A. M. Franco Pereira, M. C. Pardo Llorente, C. T. Nakas
In this work, two families of statistics are studied in order to address the problem of testing order when the response variables are binary observed in k groups. The asymptotic distributions of the proposed statistics have been obtained and a simulation study has been carried out to compare the empirical powers obtained for different members of these families, among which is the likelihood ratio statistic. The methodology developed in this work was motivated by a real problem that was raised to us about the efficacy of the aerial insecticide phosphine and in which, despite obtaining complete mortality with a dose of 1000 ppm, an unexpected survival was obtained at higher doses, suggesting that the same insecticide might have acted differently at high concentrations for specific insects.
Keywords: chi-squared test statistic, likelihood ratio test statistic, power divergence, tree ordering, umbrella ordering
Scheduled
Invited Session Recent advances in goodness-of-fit and k-sample tests
June 10, 2022 10:10 AM
A13
