Table MTE0. Two group t-test of equivalence in means (equal n’s)

This sample size table is for a one-sided two-group t-test of the null hypothesis that the means are not equivalent (i.e. that the test drug mean is “worse than” the standard drug mean by an amount Δ0 or more). The alternative hypothesis is that the means are equivalent, defined as differing by only Δ1 (usually 0). Computed using t and non-central t distributions; the non- centrality parameter is (δ)(sqrt(n))/sqrt(2) where δ is defined in the table row title. See nQuery Advisor® manual Appendix for details of computing method for t. Results for this table then will agree exactly with those for Table MTT0, the two-sample t-test, when a one-sided test is selected in MTT0 and the numeric values for effect size are the same. Figure MTE0 shows computations for two examples. Additional values tested: significance level .001, .20 one- sided by power of 50% and 99% and effect size of .05 and 6 to n, values of n equal to 3 and 2000 and effect sizes of .05 and 6 back to power. Checked against MTT0 and tables for two-group t-test in Machin D, Campbell MJ (1987) Statistical Tables for the Design of Clinical Trials, Blackwell Scientific Publications Oxford, and two-group t-test in DOS software Stat-Power version 2.0 by Bavry JL (1991) Scientific Software, Inc and Statistical System Power Analysis by Hintze JL (1991) BMDP Statistical Software.

Figure MTE0. In both columns, nQuery Advisor® 7.0 has solved for n.