Using specific research case studies I will give an overview as to how as researchers we can have a great research idea, grounded in the relevant literature, but there are problems translating it into a robust research design. Assuming that the area/question cannot be reliably researched using small but rich sample surveys I will present ways in which routine admin data can help, along with the additional challenges of creating the correct cohort to address the research question.
The examples are from the Scottish Longitudinal Study (SLS) which links together routinely collected administrative data for a 5.3% representative sample of the Scottish population (about 270,000 people). It includes a wealth of information from the censuses (1991-2011), vital events registrations (ie births and deaths), and education data from 2007 onwards. The SLS with appropriate permissions can also be linked to health data such as cancer registry and hospital admission data from the NHS in Scotland. The size and scope of the SLS make it an unparalleled resource for analysing a range of socio-economic, demographic and health questions.
I will demonstrate how despite the large number of study members owing to the constraints on various admin data being available centrally for Scotland in systems (ie health data and education data) cohorts have to be carefully considered in order to research outcomes (events/results). Examples include: (1) life-course events for a cohort of SLS women born 1959-1965 followed up from 1991, (2) setting up 2 complex cohorts of SLS members and children of the SLS (COTS) born from 1991 onwards to investigate child development including social status information from family background, and (3) constructing relevant cohort samples to investigate those not in employment, education or training (NEET).