Model Checking with Nonparametric Curves
描述了在模型检验中使用局部似然曲线估计的方法，基于偏差差异构建检验统计量，并通过非参数自助蒙特卡洛进行解释，适用于离散数据回归。
The use of local likelihood curve estimates in model checking is described. Test statistics are based on deviance differences, and are interpreted using nonparametric bootstrap Monte Carlo. Some theoretical aspects are discussed. Methods are illustrated for regression with discrete data.