- This event has passed.
National Institute for Applied Statistics Research Australia Seminar Series
March 19 @ 2:30 pm - 3:30 pm
Dr Michael David
University of Newcastle
A preventative approach for minimising the impact of attrition in longitudinal studies
The use of longitudinal data has become increasingly important in a number of research fields, especially in the medical and health sciences. This reliance on longitudinal data has been accompanied by the identification of a number of design and analysis issues, one being attrition. Like death and taxes, its impacts are unavoidable, irrespective of setting, study design or data collection mode. In addition to decreasing power, its presence generally negates generalisability and produces biased estimates. As a consequence, numerous strategies emanating from two distinct approaches have and continue to be implemented so as to minimise these impacts. From a temporal perspective, those strategies implemented prior and during data collection could be labelled as preventative. The second approach can be considered curative and includes a suite of statistical techniques that are applied during the analysis stage; the most notable of being multiple imputation. Whilst the research literature regarding preventive strategies is substantial, in the main comparative assessments of these strategies have not instigated full economic evaluations and tended to include only the most commonly used strategies and their variants, such as repeat contacts and incentives. This presentation is a response to this situation, and as such provides an overview of research that I have undertaken to this point of time.