On 3 February 2015, the RSS Social Statistics Section held a meeting titled: ‘The ONS Longitudinal Study - 40 years old and going strong’. The meeting was held at the Royal Statistical Society, and was chaired by Professor Allan Findlay of the University of St Andrews.
Three speakers were engaged to explain the nature of the survey, some of the research that was possible from it and the support available for those wishing to use the data. The event was timely as it was a year into the new linked data being available to researchers.
Nicky Rogers from the Office for National Statistics (ONS) spoke first, and began by emphasising that the census longitudinal study (CLS) was not a survey but really a study. It now comprises five censuses (for England and Wales; three for other nations), with linked data for 1% of the population based on four birth days, spaced seasonally, using household census returns and vital registration. The total sample now is 500,000 but linked for all five waves is around 200,000. Fertility, deaths, cancer registrations (site and type), partner deaths and migration are recorded and persons new to the resident population born on the four days are added to the sample. Nicky closed by higlighting the research uses in large social policy questions around ageing and pensions such as the Dilnot report and Marmot review. Long term survival in communal establishments and changes in health inequalities being some of their results.
Dr Franz Buscha of the University of Westminister talked about research on intergenerational social mobility as an example of the potential use of the CLS. He praised the support available for users of the data, once rigorous security procedures about disclosure control had been completed. Social mobility analyses of cohort studies have inherent biases but are the main evidence for social mobility so the CLS provides a new way to consider a pressing social issue. While there is no income data in the census, the household data allows for understanding of occupational class and status (measured using CAMSIS). There are now three generations with twenty years (three waves) of data and Dr Buscha explained how they showed improvements in social mobility. Further possibilities for analysis are looking at social policy reforms such as the introduction of the GCSE.
Finally, Dr Nicola Shelton of UCL detailed the user support available - covering advice on research design as well as the application process through to approval of tables for publication. Census data is subject to strict disclosure controls and so release sometimes has to be negotiatied and training is provided to users on safe practices. She emphasised the great potential of the data because of its large sample size and the opportunity to be matched with a more experienced partner to do analysis.
Discussion revealed that the CLS is unusual in covering communal establishments which most household surveys do not. Even if a person was missed from a census, because four birth days are known to holders of administrative data, details of deaths etc would not be missing. Time changes in ethnicity specific mortality could be studied, as could migrant outcomes as year of arrival was recorded in 2011. Finally there could also be methodological advances by looking at changing patterns of social status and occupations over time in the whole population.