Exact approaches to testing problems involving nuisance parameters
Consiglio, Joseph D.
MetadataShow full item record
We propose exact approaches for tests of hypotheses in the presence of at least one nuisance parameter. The procedures discussed are useful in situations where the available sample size is small, and procedures grounded in large-sample theory may not be applied without consequence. We present an exact test of monotone trend useful for dose-response studies in which the dose variable is qualitative in nature. Further approaches are developed to incorporate a quantitative predictor. Finally, we develop an exact test about the kappa coefficient in studies of inter-rater reliability, and exact tests for homogeneity of agreement across multiple studies.