Handling Missing Data in Longitudinal Surveys
Dates: 13-15th December 2011
Duration: 3 days (10am — 4:30pm)
Level: Advanced
Course Fee: £525 (£375 for those from educational institutions)
CCSR offers 5 free places to research staff and students within the Faculty of Humanities at the University of Manchester and the North West Doctoral Training Centre.
Course Leader:
Ian Plewis and Jonathan Bartlett
Course Requirements:
Participants should have some familiarity with longitudinal data, a working knowledge of STATA and a good understanding of regression and logistic regression.
Course Summary
The first day of the course will provide an overview of methods for adjusting for missing data, descriptions of missing data and their predictors in the UK birth cohort studies, and introductions to the datasets to be used in the practical sessions on days two and three. The practical work will focus on multiple imputation, although weighting methods and joint models for the process of interest and the missingness mechanism will also be discussed. All the techniques will be demonstrated using real data from the UK birth cohort studies.
Course Objectives
The course will:
- Introduce participants to patterns and correlates of missing data in the UK birth cohort studies
- Introduce the various methods proposed to adjust for missing data
- Describe the different ways of doing multiple imputation and introduce relevant software for carrying out multiple imputation for longitudinal data
- Examine the advantages and disadvantages of inverse probability weighting to adjust for non-response.
- Give a brief introduction to Bayesian methods of adjustment and to ‘Heckman’ selection models.
Target Audience
The course is designed for users of longitudinal data in general and the UK birth cohort studies in particular. It would be particularly appropriate for those who want to extend their skills as analysts of longitudinal data to take account, in a statistically principled way, of the ubiquitous problems of missing data, arising both as a result of attrition and item non-response.
Preliminary Reading
- Little, R. J. A. and Rubin, D. B. (2002) Statistical Analysis with Missing Data (2nd. Ed.) (Ch. 1). New York: Wiley.
- Plewis, I. et al. (2004) National Child Development Study and 1970 British Cohort Study Technical Report. London: Centre for Longitudinal Studies.
- Plewis, I. (Ed.) (2007) The Millennium Cohort Study. Technical Report on Sampling (4th. Ed.). London: Centre for Longitudinal Studies.
For information about the birth cohort studies (used in the practical sessions), see the CLS web site: http://www.cls.ioe.ac.uk/default.asp