Workshop details

Attendees learnt why principled methods of missing data handling are usually required to reduce bias and restore sample representativeness in long-running cohort studies and how to undertake such analyses in practice. Analyses was demonstrated using Stata, with a focus on multiple imputation. The National Child Development Study (NCDS) was used throughout as a case study. There was plenty of opportunity for questions.

Learning objectives

1. Understand why principled methods of missing data handling are usually required to obtain unbiased estimates in long-running cohort studies.

2. Understand why and how variables other than those included in the analyses should be included in principled methods of missing data handling.

3. Have gained familiarity with examples in which principled methods of missing data handling have been applied in NCDS.

4. Have observed demonstrations of applying such methods in example applications using data from NCDS (focusing on the use of multiple imputation in Stata).

Presenters

This workshop was run by staff from the UCL Centre for Longitudinal Studies:

• Prof George Ploubidis, Professor of Population Health and Statistics, Director of Research and Chief Statistician
• Dr Richard Silverwood, Associate Professor of Statistics
• Dr Martina Narayanan, Research Associate
• Brian Dodgeon, Research Fellow