### Abstract

The goal of this article is to construct doubly robust (DR) estimators in ignorable missing data and causal inference models. In a missing data model, an estimator is DR if it remains consistent when either (but not necessarily both) a model for the missingness mechanism or a model for the distribution of the complete data is correctly specified. Because with observational data one can never be sure that either a missingness model or a complete data model is correct, perhaps the best that can be hoped for is to find a DR estimator. DR estimators, in contrast to standard likelihood-based or (nonaugmented) inverse probability-weighted estimators, give the analyst two chances, instead of only one, to make a valid inference. In a causal inference model, an estimator is DR if it remains consistent when either a model for the treatment assignment mechanism or a model for the distribution of the counterfactual data is correctly specified. Because with observational data one can never be sure that a model for the treatment assignment mechanism or a model for the counterfactual data is correct, inference based on DR estimators should improve upon previous approaches. Indeed, we present the results of simulation studies which demonstrate that the finite sample performance of DR estimators is as impressive as theory would predict. The proposed method is applied to a cardiovascular clinical trial.

Original language | English (US) |
---|---|

Pages (from-to) | 962-972 |

Number of pages | 11 |

Journal | Biometrics |

Volume | 61 |

Issue number | 4 |

DOIs | |

State | Published - Dec 2005 |

Externally published | Yes |

### Fingerprint

### Keywords

- Causal inference
- Doubly robust estimation
- Longitudinal data
- Marginal structural model
- Missing data
- Semiparametrics

### ASJC Scopus subject areas

- Agricultural and Biological Sciences(all)
- Public Health, Environmental and Occupational Health
- Agricultural and Biological Sciences (miscellaneous)
- Applied Mathematics
- Statistics and Probability

### Cite this

*Biometrics*,

*61*(4), 962-972. https://doi.org/10.1111/j.1541-0420.2005.00377.x

**Doubly robust estimation in missing data and causal inference models.** / Bang, Heejung; Robins, James M.

Research output: Contribution to journal › Article

*Biometrics*, vol. 61, no. 4, pp. 962-972. https://doi.org/10.1111/j.1541-0420.2005.00377.x

}

TY - JOUR

T1 - Doubly robust estimation in missing data and causal inference models

AU - Bang, Heejung

AU - Robins, James M.

PY - 2005/12

Y1 - 2005/12

N2 - The goal of this article is to construct doubly robust (DR) estimators in ignorable missing data and causal inference models. In a missing data model, an estimator is DR if it remains consistent when either (but not necessarily both) a model for the missingness mechanism or a model for the distribution of the complete data is correctly specified. Because with observational data one can never be sure that either a missingness model or a complete data model is correct, perhaps the best that can be hoped for is to find a DR estimator. DR estimators, in contrast to standard likelihood-based or (nonaugmented) inverse probability-weighted estimators, give the analyst two chances, instead of only one, to make a valid inference. In a causal inference model, an estimator is DR if it remains consistent when either a model for the treatment assignment mechanism or a model for the distribution of the counterfactual data is correctly specified. Because with observational data one can never be sure that a model for the treatment assignment mechanism or a model for the counterfactual data is correct, inference based on DR estimators should improve upon previous approaches. Indeed, we present the results of simulation studies which demonstrate that the finite sample performance of DR estimators is as impressive as theory would predict. The proposed method is applied to a cardiovascular clinical trial.

AB - The goal of this article is to construct doubly robust (DR) estimators in ignorable missing data and causal inference models. In a missing data model, an estimator is DR if it remains consistent when either (but not necessarily both) a model for the missingness mechanism or a model for the distribution of the complete data is correctly specified. Because with observational data one can never be sure that either a missingness model or a complete data model is correct, perhaps the best that can be hoped for is to find a DR estimator. DR estimators, in contrast to standard likelihood-based or (nonaugmented) inverse probability-weighted estimators, give the analyst two chances, instead of only one, to make a valid inference. In a causal inference model, an estimator is DR if it remains consistent when either a model for the treatment assignment mechanism or a model for the distribution of the counterfactual data is correctly specified. Because with observational data one can never be sure that a model for the treatment assignment mechanism or a model for the counterfactual data is correct, inference based on DR estimators should improve upon previous approaches. Indeed, we present the results of simulation studies which demonstrate that the finite sample performance of DR estimators is as impressive as theory would predict. The proposed method is applied to a cardiovascular clinical trial.

KW - Causal inference

KW - Doubly robust estimation

KW - Longitudinal data

KW - Marginal structural model

KW - Missing data

KW - Semiparametrics

UR - http://www.scopus.com/inward/record.url?scp=33644851650&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33644851650&partnerID=8YFLogxK

U2 - 10.1111/j.1541-0420.2005.00377.x

DO - 10.1111/j.1541-0420.2005.00377.x

M3 - Article

C2 - 16401269

AN - SCOPUS:33644851650

VL - 61

SP - 962

EP - 972

JO - Biometrics

JF - Biometrics

SN - 0006-341X

IS - 4

ER -