2014 Fiscal Year Final Research Report
A new perspective for the effect decomposition problem
Project/Area Number |
23500346
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Multi-year Fund |
Section | 一般 |
Research Field |
Statistical science
|
Research Institution | The Institute of Statistical Mathematics |
Principal Investigator |
KUROKI Manabu 統計数理研究所, データ科学, 准教授 (60334512)
|
Co-Investigator(Kenkyū-buntansha) |
NOMA Hisashi 統計数理研究所, データ科学研究系, 助教 (70633486)
|
Research Collaborator |
JUDEA Pearl UCLA, Cgnitive System Lab, Professor
CAI Zhihong 日本イーライリリー株式会社, 臨床開発部, アソシエイト
SANO Natsuki 東京理科大学, 理工学部, 助教 (60568971)
|
Project Period (FY) |
2011-04-28 – 2015-03-31
|
Keywords | 未観測交絡因子 / 直接効果 / 間接効果 / 代替エンドポイント / 総合効果 |
Outline of Final Research Achievements |
In this research, I considered evaluation problems for direct and indirect effects through a new causal concept, `natural'. Especially, I developed the effect restoration methods to derive the consistent estimator of the total effects when unmeasured confounders exist. In addition, I proposed selection criteria for intermediate variables to estimate direct effects with the variation reduction, and provided a new look at the propensity score method from the proposed criteria.
|
Free Research Field |
統計科学
|