GSS methodology research

What is the Administrative Data Methods Research Programme?

The Administrative Data Methods Research Programme is a new programme led by the ONS’ Methodology Division.

Why was the Administrative Data Methods Research Programme set up?

Across the GSS there is increasing use of administrative data in the production of Official Statistics. The UK Statistics Authority has been encouraging public bodies to fully utilise administrative data for statistical purposes. As part of the ONS’ transformation, we have ambitious plans to put administrative data at the core of our statistical production and are increasing their use.

Unlike survey data, however, there is not yet a well-established and agreed statistical theory to underpin the use of administrative data in Official Statistics. In 2018, David Hand published a paper that explains that there are a range of statistical challenges associated with administrative data which need to be investigated.

What are the aims of the Administrative Data Methods Research Programme?

Our aims are to:

• address the key methodological and statistical challenges of using administrative and transactional data
• collaborate on the cross cutting methodological challenge across the GSS, academia and internationally
• create a framework of methods to effectively utilise administrative data, maximising its potential
• catalogue and share best practice
• ensure the Administrative Data Methods Research Programme has a clear impact and application on Official Statistics within the ONS, GSS and beyond
• help to ensure robustness of statistical outputs and help put administrative data at the core of statistics within ONS

How will the Administrative Data Methods Research Programme achieve these aims?

We are working collaboratively across government, academia and other research organisations. We recognise that there is already existing research in these topics. Thorough research will be undertaken to ensure we are building on existing knowledge. This is a long-term investment programme, with projects started in May 2019.

If you wish to collaborate with us and would like an informal chat around future projects and wider research please email Methods.Research@ons.gov.uk.

The research programme has a number of projects underway, working in collaboration with colleagues across the GSS and beyond.

Developing standard tools for administrative data pre-processing and linkage

The first part of this project has been completed. Pre-processing tools have been created to provide metrics on the comparability of potential links and the utility of match key variables. Linkage tools to calculate match probabilities have been developed as part of this project. The next step is to carry out more testing under real world conditions and investigating the possibility of developing additional tools to enhance those already developed.

Estimation and communication of errors in non-survey sources

The aim of this project is to develop methods to understand and communicate sources of error produced in administrative, transactional or integrated data sources. This project will complete March 2021 and is a collaborative effort between the ONS and the University of Southampton.

Investigating quality frameworks

This project sets out to understand common practices from data quality frameworks and quality indicators in this field and produce case studies and guidance in this area for use in the ONS and the wider GSS.

Estimating vehicle and pedestrian activity from town and city traffic cameras

CCTV traffic cameras are present in most town and city centres across the UK. In this collaboration between ONS Methodology and Data Science Campus we applied machine vision algorithms to ‘count’ people and vehicles across a number of urban areas in near-real time. The resulting indicators published in Faster Indicators give a revealing insight into how busy the streets are and how the levels of activity change as social distancing policies evolve during the pandemic response.

Using qualitative methods to improve Admin data

ONS Methodology Quality and Data Collection Methods (QDCM), Data Science, and Sample Design and Estimation expert groups recently worked with the Valuation Office Agency (VOA), to explore ways of improving their existing data collection of Lettings Information. Building on an existing review completed by VOA, QDCM were invited to explore the human aspects of the collection process, with the aim of findings ways to improve the overall quality of the end-data. Analysis found some common themes spanning the data collection journey and we recommended a range of small cost-effective improvements to the process which could improve the overall quality of Lettings Information.

Economic and Social Research Council

The Administrative Data Methods Research programme is collaborating with academics across the UK to deliver a number of research projects funded by the Economic and Social Research Council.

PhD students

The Administrative Data Methods Research programme is also sponsoring a number of PhD projects with various universities.