Abstract
In modeling a response as a function of a set of explanatory variables, one way to check lack of fit in a fitted model of the form E(Y) = Xβ1, when the true model is E(Y) = Xβ1 + X22β2, is to use additional check points. The question of choosing check points has not often been addressed, so we propose a method of selecting check points to maximize the power of the lack-of-fit test. Two examples illustrate the procedure.