This BETA site version 1.3 is subject to ongoing testing and improvements. For more information, or to provide feedback or report issues, please contact us .
The Centre for Longitudinal Studies runs four longstanding British cohort studies, born in 1958 (the National Child Development Study, NCDS), 1970 (the British Cohort Study, BCS70), 1989/19990 (Next Steps) and 2000/2002 (the Millennium Cohort Study, MCS), respectively. The large, representative sample sizes of each of these studies (initial samples exceeding 15,000 in each cohort) make them useful data sources for analyses of specific subgroups in the general population.
Variable and Category Selection: tabulates the number of cases in each study by survey sweep for population subgroups. Users can combine subgroups within a domain (e.g., "Homosexual" and "Bisexual" in the "Sexual Orientation" domain) to aggregate sample sizes, if they so wish.
Age Range Selection: for particularly small subgroups, users may wish to combine data across cohorts and within a cohort across sweeps (e.g., defining disability as ever reporting being disabled in adult sweeps). This tab provides functionality to calculate sample sizes within each domain, combining data from multiple cohorts and from sweep within a specified age range.
In the Age Range Selection tab, sample sizes are calculated by summing the number of cases in each cohort across the sweeps that took place within the age range specified by the users. In this calculation, an individual is defined as a case if they recorded the specific category for the selected characteristic during that time period. This means that a given individual can be counted multiple times if they meet the criteria across different sweeps - for instance, a person who lived in England in the age 23y sweep of the NCDS but in Scotland at the age 33y sweep will be counted in both categories for the sample size calculation. Given this possibility, the sample sizes in this tab can exceed the total number of individuals in a given cohort.
Some variables have few observations in specific categories. For data sharing reasons, we cannot share the exact cell counts in these cases. Instead, we partially mask the true value as < 10, and provided ranges for any aggregate values based upon these.
The characteristics included on this website are not always measured in precisely the same way across the four cohorts or within a cohort, in each survey sweep. The code used to clean the data can be viewed on the GitHub repository page.
Sex and ethnicity are treated as time invariant variables. Both are collected on multiple occasions in each cohort. We use the first recorded observation per individual, regardless of the sweep when it was recorded. One exception to this is that we use pre-cleaned 'longitudinal' sex (BCS70) and sex and ethnicity (NCDS) where these were available. The sample size calculations in the Age Range Selection and Variable and Category Selection tabs for these variables therefore represent the number of individuals at a given sweep (or at sweeps in a given age range) who have the particular characteristic. This differs slightly from the treatment of the time variant variables, where it is the number of individuals with that characteristic collected at a given sweep (or in any of the sweeps in a given age range).
Cross-tabulations are not possible due to data sharing rules. However, researchers can access CLS cohort data through the UK Data Service. Please see the CLS data access website for more details.
If you would like to see another characteristic included on this website, please submit an issue on GitHub. Questions can otherwise be directed to Liam Wright.
