Cr. Blyth et Rg. Staudte, HYPOTHESIS ESTIMATES AND ACCEPTABILITY PROFILES FOR 2X2-CONTINGENCY-TABLES, Journal of the American Statistical Association, 92(438), 1997, pp. 694-699
By estimating, rather than testing hypotheses regarding the degree of
dependence between the factors in 2 x 2 tables, the technical difficul
ties associated with small sample sizes are avoided. The estimators pr
oposed here attempt to estimate I when the alternative hypothesis is t
rue and 0 when the null hypothesis is true, subject to a bound on the
squared error loss under the hypothesis. Such estimators provide guard
ed weights of evidence for the alternative hypothesis. Guarded weights
of evidence based on the likelihood ratio are compared with those bas
ed on the p value or mid-p value, and they are shown to have lower ris
k functions except when the alternative is far from the hypothesis. Fo
r the case of two independent binomial distributions, it is shown that
the conditional likelihood ratio estimator for the hypothesis of homo
geneity against the two-sided alternative has a smaller unconditional
risk than the unconditional likelihood ratio estimator, except when th
e binomial probabilities are far apart. Inversion of a family of guard
ed weights of evidence leads to acceptability profiles. These profiles
provide more information than traditional confidence intervals regard
ing the unknown parameter. Two-sided profiles are found for the degree
of dependence as measured by the odds ratio and log-odds ratio, and o
ne-sided profiles are found for Yule's Q.