SINDSCAL source weight transformations are not always necessary or desirable: Reply to Hodgkinson

Hodgkinson (1998) argues that we should have transformed the SINDSCAL source weights we used for an analysis presented on page 15 of our paper on the judgement of HIV cases (Irwin, Jones, and Mundo, 1996). In this reply we present theoretical and empirical arguments for why we did not transform the weights. Our statements are presented in the context of a more general discussion of how source weights should be interpreted and used. (C) 1998 John Wiley & Sons, Ltd.