本刊1990年3期刊登的陈森同志的文章《多重比较的应用》,文中谈到“由于Q检验要求各水平的试验次数相同,假若方差分析结果各水平间差异显著,然而因各水平的试验次数不同,为比较x_i与x_j)i,j=1,2,…,a)之间差异是否显著,此时Q检验就无法进行比较了,这是Q检验的不足之处,为此可用S检验…”。其实,当各水平的试验次数不相同,设为n_i,若能求出一个 The article “Combinations of Multiple Comparisons” published by Chen Chen in 1990 issue 3 states that “since the Q test requires the same number of tests at each level, if there is a significant difference between the levels of the analysis of variance, The number of tests is different. In order to compare whether the difference between x_i and x_j) i, j = 1, 2, ..., a) is significant, the Q test can not be compared at this time. This is a deficiency of the Q test S test ... ”. In fact, when the number of tests at different levels is not the same as n_i, if we can find one