The author studies the linear rank statistics for testing the pypothesis of randomness against the alternative of two samples provided both are drawn grom discrete (integer-valued) distributions. The weak law of large numbers and the exact slope are obtained for statistics with randomized ranks of with averaged scores.
The equivalence of the symmetry of density of the distribution of observations and the oddness and evenness of the score-generating functions for the location and the scale problem, respectively, is established at first. Then, it is shown that the linear rank statistics with scores generated by these functions are asymptotically independent under the hypothesis of randomness as well as under contiguous alternatives in the last part of the paper. The linear and quadratic forms of these statistics...
The paper deals with the experimental design which is optimal in the following sense: it satisfies the cost requirements simultaneously with a satisfactory precision of estimates. The underlying regression model is quadratic. The estimates of unknown parameters of the model are explicitly derived.
Page 1