Opportunities for Cross Cohort Comparisons using UK Datasets: The Work of CLOSER

by Tomos Robinson

In the UK, researchers in the health and social sciences are privileged to have a wide variety of large, representative cohort and household panel studies to utilise. For instance, cohort studies such as the National Survey of Health and Development (1946) and the National Child Development Study (1958) have tracked a sample of individuals across their life course, while recent cohort studies such as the Avon Longitudinal Study of Parents and Children (1991) and the Millenium Cohort Study (2000) provide fantastic opportunities to analyse recent cohorts of children. Furthermore, the British Household Panel Survey (1991) (now known as ‘Understanding Society’) has followed a representative sample of individuals every year since 1991.

While each individual cohort/household survey is a fantastic resource for analysing a range of social, economic and health issues, the ESRC and MRC funded ‘Centre & Longitudinal Studies Enhancement Resources’ (CLOSER) has recently argued that in order to make full use of the cohort and longitudinal studies we have in the UK, researchers should be able to make comparisons between the different studies. As argued by CLOSER, the goal of such ‘cross cohort’ research could be either to test whether results are consistent across studies (as a form of sensitivity analysis), or to see how the results differ in different time periods and social conditions.

To aid the comparison of results across cohorts, CLOSER has launched a series of work packages in an attempt to harmonise various key measures, examples of which include: the harmonisation of body size and body composition; the harmonisation of socio-economic status and qualifications and the harmonisation of biomarker data. Studies utilising the newly harmonised measures are beginning to be published, for instance Johnson et al (2015) who utilised five British based cohort studies to examine the trajectory of obesity over time in the UK. CLOSER have also organised a series of workshops and conferences concerning cross cohort research, designed to both inform early career researchers of the opportunities and challenges that may be faced when doing cross-cohort research and give researchers the opportunity to present their research whilst considering the potential for cross-study comparisons.

I have been fortunate enough to attend one of the CLOSER workshops and, additionally, present my research at a CLOSER conference. In early September 2015 I attended a workshop entitled “Cross-cohort research: opportunities, challenges and example”, which opened my eyes to the range of variables available for comparison across the different cohort surveys, as well as the potential pitfalls in using such methods.

At the end of November, I was lucky enough to be selected to give a short oral presentation at the CLOSER conference entitled “The importance of early years, childhood and adolescence: Evidence form longitudinal studies”, held at the British Library conference centre in central London. The presentation concerned some work from my ongoing PhD research, which compares socioeconomic inequalities in cognitive ability in the National Child Development Study (1958) and the UK Millennium Cohort Study (2000/1). Once more, the experience was extremely beneficial. As well as allowing me an opportunity to practice my presenting skills, it was valuable to receive feedback on my research from an inter-disciplinary audience consisting of a variety of academics from across the social sciences.

Overall, I think that the work being undertaken by CLOSER and its affiliates is exciting for researchers across the health and social sciences using UK based data. Although harmonising variables may be a challenging process, better access to the combined variables from a variety of different cohort and panel surveys could have significant, positive implications for research. As well as potentially improving the robustness of results from work using single data sources, the CLOSER harmonised variables may also be able to increase the understanding of the ways that different institutional contexts and time periods affect the specific research findings.