On the Testability of Models with Missing Data
On the Testability of Models with Missing Data
复制标题
关于缺失数据模型的可测试性
DOI：
--
复制
发表时间：
2014
期刊：
影响因子：
--
通讯作者：
J. Pearl
中科院分区：
文献类型：
--
作者：
Karthika Mohan;J. Pearl
描绘数据丢失过程的图形模型有助于从缺失数据中恢复信息。我们探讨的问题是，鉴于现有数据因缺失而受损，是否任何此类模型都可进行统计检验。我们给出了缺失数据应用中可检验性的充分条件，并指出当数据受缺失项污染时可检验性面临的障碍。我们的结果强化了对完全随机缺失（MCAR）和随机缺失（MAR）的现有检验，并进一步提供了非随机缺失（MNAR）类别中的检验。此外，我们给出了检测变量与其缺失机制之间存在相关性的充分条件。我们利用我们的结果表明，模型敏感性在几乎所有通常归类为非随机缺失的模型中都存在。
Graphical models that depict the process by which data are lost are helpful in recovering information from missing data. We address the question of whether any such model can be submitted to a statistical test given that the data available are corrupted by missingness. We present sucient conditions for testability in missing data applications and note the impediments for testability when data are contaminated by missing entries. Our results strengthen the available tests for MCAR and MAR and further provide tests in the category of MNAR. Furthermore, we provide sucient conditions to detect the existence of dependence between a variable and its missingness mechanism. We use our results to show that model sensitivity persists in almost all models typically categorized as MNAR.