摘要
本文探讨了删失数据下带固定效应部分线性面板数据模型在协变量含测量误差情况下的统计推断。首先用Kaplan-Meier估计来处理删失数据带来的估计偏差,利用虚拟变量方法来消除未知固定效应;然后运用“衰减校正”来解决测量误差引起的不一致问题,构造了协变量含测量误差情况下未知参数的修正经验对数似然比统计量,证明了其服从卡方分布。最后,通过实验模拟验证了方法具有良好的效果。
This paper discusses the statistical inference of the partial linear panel data model with fixed effects under censored data in the case of covariates with measurement errors.Firstly,the Kaplan-Meier estimation is used to deal with the estimation bias caused by censored data,and the dummy variable method is used to eliminate the unknown fixed effects.Then,the'attenuation correction'is used to solve the inconsistency problem caused by measurement errors,and the modified empirical log-likelihood ratio statistics of unknown parameters in the case of covariates with measurement errors are constructed,which is proved that it obeys chi-square distribution.Finally,the experimental simulation verifies that the method has a good effect.
作者
姚钰琦
何帮强
YAO Yuqi;HE Bangqiang(School of Mathematics-Physics and Finance,Anhui Polytechnic University,Wuhu Anhui 241000,China)
出处
《阜阳师范大学学报(自然科学版)》
2025年第1期8-14,共7页
Journal of Fuyang Normal University:Natural Science
基金
国家社会科学一般基金项目(18BTJ034)
国家自然科学基金面上项目(72271003)。
关键词
部分线性模型
经验似然
固定效应
变量误差
删失数据
partially linear model
empirical likelihood
fixed effect
errors-in-variables
censored data
