In these times of alternative facts Tim Harford’s quietly spoken commitment to the notion of truth is enormously welcome. His programme, More or Less, on BBC’s radio four is an enlightening analysis of statistical claims made on the news, often by politicians. Attempting and succeeding to distinguish the soundly based from the mildly misleading, the recklessly inaccurate and the straightforward propaganda lies.

For fans of the programme his book “How to Make the World Add Up” is a must read and for a much wider audience it is a should read. The first thing to say is that it is a book about statistics that those who did not even do O Level Maths can read with pleasure. There are no complex formula. On the contrary it is a limpid paean to the importance of statistics and its power to deepen and clarify our understanding of so many aspects of the world we live in.

Throughout there is a reassuring tone of common sense and inspiring presentation of how maths and statistics can be used to penetrate complex issues and abused to mislead and confuse. It is like being back in secondary school with that teacher who never raised his/her voice but never lost control of the class. Who maintained everyones interest through their enthusiastic mastery of their subject and ability to communicate in a manner accessible to all.

The book sets out eleven rules about how you should approach statistical claims. They all attempt to help you avoid confirmation bias which in the age of social media echo chambers leveraged by bots trolls is a vital skill. Asking you to think how you feel when you see a statistical claim. Does it give you a positive emotional boost as you find evidence to support your core beliefs/prejudices or does it make your hackles rise by contradicting them?

His advice is to treat those two imposters the same way and apply an objective set of challenges to both to try to discern the gold from the dross. This involves a deal of common sense and attempting to secure some kind of contextual framework. One which combines a birds eye view with a worms eye view. And one which understands what question any set of numbers is supposed to be answering. What are the definitions of the things that are counted. If it is gun deaths in the US is that about, mass killings, accidents, suicide or all the above?

it is important to avoid being enticed by the unusual nature of the findings in a report. Science progresses by findings exceptions to what is known. Scientific publications are would be tedious if all they ever published were results confirming what is already known. This means they look for the unusual. When combined with the academic imperative to publish or perish this can lead to the publication of findings which are unusual.

These may be a genuine step forward in scientific knowledge or they may be a result which is either unreplicable or designed in such a way as to provide eye catching if not knowledge enhancing results. Things like red wine prevents (yeay) or causes (boo) cancer. Along the way Hartford provides pointers to helpful tools which present a comprehensive picture of the state of peer reviewed research on medical matters, eg. the Cochraine Library.

Another target is the extravagant claims made for big data. The claim that google’s search engine was better at predicting flu epidemics than more traditional surveys, or that loyalty cards can predict your future shopping needs and that because such enormous data sets are being used the need for sample design was avoided, the facts would speak for themselves. At this point Harford quotes Professor Sir David Spiegalhalter of Cambridges University who said all this was “complete bollocks”.

This assessment may be of limited import when related to predictions about your need to buy nappies on the back of your previous purchase of follic acid. However, when such techniques are integrated into algorithms to guide recruitment or predict criminal behaviour or which teachers are failing to perform and need to be sacked, then they become of real concern. What they seem to end up doing is replicating and legitimating unfounded biased decisions.

The problem does not so much reside in the data sets or the use of algorithms in principle. Rather, it is the secrecy clothed in commercial confidentiality which means it is difficult to challenge results which seem to be producing bizarre outcomes. If big data is going to become a useful tool for analysing consumer preferences, criminal propensity, or professional competence it needs to be done in an open and transparent manner so that the logical steps in the chain of reasoning can be understood rather than hidden under a mountain of data.

If the misuses of statistics are of concern in civil society their manipulation by the state are a much greater concern. The book makes the point there are times when the bedrock statistics from national governments cannot be relied upon. This can never be right but is often ominous. Clear, reliable statistics about crime, health, the state of the national economy are fundamental elements of open government. They allow opposition parties, the media, academics and concerned citizens to question the government about the impact their policies are having. The book rightly draws attention to the geeks who try to protect the

Harford clearly articulates how critical reliable data is in modern democracies. Without a solid base to build knowledge all claims are equal. This is the realm of alternative facts. It is the logic of ex President Trump that if you don’t count the number of infections they go away.

The easy readability of this book belies the urgency, importance and contemporary relevance of it messages. One of which is the importance of sound maths and sound statistical techniques. But the other rather more profound one is the importance of bringing a critical mind to what the numbers purport to tell you. The need to understand what it is the numbers actually relate to and how accurate they can possibly be. What the standing of the authors is and the transparency of their sources, data and techniques.

Most important however is the need to examine your own motivations and biases. Carefully consider the extent to which your understanding of the numbers is guided more by your deep seated beliefs and prejudices than by what they actually say. In a world where some see evidence as a weapon to be tailored to promote a preconceived theory, or where dispassionate review of facts is dismissed as the tyranny of experts this book speaks to a key issue of our age, how we secure sound knowledge. Should be on the reading list of every first year University course in the country.

How to Make the World Add Up: Ten Rules for Thinking Differently About Numbers. The Bridge Street Press 2020. Tim Harford.