多组均数比较的样本含量估计

Estimation of sample size for comparing more than two treatment groups

Estimation of sample size is a necessary step in clinical trials and experimental study. If the sample size is too small, the power of test would not be enough and the variation among groups may not be found. If the sample size is too large, it would make the practical work difficult and would cause unnecessary waste, moreover make the experiment conditions difficult to control. The variation among groups may not have practical meaning even if the statistical result is of significance. So, we present two tables to detect the sample size when the numbers of groups g and the parameter are known, give the usage of tables for randomized design with samples of unequal size and for randomized block design, and discuss the relations of these tables and the tables of t test.