Innovative approaches to methodological challenges facing ageing cohort studies

Ageing cohort studies around the world face common methodological challenges of data collection, measurement and analysis, which become increasingly problematic as participants grow older. While these challenges are common to all longitudinal studies, ageing cohort studies in particular highlight complex methodological issues due to the nature of the population. The National Centre for Research Methods (NCRM) funded a series of workshops during 2012 that brought together experts and researchers in longitudinal and ageing cohort studies to discuss some of these methodological challenges. The series was divided into workshops around the challenges of data collection, measurement and analysis in ageing cohort studies. The workshops and international conference brought together over 150 researchers working primarily in the social and medical sciences. Within the social sciences, social statistics and sociology were the main disciplines represented; while within the medical sciences, epidemiology and gerontology were the main disciplines. Principal investigators and researchers from a number of ageing cohort studies attended the events, including the 1946, 1958, 1970 British Birth Cohort studies, English Longitudinal Study of Ageing (ELSA), Whitehall II study, the Cognitive Function and Ageing Studies, the Gas and Electricity Workers cohort study (GAZEL) and the Lothian birth cohort. There were presentations from early career researchers at all the events. These workshops were successful in generating discussions about the methodological contributions from different disciplines on the challenges related to data collection and analysis for ageing cohort studies. The seminar series was responsive to suggestions from attendees, for example, the theme of workshop 2 emerged from discussions at workshop 1. The themes of the conference and related training workshops emerged from the evaluations from previous workshops. Evaluations for each event were very positive, with a strong appreciation for the training workshops. Attendees were very appreciative of the provision of STATA .do files for complex analyses and they indicated a strong preference for more training events on the topics of missing data and life course analysis.

This report summarises some of the work underlying the workshops and highlights some of the innovative solutions researchers have adopted to overcome these challenges. These include using mixed modes of data collection to deal with respondent burden, using the Life Grid history method to deal with recall bias for proxy respondents, using auxiliary variables to adjust for ‘missing not at random’ mechanisms, and using a range of missing data analysis methods and simulation studies to assess the performance of a different analytical strategies to deal with missing data.