ABSTRACT
Suppose first that the null hypothesis is that the random variables E l , . .., Yn are i.i.d. in some unknown density. The type of departure to be tested may, for example, be (a) trend with serial order, (b) de pendence on some explanatory variable, (c) stochastic difference be tween observations 1 ,. . . , «i and observations nx + 1 ,. . . , n.