Case Study E: The Policy Researcher
[This case study has been anonymised at the request of the interviewee, particularly because it contains potentially sensitive information regarding their opinions of and working relationship with statistical producers.]
Researcher E leads an analytical team at a relatively well-resourced third-sector organisation with strong, in-house technical expertise. Their team are primarily engaged in the production of social research with policy relevance and regularly make use of data relating to poverty, public economics, and living standards. They have strong relationships with government and their publications regularly influence national debate.
In the course of their policy work, Researcher E has to produce timely analyses of announced government policies at pace. The speed at which their outputs have to be produced is key to their organisation’s relevance and impact. Typical analyses include costing of economic policies or understanding the impact of policy changes on specific groups.
One activity they find particularly important is rapidly responding to fiscal events like the 2025 Autumn Budget in which the government announced that it would be terminating the two-child benefit limit. Modelling the impact of large proposed changes like that in the current system often has to take place after the announcements, but getting robust analysis out quickly is critical for promoting informed public debate.
Getting access to the most up-to-date and detailed data required for modelling these policy changes generally requires submitting applications to data owners like DWP, DfE, the UK Data Service, or others (depending on the desired analysis) proving that they have a reasonable business case for the requested data. Typically, these applications are necessary because the detailed analyses desired by Researcher E’s organisation make use of survey microdata or sometimes non-public administrative datasets. A burden for researchers like Researcher E is that, under the current system, applications have to be made repeatedly, even when the requested data are merely updated versions of previously used datasets and the proposed analyses are the same as previous applications.
Because of this friction in accessing data, people like Researcher E may have to wait for months before being able to robustly scrutinise proposed government policies. Their organisations also have to expend staff time and resources on duplicated efforts to access data, which could otherwise be spent on producing more thorough or innovative reports. In the worst instances, robust analyses cannot be completed before relevant political timelines expire.
Depending on the analysis, Researcher E may be reliant on accessing data through Freedom of Information (FOI) requests. Unfortunately, in their experience the quality of responses is inconsistent at best from most government departments.
They noted one instance when they requested data from the Cabinet Office that exemplifies some of the systemic issues here. In the example that they shared, they knew the requested dataset existed because of prior public releases and interactions with the Office. However, the data shared in response to the FOI request was of such poor quality and documentation that the originally intended analysis was not possible. Researcher E suspects that, given the discrepancy between the known dataset and what was received, there may have been errors in handling the request.
Additionally, Researcher E found that the Cabinet Office did not have staff who could support with queries about how to interpret the dataset. Generalising to their experience with various government departments, Researcher E ascribes the problem to a scarcity of statistically trained staff who can support with these requests.
To summarise, researchers in organisations like Researcher E’s are forced to spend their finite resources on repetitive administrative tasks just to access data necessary for their work. Frictions in their ability to access undermine their ability to produce research in the public interest in a timely fashion. In some cases, timely analysis is not possible, and in other cases there is inadequate support from the statistical system to enable important questions to be answered.