Deriving variables that predict missing data in the CLS cohort studies: NCDS, BCS70, Next Steps, and MCS

Centre for Longitudinal Studies

---

Overview

• This repository provides Stata do files that derive variables that predict non-response or missing data in each of the CLS cohort studies: NCDS, BCS70, Next Steps, and MCS. These variables have been identified in an extensive programme of work that informs users on how best to deal with missing data in longitudinal cohort studies.
• The handling missing data in the CLS cohort studies user guide provides full information on how the variables predicting missing data have been identified and how they should be used in analysis of cohort data.

---

Syntax and data availability

• Source data: Download raw data from each of the cohort studies from the UK Data Service. For each cohort study, data files for all study sweeps are needed for deriving the final dataset of variables predicting missing data. The do files specify the files needed. Place the raw data files in a folder on your computer.
• Syntax: Each cohort study has its own do file that reads the raw data and generates a single dataset for the given cohort. Note that raw cohort study data is occassionally updated with the UK Data Service, and therefore the code may not run seamlessly on future versions of data. Users would therefore need to make small adjustments to the code to accommodate.

---

User feedback and future plans

We welcome user feedback. Please open an issue on GitHub or email [email protected].

Authors

• Aase Villadsen

Licence

Code: MIT Licence (see LICENSE).

---

© 2025 UCL Centre for Longitudinal Studies