Today, OSR has published its first report on presenting uncertainty around estimates, a fundamental aspect in the appropriate use and interpretation of statistics.
OSR’s work considers aspects of the cause and effect of uncertainty in statistics and highlights good practice that already exists and is intended to be part of an ongoing series of outputs from OSR’s Insight programme. This will add to the wide and increasing range of advice already available to help producers think about how to best present uncertainty.
Depicting uncertainty can be challenging for producers as they need to think about many aspects such as different audiences, how the statistics will be used, as well as ensure they are working within accessibility guidelines. It is evident that conveying uncertainty in data tables is particularly challenging.
OSR found that uncertainty is typically better depicted and described in statistical bulletins and methodological documents than it is in data tables, data dashboards and downloadable datasets. There has been a rise in the use of infographics and other ways of visualising uncertainty and OSR saw great examples from some producers of where this has been done well. But, OSR feels that more needs to be done in this space as it can be particularly useful when producers show uncertainty in a graphical form.
Ed Humpherson, DG Regulation said:
OSR will do more to promote and support good practice around presenting uncertainty and consider what this means for our regulatory work. We recognise that this is an area where practice is developing all the time. We will focus on helping to share that good practice, enabling producers to learn from each other and improve the presentation of uncertainty.
During OSR’s research for this report, we were honoured to collaborate with the expert Professor and author, Sir David Spiegelhalter OBE FRS, an expert on communicating risk and uncertainty.
Professor David Spiegelhalter said:
I am delighted that OSR has focussed on the vital issue of communicating uncertainty about claims based on data, whether using graphics, numerical intervals or acknowledging deeper uncertainty. Presenting statistics with a gripping narrative is essential to gain attention and confidence, but must be carefully balanced with judgements about the ability of the data to answer the relevant questions.