on biased inferences about variance components in the binary threshold model在偏见推断二进制阈值的方差分量模型.pdfVIP

Original article On biased inferences about variance components in the binary threshold model C Moreno1D Sorensen LA García-Cortés1 L Varona J Altarriba 1 Unidad de Genética Cuantitativa y Mejora Animal, Facultad de Veterinaria, Universidad de Zaragoza, Miguel Servet 177, 50013 Zaragoza, Spain; Danish2Institute of AnimaL Science, Research Center Foulum, PO Box 39, DK-8830 Tjele, Denmark (Received 2 September 1996; accepted 6 February 1997) Summary - A simulation study was conducted to study frequentist properties of three estimators of the variance component in a mixed effect binary threshold model. The three estimators were: the mode of a normal approximation to the marginal posterior distribution of the component, which is denoted in the literature as marginal maximum likelihood (MML);the mean of the marginal posterior distribution of the component, using the Gibbs sampler to perform the marginalisations (GSR); and third, the mode of the joint posterior distributions of location and the variance parameter, used in conjunction with the iterative bootstrap bias correction (MJP-IBC). The latter was recently proposed in the literature as a method to obtain nearly unbiased estimators. The results of this study confirm that MML can yield biased inferences about the variance component, and that the sign of the bias depends on the amount of information associated with either fixed effects or with random effects. GSR can produce positively biased inferences when the amount of data per fixed effect is small. When fixed effects are poorly estimated, the bias persists, despite the fact that posterior distributions are guaranteed to be proper, and that the a