Shaping debate on religion in public life.

Lies and damned lies about statistics

31 May 2019

John Henry, an ordinand at Ripon College Cuddesdon, makes a passionate moral case for increasing our statistical literacy.

Do you love statistics? Do you get excited at the idea of analysing a couple of million rows of data to understand what is going on? Or examining a complex set of charts with lots of correlation coefficients? Not likely, I suspect. And why bother anyway? After all, there are only lies, damned lies, and statistics. Right?

The origin of this well-known quip is often assumed, but in truth it is unknown. The validity of this statement is similarly assumed by most people, but on further reflection, this is equally misguided. The tragic irony of this endlessly-parroted ideology is that the proper use of statistics is a critical tool for countering falsehood and providing a firm foundation for good decision making and public policy.

The aim and function of statistics is to understand probability and uncertainty in our world. From the probability that the jet engine on your next flight won’t explode at 33,000 feet, to the probability that the amount of vaccine injected into your child won’t kill them, to the probability that global temperatures will rise by 2 degrees Celsius in the next fifty years—statistical methods keep billions of people safe and make life better.

Theologically, statistics (and our ability to understand and apply mathematics more generally) must be seen as one of the central gifts of mind that our Creator has given us. Like all our gifts and powers, we believe we have a moral obligation to use them for good.

But statistics, of course, can also be abused. Statistical findings, which often reach the public sphere as single bits of ‘data’, can form the foundation of corrupt initiatives by individuals and organisations whose objective is to convince, or obfuscate, or both. ‘Let’s spend £350m more on the NHS’ comes to mind at this moment in time.

But abuse through statistics requires two groups of conspirators: the creators of the false narrative based on bad data and poor statistical analysis, and the recipients of the false narrative. Unlike robbery or assault, ‘victims’ of statistical abuse cannot avoid bearing some responsibility for the crime—for it is our lack of understanding of solid statistics and data analysis which makes us vulnerable to abuse.

Whether it be the terrible mathematics teacher we had at high school, or our struggles with frequently abstract concepts, the vast majority of people lack the knowledge and understanding of basic statistical concepts that are required to sufficiently understand the complexity of the modern world. This is especially so when data and statistics are explained to us via the media.

Statistical methods and data analysis are the tools we use to extract meaning from data. But our statistical capabilities have not kept pace with the volume, complexity and importance of data in our world. People frequently complain that too many statistics are bandied about. But this could not be further from the truth—we need more and better statistics, not less.

So, what is the solution? Statistics is hard, and our time and capabilities are limited. It is obviously naive to expect us to all become data journalists or statistical experts. We clearly have no choice but to rely on organisations and individuals whom we trust to analyse and synthesise the complexity of the world and then communicate their insights to us. But given the complex world we live in, we have no option but to increase the statistical standards we expect of ourselves and each other. And I think we need to demand this improvement in two distinct ways.

Firstly, we must demand better statistical rigour from our day-to-day media. When a journalist or commentator uses data in an argument, we must demand a source and a statistical context. Failing that, we must use and support organisations like FullFact.org, whose purpose is to analyse the strength and validity of the data presented to us in the media. On our part, we ought to remind ourselves, or learn for the first time, basic concepts of statistical confidence. We should try to understand correlation coefficients, sample sizes, t-values, and compound annual growth rates. It’s all there on Wikipedia and YouTube. Then we should demand these from our media—and ignore those who refuse to offer the most basic level of statistical rigour.

Secondly, we must demand better data visualisation and properly employ the massive visual processing power of our brains. Professor Edward Tufte of Princeton University has spent his career developing the principles of how to do this well. Statistical analysis is not all about equations. And there are early signs of hope in this regard. ‘Data journalism’ is now a trendy topic. We frequently see ‘infographics’ in our media today. And we increasingly see the use of ‘micro-charts’ to give an immediate visual description of trends of a particular metric, rather than an absolute number or a single growth rate.

But we have a long way to go. Journalists, commentators and politicians continually get away with using single-number sound bites without any statistical context. The quality of ‘infographics’ is frequently so poor that you get little more insight than if the data was listed in a table.

At its heart, this is a profoundly moral issue of the responsibility that we all have to each other as citizens. As a society we have both rights and responsibilities. We live under laws and social constructs which make demands of us for the safety, security and wellbeing of our communities. For many of us, we believe in a divine call to love one another. But for these ideals and standards to be met, we accept the need for training and education. Our safety and the safety of others is protected by demanding sufficient training and testing before we are allowed to participate in activities which impact others (e.g. driving).

The risks, consequences and potential loss of a car accident are clear and visceral, but the risks, consequences and potential loss of poor statistical understanding and data analysis are much more abstract. Yet the risk and potential damage of millions of people misunderstanding the facts about the world is far greater. Our political and social systems, and more importantly, the global ecosystem, are at stake.


More blogs on religion and public life…

Whose “bloody GDP” is it anyway? by Tim Howles

Come the Resurrection…? by Rosie Dawson

Chinese Christian Schools in the 21st Century by Oscar Siu

Tell the truth and act as if the truth is real by Matt Stemp

Share this page:

Discuss this

Discuss this

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.