Rejoinder

We thank the discussants for their varied and thought provoking responses although their number and variety makes a short response difficult. One of our goals, amply fulfilled, was to have extensive discussion of the many questions we raised but did not try to answer. In particular, Fan, Klaassen, and McNeney and Wellner all addressed questions A and B which we respond to further below. Another goal of ours was to have additional important issues raised and here again, we were far from disappointed. Fan addressed the testing question. Klaassen, van der Laan and Yu, and Robins and Rotnitzky raised the important issue of robustness. Shen and Li discuss the estimating equation paradigm, relating the information calculus approach to estimating equa-