TY - JOUR AU - Nash, John C. AB - Letters to the Editor Fabian dismissed their suggestion as irrelevant because of the different goals ACCOUNTING FOR INTERACTIONS addressed by the literature on shrinkage estimation. Note, however, that the Fabian (1991) considered the problem of estimating the population cell point estimator suggested by Casella and Hwang, means Ilij in a two-way ANOY A model in which additivity is conjectured but the presence of interactions cannot be ruled out. His purpose was to flij = YOij + h( ~ W~)Wij, h(x) = (I -~r show that a common recommendation (Ro in his notation) to test for inĀ­ teractions and then proceed with estimates based on the additivity model if is a convex combination this test fails to reject additivity is seriously flawed. To do so, he introduced what he called an improvement to Ro (R in his notation; specifically, the procedure in his Sec. 3.1). He then showed that this improved procedure is inferior to direct estimation ofthe Ilij ignoring possible additivity. Although his criticism of the approach Ro is valid, his broader criticism against using of the additive estimator YOij and the usual estimator Yij. For small the structure of the two-way ANOY A model in estimation is flawed TI - Comment on Efron JF - Journal of the American Statistical Association DO - 10.1080/01621459.1992.10475297 DA - 1992-09-01 UR - https://www.deepdyve.com/lp/taylor-francis/comment-on-efron-61PCSUub0G SP - 912 EP - 913 VL - 87 IS - 419 DP - DeepDyve ER -