Using mixtures of t densities to make inferences in the presence of missing data with a small number of multiply imputed data sets
Using mixtures of t densities to make inferences in the presence of missing data with a small number of multiply imputed data sets
Strategies for making inference in the presence of missing data after conducting a Multiple Imputation (MI) procedure are considered. An approach which approximates the posterior distribution for parameters using a mixture of tt-distributions is proposed. Simulated experiments show this approach improves inferences in some aspects, making them more stable over repeated analysis and creating narrower bounds for certain common statistics of interest. Extensions to the existing literature have been executed that provide further stability to inferences and also a strong potential to identify ways to make the analysis procedure more flexible. The competing methods have been first compared using simulated data sets and then a real data set concerning analysis of the effect of breastfeeding duration on children’s cognitive ability. R code to implement the methods used is available as online supplementary material.
missing data, multiple imputation, bayesian statistics, disclosure avoidance, mixture distribution, monte carlo
84-96
Rashid, S.
ced9fe5b-c8c0-49e1-a946-e411535c011b
Mitra, R.
2b944cd7-5be8-4dd1-ab44-f8ada9a33405
Steele, R.J.
b7e08aec-9d93-4433-a2cb-e3b6c0db33c6
December 2015
Rashid, S.
ced9fe5b-c8c0-49e1-a946-e411535c011b
Mitra, R.
2b944cd7-5be8-4dd1-ab44-f8ada9a33405
Steele, R.J.
b7e08aec-9d93-4433-a2cb-e3b6c0db33c6
Rashid, S., Mitra, R. and Steele, R.J.
(2015)
Using mixtures of t densities to make inferences in the presence of missing data with a small number of multiply imputed data sets.
Computational Statistics & Data Analysis, 92, .
(doi:10.1016/j.csda.2015.05.009).
Abstract
Strategies for making inference in the presence of missing data after conducting a Multiple Imputation (MI) procedure are considered. An approach which approximates the posterior distribution for parameters using a mixture of tt-distributions is proposed. Simulated experiments show this approach improves inferences in some aspects, making them more stable over repeated analysis and creating narrower bounds for certain common statistics of interest. Extensions to the existing literature have been executed that provide further stability to inferences and also a strong potential to identify ways to make the analysis procedure more flexible. The competing methods have been first compared using simulated data sets and then a real data set concerning analysis of the effect of breastfeeding duration on children’s cognitive ability. R code to implement the methods used is available as online supplementary material.
This record has no associated files available for download.
More information
Accepted/In Press date: 29 May 2015
e-pub ahead of print date: 19 June 2015
Published date: December 2015
Keywords:
missing data, multiple imputation, bayesian statistics, disclosure avoidance, mixture distribution, monte carlo
Organisations:
Statistics
Identifiers
Local EPrints ID: 389853
URI: http://eprints.soton.ac.uk/id/eprint/389853
ISSN: 0167-9473
PURE UUID: 7bf7277c-d900-4732-9d2f-77faf78c0e6f
Catalogue record
Date deposited: 17 Mar 2016 10:12
Last modified: 14 Mar 2024 23:10
Export record
Altmetrics
Contributors
Author:
S. Rashid
Author:
R.J. Steele
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics