For the test of two means, to measure the deviation from the null hypothesis, it is logical to take (x 1 – x 2 ) – (μ 1 – μ 2 ) which has a standard d