Title: Asympirical method: A new paradigm for the statistical analysis of large samples
Authors: Ping Ma - University of Georgia (United States) [presenting]
Abstract: Traditional statistical theory and methods are developed for small and mild size samples. In particular, statistical model fitting and inference are conducted in the small and mild size samples to get empirical results. Asymptotic theory is established to extrapolate the performance of the empirical results to large samples. However, this traditional coherent statistical analysis paradigm falls apart in large samples. The key challenge is that many traditional statistical methods are computational too expensive to get meaningful empirical results. A new statistical paradigm is in urgent need for the statistical analysis in large samples. We will present an asympirical (asymptotic + empirical) method, which is designed by the principle that theory informs practice. We will present it in the context of smoothing spline ANOVA models. Simulation and real data analysis will be used to demonstrate the performance of the new paradigm.