BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Congenial multiple imputation of partially observed covariates wit
 hin the full conditional specification framework - Jonathan Bartlett\, LSH
 TM.
DTSTART:20120515T133000Z
DTEND:20120515T143000Z
UID:TALK37217@talks.cam.ac.uk
CONTACT:Dr Jack Bowden
DESCRIPTION:Missing covariate data is a common issue in epidemiological an
 d clinical research\, and is often dealt with using multiple imputation (M
 I). When the analysis model is non-linear\, or contains non-linear (e.g. s
 quared) or interaction terms\, this complicates the imputation of covariat
 es. Standard software implementations of MI typically impute covariates fr
 om models that are uncongenial with such analysis models. We show how impu
 tation by full conditional specification\, a popular approach for performi
 ng MI\, can be modified so that covariates are imputed from a model which 
 is congenial with the analysis model. We investigate through simulation th
 e performance of this proposal\, and compare it to passive imputation of n
 on-linear or interaction terms and the `just another variable’ approach.
  Our proposed approach provides consistent estimates provided the imputati
 on models and analysis models are correctly specified and data are missing
  at random. In contrast\, passive imputation of non-linear or interaction 
 terms generally results in inconsistent estimates of the parameters of the
  model of interest\, while the `just another variable' approach gives cons
 istent results only for linear models and only if data are missing complet
 ely at random. Furthermore\, simulation results suggest that even under im
 putation model mis-specification our proposed approach gives estimates whi
 ch are substantially less biased than estimates based on passive imputatio
 n. The proposed approach is illustrated using data from the National Child
  Development Survey in which the analysis model contains both non-linear a
 nd interaction terms.
LOCATION:Large  Seminar Room\, 1st Floor\, Institute of Public Health\, Un
 iversity Forvie Site\, Robinson Way\, Cambridge
END:VEVENT
END:VCALENDAR