Exact conditional tests of quasi-independence for triangular contingency tables: estimating attained significance levels

摘要: SUMMARY Some investigations of whether disease is related to birth order result in a triangular table counts diseased individuals classified by and sibship size. Here testing for an interaction corresponds quasi-independence the classification factors. We propose Monte Carlo algorithm estimating significance level this test, which should be used when asymptotic results are suspect. describe procedure using classic classifying stroke patients. Unlike other researchers, we conclude that there moderate evidence rejecting here. Our main application concerns relationship between neurosis order. strong favour uniform association model.