The jointly developed ‘Evidence Transparency Framework’ asks whether someone outside of government can tell what the government is proposing to do, and why. The report shows where transparency about the use of evidence is being achieved and warns against practices that obscure it. The findings include:
- Departments often don’t share the research and reviews they have conducted to inform a policy, making it difficult for the public to understand the rationale.
- Manifesto-derived and values-based policies can be transparent. They were found to be among the highest and lowest scoring policies. Government seems to be more transparent when developing policies to deliver promised outcomes (ends) and less transparent when promises have been made about specific measures to achieve them (means).
- The best proposals demonstrated the chain of reasoning as to what the problem was and why the policy was the chosen response, and included discussion about the limitations of the evidence.
- Policies announced in the Budget and Autumn Statement were less transparent about the underlying evidence than other policies.
Researchers counted 593 policy announcements between May 2015 and May 2016 from 13 domestic-facing government departments. They scored a sample of these against the Evidence Transparency Framework, asking ‘Can we tell what evidence has been used? Can we tell how the government has assessed or used this evidence?’ in each of the following:
- Diagnosis (the issue the policy is designed to address)
- Proposal (the government’s chosen intervention)
- Implementation (how the intervention will be introduced and run)
- Testing and evaluation (plans to assess whether the policy has worked)
Today’s report highlights good practice and warns what needs to change before a full review in 2017 that will see departments scored and ranked on the extent of their transparency about the evidence they are using.
Join us (it’s free and open to all) and find out more about the how we champion the use of evidence in social policy and practice.