Answers are grounded in this paper and cite their sources.
SUMMARY Goodman's work on quasi-independence and missing values for contingency tables is unified so as to cover in general incompleteness, as here defined. Cyclic computational procedures are provided for maximum likelihood fitting of iincomplete data. Methods are given for determining appropriate degrees of freedom, whether by identifyinig separable or semiseparable subtables or by elimination of cell isolates. There is discussion of how inicompleteness might arise in other contingency-table situations, e.g. in testing symmetry for square contingency tables. But it is suggested that use of some power-increasingf test might be preferable for detecting departures from symmetry.
No takes yet. Share an insight, caveat, or question.
Nathan Mantel (1970) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: