Home > Research > Publications & Outputs > Are independent parameter draws necessary for m...
View graph of relations

Are independent parameter draws necessary for multiple imputation?

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Published

Standard

Are independent parameter draws necessary for multiple imputation? / Hu, Jingchen; Mitra, Robin; Reiter, Jerome P.
In: American Statistician, Vol. 67, No. 3, 2013, p. 143-149.

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Harvard

Hu, J, Mitra, R & Reiter, JP 2013, 'Are independent parameter draws necessary for multiple imputation?', American Statistician, vol. 67, no. 3, pp. 143-149. https://doi.org/10.1080/00031305.2013.821953

APA

Vancouver

Hu J, Mitra R, Reiter JP. Are independent parameter draws necessary for multiple imputation? American Statistician. 2013;67(3):143-149. Epub 2013 Jul 12. doi: 10.1080/00031305.2013.821953

Author

Hu, Jingchen ; Mitra, Robin ; Reiter, Jerome P. / Are independent parameter draws necessary for multiple imputation?. In: American Statistician. 2013 ; Vol. 67, No. 3. pp. 143-149.

Bibtex

@article{6370faee70c2483e9907b934ee0b366d,
title = "Are independent parameter draws necessary for multiple imputation?",
abstract = "In typical implementations of multiple imputation for missing data, analysts create m completed datasets based on approximately independent draws of imputation model parameters. We use theoretical arguments and simulations to show that, provided m is large, the use of independent draws is not necessary. In fact, appropriate use of dependent draws can improve precision relative to the use of independent draws. It also eliminates the sometimes difficult task of obtaining independent draws; for example, in fully Bayesian imputation models based on MCMC, analysts can avoid the search for a subsampling interval that ensures approximately independent draws for all parameters. We illustrate the use of dependent draws in multiple imputation with a study of the effect of breast feeding on children{\textquoteright}s later cognitive abilities.",
keywords = "Bayesian, Missing, Nonresponse, Survey",
author = "Jingchen Hu and Robin Mitra and Reiter, {Jerome P.}",
year = "2013",
doi = "10.1080/00031305.2013.821953",
language = "English",
volume = "67",
pages = "143--149",
journal = "American Statistician",
issn = "0003-1305",
publisher = "American Statistical Association",
number = "3",

}

RIS

TY - JOUR

T1 - Are independent parameter draws necessary for multiple imputation?

AU - Hu, Jingchen

AU - Mitra, Robin

AU - Reiter, Jerome P.

PY - 2013

Y1 - 2013

N2 - In typical implementations of multiple imputation for missing data, analysts create m completed datasets based on approximately independent draws of imputation model parameters. We use theoretical arguments and simulations to show that, provided m is large, the use of independent draws is not necessary. In fact, appropriate use of dependent draws can improve precision relative to the use of independent draws. It also eliminates the sometimes difficult task of obtaining independent draws; for example, in fully Bayesian imputation models based on MCMC, analysts can avoid the search for a subsampling interval that ensures approximately independent draws for all parameters. We illustrate the use of dependent draws in multiple imputation with a study of the effect of breast feeding on children’s later cognitive abilities.

AB - In typical implementations of multiple imputation for missing data, analysts create m completed datasets based on approximately independent draws of imputation model parameters. We use theoretical arguments and simulations to show that, provided m is large, the use of independent draws is not necessary. In fact, appropriate use of dependent draws can improve precision relative to the use of independent draws. It also eliminates the sometimes difficult task of obtaining independent draws; for example, in fully Bayesian imputation models based on MCMC, analysts can avoid the search for a subsampling interval that ensures approximately independent draws for all parameters. We illustrate the use of dependent draws in multiple imputation with a study of the effect of breast feeding on children’s later cognitive abilities.

KW - Bayesian

KW - Missing

KW - Nonresponse

KW - Survey

U2 - 10.1080/00031305.2013.821953

DO - 10.1080/00031305.2013.821953

M3 - Journal article

VL - 67

SP - 143

EP - 149

JO - American Statistician

JF - American Statistician

SN - 0003-1305

IS - 3

ER -