparison procedure (MCP) that write an article for an un