Cantoni Eva, Flemming Joanna Mills, Ronchetti Elvezio
Department of Econometrics, University of Geneva, CH-1211 Geneva 4, Switzerland.
Biometrics. 2005 Jun;61(2):507-14. doi: 10.1111/j.1541-0420.2005.00331.x.
Variable selection is an essential part of any statistical analysis and yet has been somewhat neglected in the context of longitudinal data analysis. In this article, we propose a generalized version of Mallows's C(p) (GC(p)) suitable for use with both parametric and nonparametric models. GC(p) provides an estimate of a measure of model's adequacy for prediction. We examine its performance with popular marginal longitudinal models (fitted using GEE) and contrast results with what is typically done in practice: variable selection based on Wald-type or score-type tests. An application to real data further demonstrates the merits of our approach while at the same time emphasizing some important robust features inherent to GC(p).
变量选择是任何统计分析的重要组成部分，但在纵向数据分析的背景下却有些被忽视。在本文中，我们提出了一种适用于参数模型和非参数模型的广义 Mallows's C(p)（GC(p)）版本。GC(p) 提供了对模型预测充分性度量的估计。我们使用流行的边际纵向模型（通过广义估计方程（GEE）拟合）来检验其性能，并将结果与实际中通常所做的进行对比：基于 Wald 型或得分型检验的变量选择。对实际数据的应用进一步证明了我们方法的优点，同时强调了 GC(p) 固有的一些重要稳健特征。
