ADR UK Research Fellowships: £200k to Unlock Public Data

Data is the new oil, but in the UK public sector, most of it is still trapped underground. The government holds vast amounts of data—on schools, courts, hospitals, and tax returns—that could solve massive social problems. But due to privacy concerns and bureaucratic silos, researchers rarely get to see it.

Administrative Data Research UK (ADR UK) is changing that. They have built secure pipelines to link these datasets together. Now, they need brilliant researchers to mine them for insights.

The ADR UK Research Fellowships 2025 offer a massive £200,000 grant (at 100% Full Economic Cost) for researchers to spend 18 months diving deep into these “Flagship Datasets.”

This is not just about writing a paper for a journal that five people will read. This is about “Public Good.” It is about finding out why some kids get expelled from school, or what happens to prisoners after they are released, or how the cost of living crisis is affecting mental health. It is research that lands on a Minister’s desk.

Key Details at a Glance

What This Opportunity Offers

Access to “Gold Dust” Data You get access to linked administrative data that is usually impossible to get. For example:

• Data First: Linked data from the Ministry of Justice (Courts, Prisons, Probation).
• ECHILD: Linked data between Hospital Episode Statistics and the National Pupil Database.
• Wage and Employment Dynamics: Linked tax and benefits data. This allows you to answer longitudinal questions that surveys simply can’t touch.

Full Economic Cost (FEC) Unlike many charities that only pay “direct costs,” this is a UKRI grant. It pays 100% of the Full Economic Cost. This means it covers your salary, your university’s overheads (estates, indirect costs), and your research expenses. It is a “profitable” grant for your department.

Policy Impact ADR UK has a direct line to Whitehall. As a fellow, you are not just a researcher; you are a “Policy Advisor.” You will be supported to write policy briefs, present to government departments, and engage with the “What Works Centres.” If you want your research to change the law, this is the vehicle.

Who Should Apply

This is for quantitative social scientists who know how to handle big, messy data.

Ideal Candidates:

• The Criminologist: You want to track the “revolving door” of reoffending using the Data First dataset.
• The Health Economist: You want to analyze the long-term earnings of people who had childhood surgeries using ECHILD.
• The Education Researcher: You want to see how school exclusions link to later benefit claims.

Eligibility Checklist:

• Institution: You must be based at a UK “Research Organisation” (University or Independent Research Institute) that is eligible for ESRC funding.
• Experience: You can be an Early Career Researcher (ECR) or a Senior Professor. The panel assesses you relative to your career stage.
• Skills: You must be an “Accredited Researcher” (AR) under the Digital Economy Act (or be willing to become one). You need strong coding skills (R, Python, Stata).

Insider Tips for a Winning Application

I have seen the feedback from ESRC panels. Here is what they kill applications for.

1. The “Public Good” Test This is the most important criterion. You must explicitly state how your project serves the public good. “I am curious about X” is not enough. “Understanding X will help the Department for Education reduce exclusions by 10%” is what they want.

2. Feasibility is King Administrative data is messy. It has missing variables. It has coding errors. If you promise to do a complex machine learning model on a dataset you have never seen, the panel will doubt you.

• Tip: Read the “User Guide” and “Data Dictionary” for the dataset before you write the proposal. Quote the variable names. Show you know the data limitations.

3. Stakeholder Engagement Don’t leave “Impact” for the last month. Build it in from Day 1.

• Tip: Get a “Letter of Support” from a relevant charity or government unit. If you are studying prisons, get a letter from the Prison Reform Trust. It shows your research has an audience.

4. The “Safe People” Check You will be working with highly sensitive data (de-identified, but still sensitive). The panel needs to trust you. Highlight your experience with “Safe Pods” or “Secure Research Environments” (SREs). If you are already an Accredited Researcher, put that in bold.

5. Value for Money £200k is a lot. Justify it. Why do you need 18 months? Why do you need a Research Assistant? If you pad the budget, they will cut it.

Application Timeline

December 2025: The Idea

• Action: Browse the ADR UK Data Catalogue. Pick your dataset.
• Action: Check if you need “pre-approval” to access the data. Some datasets require a “Research Accreditation Panel” (RAP) application before you apply for the funding.

January 2026: The Draft

• Action: Write the “Case for Support.” This is the main 6-page document.
• Action: Talk to your university’s Research Office. They need time to calculate the “fEC” (Full Economic Cost). Do not email them 2 days before the deadline.

February 1-15, 2026: The Polish

• Action: Get a colleague to read it. Ask them: “Is the methodology clear? Is the impact convincing?”
• Action: Finalize your “Data Management Plan.” How will you handle the code?

February 26, 2026: Submission

• Action: Submit via the UKRI Funding Service (TFS). Note: Your Research Office has to hit the final “Submit” button. You usually need to send it to them 5 days early.

Required Materials

• Vision and Approach: The core proposal (methodology, objectives).
• Applicant Capability: Your CV and track record (narrative format).
• Resources and Cost Justification: The budget explanation.
• Data Management Plan: How you will work in the Secure Research Environment (SRE).
• Ethics Assessment: How you will handle ethical risks.

What Makes an Application Stand Out

Novel Linkage Proposing to link two datasets that have never been linked before (e.g., linking “Homelessness Data” with “Hospital Data”) is high risk but high reward. If you can pull it off, it’s a winner.

Capacity Building ADR UK wants to build a community. If you propose to “open source” your cleaning code or run a workshop for other researchers on how to use the data, you get bonus points.

Clear Policy Hook “This research directly addresses the question raised in the 2024 Green Paper on Social Care.” Connecting your work to a specific government inquiry or strategy makes it “timely.”

Common Mistakes to Avoid

Assuming Data Access is Instant It is not. Getting access to the ONS Secure Research Service (SRS) can take months. Build this delay into your Gantt chart. If you say “I will start analysis on Day 1,” the panel knows you are naive.

Ignoring the “Five Safes” The UK data framework is built on the “Five Safes” (Safe People, Safe Projects, Safe Settings, Safe Data, Safe Outputs). Mention this framework in your proposal. It shows you speak the language of data governance.

Over-Complicating the Stats Sometimes a simple regression is better than a neural network. Don’t use fancy math just to impress. Use the right tool for the job.

Frequently Asked Questions

Can I use my own data? You can link your own data to the ADR datasets, but the core of the project must be the ADR flagship data.

Can I work from home? Yes, if you have an “Assured Connection” to the ONS Secure Research Service. Many universities have this. If not, you might have to go to a physical “Safe Pod.”

Is this open to PhD students? No. This is a fellowship for staff. However, you can hire a PhD student as a Research Assistant if the budget allows.

What happens if the data quality is bad? That is a risk. You should have a “Plan B.” “If variable X is missing, I will use variable Y as a proxy.”

How to Apply

1. Log in to TFS: Go to the UKRI Funding Service.
2. Find the Opportunity: Search for “ADR UK Research Fellowships 2025.”
3. Start the Application: Fill in the sections.
4. Submit to RO: Send it to your Research Office for approval.

The data is there. The money is there. The only thing missing is the question. What will you ask?