Simulation studies of smoothing parameter estimates and bayesian confidence intervals in bernoulli ss anova models

Abstract

Wahba, Wang, Gu, Klein and Klein (1994b) introduced Smoothing Spline ANalysis of VAriance (SS ANOVA) method for data from exponential families. They used GCV and UBR methods to choose smoothing parameters. They also constructed Bayesian confidence intervals for SS ANOVA estimates. In this paper, we report simulation results concerning the performance of SS ANOVA estimates with smoothing parameters chosen by GCV or UBR method and their Bayesian confidence intervals. We also compare the performace between GCV and UBR