File(s) under permanent embargo
Missing data: the importance and impact of missing data from clinical research
journal contribution
posted on 2023-05-18, 03:17 authored by Christine PadgettChristine Padgett, Skilbeck, CE, Mathew SummersMathew SummersThere is compelling evidence that traditional methods used to address the detrimental impacts of missing data are inadequate. Despite this, researchers have been slow to utilise newer statistical approaches known to be more effective. The aim of the current article is to offer a conceptual explanation of the rationale for using newer missing data techniques, with a focus on multiple imputation (MI). To illustrate the relative efficacy of deletion, single imputation and multiple imputation techniques in the clinical setting, 20 cases were selected randomly from a population study investigating the cognitive sequelae of traumatic brain injury (TBI), and 8 out of 20 cases had scores on one variable deleted to simulate a missing data set. Comparing the parameter estimates obtained by each technique to the known parameters of the complete data set revealed that MI outperformed deletion and single imputation approaches. It is therefore recommended that more sophisticated techniques such as MI should be considered in clinical research.
History
Publication title
Brain ImpairmentVolume
15Pagination
1-9ISSN
1839-5252Department/School
School of Psychological SciencesPublisher
Cambridge University PressPlace of publication
United KingdomRights statement
Copyright 2014 Australasian Society for the Study of Brain ImpairmentRepository Status
- Restricted