Title: Testing marginal homogeneity for paired ordinal data: A robust likelihood approach
Authors: Tsung-Shan Tsou - Institute of Statistics, National Central University, Taiwan (Taiwan) [presenting]
Abstract: Pairing lessens heterogeneity between subjects but introduces correlation that is nuisance to our interest. This makes modeling the underlying probability structure more complex and inference more challenging. The problem is more severe especially for discrete data. Instead of confronting the nuisance parameter, we use the simpler model for the parallel design and develop from it a robust score statistic for testing the marginal homogeneity of two distributions of paired ordinal data. The effect of the overlooked correlation is recuperated externally from data. We provide simulations and real data analyses to demonstrate the effectiveness of the robust test.