The RSS Data Manifesto: Statistics training for policymakers

The RSS Data Manifesto sets out ten recommendations to the UK government on how it can improve data for the good of society.

Former RSS president David Spiegelhalter, Winton Professor of the Public Understanding of Risk at the University of Cambridge, explains why we need more data-savvy policymakers, as we call for in recommendation number 3: Champion basic training in data handling and statistics for politicians, policymakers and other professionals working in public services.

There are two components to what might be called data literacy. First, recognising the opportunities, and also the challenges, of using data in the whole cycle of problem solving – this involves skills in planning what might be useful and available, collecting and cleaning data, carrying out exploratory and confirmatory analysis, and finally communicating conclusions to interested parties. Second, even if not conducting the investigation oneself, it is important to understand the process and its limitations, and so be able to critique the claims being made on the basis of data analysis, whether they be scientific papers, official reports, or political or media claims.

Many professionals already have considerable skills in the appropriate use of data-based evidence for making better decisions, although there is a pressing need for better training and support, particularly as the growing number of people employed as ‘data scientists’ still lack a clear professional structure and skill inventory. There are various groups of interest:  

  • Science and medical professionals should already get some statistical training, but increasing focus on improving the quality of research, and hence the reproducibility of scientific claims, has identified that much better statistical standards are necessary. The UKRI has a strong role to play in establishing better practice across all the research councils, although it is challenging to improve statistical peer review in scientific journals when the number of willing volunteers is necessarily limited: some form of paid statistical review may be reasonable.
  • Politicians, and their aides, often appear woefully ignorant of good statistical practice, seeing numbers as rhetorical devices to persuade people to support their views, rather than to provide a basis for reasoning between alternative policies. This group is difficult to access, although the RSS has worked hard to get politicians to agree to have statistical training.  The statistical ‘stick’ of the UK Statistics Authority, writing critical letters to put politicians and their advisers on the ‘naughty step’, also needs to be continued with vigour.
  • Civil servants require strong standards in both aspects of data literacy, whether statisticians and others who conduct the analyses and communicate the results, or the policy profession who need to be able to critique and compare analyses comparing alternative policies. ONS have been making great improvements, with the instigation of their data campus, personalized narratives from statisticians, and more attention to communicating uncertainty.
  • Legal professionals require training in ‘forensic statistics’, to allow them to adequately address evidence coming from increasingly sophisticated forensic science. The RSS Statistics and the Law Section has always played a prominent role in this area, and the Winton Centre for Risk and Evidence Communication in Cambridge is about to embark on a major collaboration with the Leverhulme Research Centre for Forensic Science at Dundee and Northumbria Law School to produce flexible online training materials.
  • Journalists and press officers are requiring increased statistical skills, exemplified by BBC News now having a Head of Statistics (who is also an RSS Ambassador). There is some training organised by the Science Media Centre and provided by RSS Fellows, but there is strong demand for more.

The Royal Statistical Society has a strong role to play in all these endeavours, including establishing professional standards within data science, providing training courses, working through its Sections such as the Statistics and the Law Section, and encouraging statistical ambassadors and others to help the media, such as working on the Science Media Centre’s Before the Headlines process. These are all vital activities in improving data literacy.

Load more