ABSTRACT
This chapter addresses the question of two-sample testing. Two-sample Gaussian-theory inference primarily concerns expectations; however, one might also compare other aspects of distributions. Under the assumption of an approximate Gaussian distribution, the only additional aspect of the distributions to be compared is their dispersion. The statistic TG(1) is designed to take on large values when items in group one are generally larger than the remainder of the observations and to take small values when items in group one are generally smaller than the remainder of the observations. Westenberg presented the equal-sample case of the statistic, and Mood detailed the use in the case with an even combined sample size. Mood’s test is of sufficiently low importance in applications that the early references did not bother to present the slight complication that arises when the combined sample size is odd.
