Statistics in biomedical and health research
Some reading to get you thinking
Contents
Statistical thinking
At the time of writing in October 2020, we remain in the grip of the covid-19 pandemic that has affected the world so badly for much of the year. Furthermore, it looks likely that covid-19 and its consequences will be with us for a considerable time to come. In this setting it could be argued that the need for the discipline of statistics has never been greater.
Statistics can be described as the science of learning from data. Sometimes people think of statistics as a bag of tools, a set of loosely-related mathematical methods which are applicable in different data scenarios. However, the American Statistical Association suggest that statistics should be viewed as
“an investigative process of problem-solving and decision making”
GAISE College Report ASA Revision Committee, “Guidelines for Assessment and Instruction in Statistics Education College Report 2016,” https://www.amstat.org/asa/education/Guidelines-for-Assessment-and-Instruction-in-Statistics-Education-Reports.aspx
Making decisions based on conclusions drawn from the careful analysis of appropriate data seems highly relevant to how we address the problem of covid-19. David Spiegelhalter, Chair of the Winton Centre for Risk and Evidence Communication at the University of Cambridge, has been very active on social media and in the traditional news media on the subject of covid-19 and statistics. Among David’s considerable output are two articles in the Guardian from earlier in the pandemic:
- How to interpret coronavirus statistics
2. The role of statistics, and of statisticians, in the response to the pandemic
Prior to the pandemic, David also published the excellent and highly accessible book ‘The Art of Statistics’, which provides an introduction to the modern practice of statistics:
While this is not a resource that can be accessed instantly online for free, it can be purchased for under £10 from all good book shops, as well as most of the bad ones too. On the subject of books, another excellent paperback published before the advent of covid-19, and also available for less than £10, is ‘Factfulness’ by Hans Rosling:
Rosling was a Swedish global health physician and public educator who sadly died in 2017. ‘Factfulness’ makes the case for a fact-based worldview achieved through data and outlines how our instincts can trip us up and impair our perspective.
Both ‘The Art of Statistics’ and ‘Factfulness’ are books which advocate for greater levels of statistical literacy, the ability to understand and critically assess the arguments and conclusions of others made using data (rather than necessarily the ability to carry out statistical analyses oneself). The importance of this skill has been recognised for a long time, with the following quote usually attributed to H G Wells, author of Victorian science fiction classics ‘The Time Machine’, ‘The Invisible Man’ and ‘The War of the Worlds’:
“Statistical thinking will one day be as necessary for efficient citizenship as the ability to read and write”
It’s a good quote, even if not strictly what H G Wells said (see the introduction to The future of statistical thinking (RSS) to see what Wells actually said and who actually said the line above).
Another good line on the importance of statistics is attributed to Florence Nightingale:
“Statistics … is the most important science in the whole world, for upon it depends the practical application of every other [science] and of every art”
(See Nightingale on Quetelet on JStor, if you really want to track the origins of this quote down.)
Known to most people as the “lady with the lamp”, a pioneer of nursing during the Crimean War (1853–56), Florence Nightingale was also a statistical pioneer and the first female fellow of the Royal Statistical Society. Recently there has been a growing interest in the statistical aspects of Nightingale’s revolutionary role in health reform, with 2020 being the bicentenary of her birth. You can learn more about Nightingale’s use of data, and particularly her use of novel data visualisations at This is Statistics and in the following articles:
The fact that the temporary critical care hospitals established in England in response to covid-19 have been called Nightingale Hospitals may owe more to the “lady with the lamp” view of Nightingale rather than her being the “lady with the data”, but the latter may be just as apt.
All good?
So, if the importance of statistical thinking has been appreciated since at least Victorian times, both in general and specifically for health research, has it been one smooth upward trajectory since then? Well, the development of statistical science and the growth of information technology as a tool for collecting, processing and analysing data has undoubtedly had a huge impact on the biological, medical and health sciences, enabling new questions to be posed and for those questions to be answered in new ways. But statistics has recently been doing some soul-searching too. One uncomfortable issue concerns the links between the growth of statistics as a discipline in the 19th and 20th centuries and the now discredited science of eugenics. Many of the leading figures in the development of statistics were also leading eugenicists, holding views which would now be considered racist. See The troubling legacy of Francis Galton for an introduction, focusing on Francis Galton, the polymath half-cousin of Charles Darwin who coined the term “eugenics”.
The legacy of these figures is now being reassessed, with, for example, some institutions “denaming” buildings named after eugenicists such as Galton:
Angela Saini’s book ‘Superior: The Return of Race Science’ is essential reading for anyone interested in race and science, which should really include everyone connected with science, shouldn’t it?
The Reproducibility Crisis
Another issue facing statistics is the role that the discipline plays in the reproducibility crisis in science. Could it be that the widespread misuse of statistical ideas and methods has contributed to, or even been the leading cause of, an explosion in spurious research findings? John Ioannidis famously claimed in 2005 that “most published research findings are false”:
While as far back as 1994, Doug Altman warned of “The scandal of poor medical research”:
If there are fundamental issues with the way that scientific research is conducted and reported, then there’s unlikely to be a quick fix. Statisticians are considering what a better future might look like though, for example in The reproducibility crisis in science: A statistical counterattack (RSS) and in the blog below:
Many responses to current situation call for a greater recognition of the importance of collaboration between scientific researchers and statisticians, and for improvements in statistical training for scientists, including a focus on statistical thinking.
Over to you
If you’re still here, thanks for making it this far. Hopefully some of the articles above will have given you food for thought. What’s your take on the role of statistics in your discipline or on the need for statistical thinking? Is statistics part of the problem or part of the solution to the reproducibility crisis? Can you find any articles on the practice of statistics in your own area of interest that you’d like to share with others? Please respond to this post to join the conversation.
