响应变量缺失下部分线性单指标模型的序列相关性检验

Serial Correlation Test for Partial Linear Single-Index Model with Missing Response Variables

研究了在响应变量随机缺失下的部分单指标模型的序列相关检验问题。首先采用借补的方法对缺失响应变量进行处理,再运用经验似然方法对残差部分进行序列相关性检验,构造了经验似然比统计量,并证得其为渐近分布。数值模拟结果表明:该检验方法具有较为理想的检验功效。

We considered the serial correlation test for partial linear single-index model with response variables missing at random（ MAR）. Firstly,we filled in the missing response variables by the imputation method. Then we applied the empirical likelihood method to establish the test statistic,and constructed the ratio statistic of empirical likehood and derive the asymptotic distribution of the statistic. Simulation results indicate that the test method performs well.