作者: Yvonne M. M. Bishop , Stephen E. Fienberg
DOI: 10.2307/2528683
关键词:
摘要: This paper examines the estimation of expected cell counts in two-dimensional contingency tables when some cells are known a priori to be empty. A log-linear model, corresponding model independence or homogeneity for 'complete' tables, is fitted counts. The resulting estimates can always computed by an iterative proportional fitting method, although directly. direct method applied set medical data. It pointed out that readily extended higher dimensionality.