4.7 The Wilcoxon-type test of hypothesis H : (2) = 0 in the model

(see (4.69), X(1) is n × p and X(2) is n × q), based on the regression rank scores, is based on the following criterion. Assume that (1) contains an intercept, i.e., the first column of X(1) is 1n. Calculate the regression rank scores (ˆan1(), . . . , ˆann()), 0 1, for the hypothetical model Y = X(1)(1) +

e, the Wilcoxon scores

where 2q (.95) is the 95% quantile of the 2q distribution (Gutenbrunner et al. (1993)).

Transcribed Image Text:

Y=X)+X2(2)+e