The RSS Data Manifesto sets out ten recommendations to the UK government on how it can improve data for the good of society.
David Spiegelhalter, former RSS president, chair of the Winton Centre for Risk and Evidence Communication and co-chair of the RSS Covid-19 Task Force, explains how we might address recommendation number 6: prepare for the data economy by skilling up the nation.
It is not only professionals that require data literacy – it is a basic requirement for informed citizens. But statistics education, whether in school, further education, university or for the general public, needs massive improvement. The ongoing Covid-19 pandemic where the public have been faced daily with new statistics and data, shows just how important improving data literacy is.
There is a continuing struggle to determine the best way to introduce ideas of data science into the school curriculum. Statistics has traditionally been seen as part of mathematics, but this is increasingly being seen as too narrow a perspective, and that data skills are required across the curriculum. It is also recognised that data skills are best taught within a specific context in which its relevance for solving problems can be emphasised. New Zealand has led the way in statistics education. It follows a model of being problem-driven – motivated by an interesting question where establishing a plan of investigation is vital – and it is not assumed that the available data can answer the question. A number of exciting initiatives are underway, for example the International Data Science in Schools Project.
University and further education courses often tend to teach basic mathematical statistics, following a traditional structure of summary statistics, probability theory, sampling distributions of statistics under an assumed probability model, then a bag of tests and procedures to apply in different circumstances. Again, restructuring as a problem-based approach seems appropriate, except perhaps for the more mathematical students. The American Statistical Association’s GAISE guidelines are full of good suggestions.
Education for the general public is more difficult, although we are fortunate to have programmes such as BBC Radio 4’s More or Less and some journalists have done a great job during the pandemic in helping the public understand the numbers. My experience is that many people are delighted to learn about the tricks people play when using statistics, and the questions that should be asked in order to call them out. The basic list I usually give is:
- Why am I hearing this? What is the motivation of the source, and what filters has it gone through?
- Is the number reliable? Could there be random or systematic biases?
- How trustworthy is the claim being made?
Again, the RSS and others all have their part to play in taking the message out to a larger audience. The RSS Statistical Ambassadors guide to the numbers on coronavirus
is well worth a read. I am also shameless enough to plug my book, The Art of Statistics
, which tries to exemplify all the approaches I’ve outlined in this blog.
Have a look at the other recommendations in our Data Manifesto