Tom Chivers is a science journalist, former student of ethics and philosophy and author of How to Read Numbers: A guide to statistics in the news. In conversation with Panoramic’s Anya Gera and Sophie Williams-Dunning, Chivers discusses the statistical and ethical debates that have arisen during the pandemic, including the highly taboo question - whose lives are worth saving?
The events of the last year have brought complex ethical questions from the abstract into the everyday. In many countries, as in the United Kingdom, emergency laws and lockdowns have demanded citizens sublimate their individual desires for the greater good of society. One year into the pandemic, we have all become habituated to this state of affairs, and it is easy to forget the ethical equation that underpins the imposition of a lockdown: this equation balances x - the majority’s freedoms - against y - a minority of lives - and concludes that y is more valuable than x. Obviously, I’m simplifying, but I hope framing the question of lockdown in this way serves to illustrate the ethical debate at the heart of pandemic politics. The equation analogy also highlights that, in taking these huge decisions, information is key, and it really is a numbers game. R numbers, case numbers, excess mortality rates; these are the statistical measures with which we’ve become all too familiar in these most unfamiliar of times. Ethics and statistics, therefore, provide two key academic frameworks through which to tackle the political debates of the moment, and who better to discuss them with than Tom Chivers; science journalist, former student of philosophy and ethics, and author of How to Read Numbers: A guide to statistics in the news (and knowing when to trust them).
When talking to Tom, the first thing that becomes clear is his firm belief that whilst statistics are numerical they do not reveal objective truths to us. Rather, statistics provide answers to the questions you ask of them, and the process of framing and defining the question leaves a lot of room for the shifting of goal posts. “Let’s take the example of COVID deaths,” explains Tom, “this might sound like a really obvious statistic […] but actually, you could count it in many ways. You could count the number of confirmed COVID deaths, the number of excess deaths as above a five-year average, or the number of deaths with COVID recorded on the death certificate. So there’s no simple straightforward answer.” These measures can, and do, give very different results.
Therefore, whilst statistics can guide actions and policies, they also have a great capacity to mislead, or to be misused. This is why Tom believes it is so important for people to be able to decode the use of statistics in the news and politics, “You always have to ask yourself,” emphasises Tom, “why does this number exist? How has it been made? Where does it come from and how trustworthy is it?” It’s also important for those making and using statistics to question their own intentions as well, and politicians and journalists alike have a duty to treat and transmit statistics in an ethical way. “Either you can use statistics as a weapon with which to fight a sort of war of information, as a way to convince people to action, or as a tool to try and understand the world. The ethical use of statistics is as a tool for enlightenment rather than to provoke action.” (The words bus, European Union and £350 million a week spring to mind). Or, as Tom puts it, never use statistics like a drunk uses a lamppost, “for support rather than for illumination.”
So how does Boris Johnson’s government score for its ethical use of statistics so far in the pandemic? Have we seen any statistical measures misused and has the government really been “following the science” in its decision-making? In particular, we wanted to know what Tom made of those who have questioned the government’s use of the R number as the holy grail indicator of where the pandemic might be advancing or retracting. Tom uses Goodharts’ Law, to illustrate that over-dependence on any one statistical measure is bad. According to this Law, if any statistical measure becomes a target in and of itself, it ceases to be a good measure because people begin to find ways to artificially control it, rather than control the outcome it is meant to be measuring. In light of this, Tom still assesses the R value to have generally proven itself as an important and a useful measure.
A less well-known statistical measure that Tom wishes was more used in decision-making processes is quality adjusted life years, or QALYs. In layman's terms, a QALY is a generic measure which takes into account both the quantity and quality of life lived in order to, quite bluntly, assess the relative value of keeping someone alive. The notion of QALYs seems intuitive - most people would react differently to the death of a healthy 30 year-old, then to the death of an ailing octogenarian - but also, somewhat ethically uncomfortable. It is hard to reconcile the cold hard maths behind QALYs with the gut feeling that one should not accord different human lives different values. Tom’s hardened statistician’s mentality notwithstanding, he recognises the dilemma: “Imagine arbitrarily that we all have a life expectancy of 80. If a 30 year old dies, that is 50 years of expected life lost, and if a 79 year old dies, that’s one year. Not even taking into account the quality adjusted side of things, is that 30 year old dying 50 times worse than the 79 year old dying? Are we comfortable with that as a society? Is that a conversation we’re willing to have? In my weird nerdy way, I wish that society would be able to have it, but actually I think it’s really, really hard.”
So what might policy-making that relies on QALYs look like? Applying QALY logic to the vaccine rollout might rationally lead to prioritising a slightly younger age group for vaccination above the very elderly. After all, people aged 50-70 are more likely to catch COVID than the very elderly (as many still work or live in intergenerational households with family members who might expose them to the virus) and they are still fairly likely to suffer a severe illness. Given the high risk to this age group, and the greater number of QALYs lost when a 55 year old dies in comparison to a 75 year old, the logical conclusion would be to favour vaccine rollout to this younger group. This kind of decision-making, however, is not ethically unproblematic and is reminiscent of the old trolley dilemma: deliberately diverting the vaccine away from those who are most obviously at risk, even if it resulted in less QALYs lost overall would be a hard decision for anyone to take, let alone a politician facing a discontented public. As it turns out, Tom tells us, the Joint Committee on Vaccination and Immunisation did take QALYs somewhat into account when calculating its vaccine priority list, and in fact, the realities of implementing a wholly QALY based system are more complex than they first seem. Nonetheless, thinking in terms of QALYs has its uses as a comparative tool. Furthermore, it demonstrates that even if lockdowns or vaccines are supposedly the result of a utilitarian line of reasoning that attempts to bring the most good to the greatest number of people, changing which statistical measures we use will result in different outcomes.
For instance, the ethical equation behind imposing a lockdown becomes more dubious when we consider broader repercussions 5, 10 or 20 years into the future. Considering the impacts lockdowns have had and will continue to have on everything from geopolitics and the global economy to child development and mental health, we asked Tom whether lockdown is still a sound utilitarian solution when we look further ahead. “You can only be utilitarian about things that are knowable;” he tells us, “you can’t say: I should go and stop that butterfly from flapping its wings because it’ll cause a storm in Brazil in three weeks. You can’t predict those outcomes, it just gets too complicated. We can only focus on the lives and realities in front of us.” Whilst he admits that lockdown causing a huge drop in GDP and a serious recession is a knowable fact, he calls for a consideration of the counterfactuals. “It’s not that if we didn’t lock down everyone would be going about their daily lives - going to cinemas and pubs - because there would still be a horrible deadly disease killing loads of people. If you look at places like Sweden, which didn’t have an official lockdown, their economy still shrank massively. It wasn’t a choice between economic damage from lockdown and business as usual; it was a choice between economic damage from lockdown and less economic damage but then thousands more people dead. I suspect it’s a fairly straightforwardly utilitarian [approach] that we locked down.”
On some issues, Tom’s vision of an ethical system for society is quite easily applicable. The AstraZeneca vaccine, its possible side effects and the tensions that arose with the EU (“It’s darkly amusing but their attitude seems to be ‘this vaccine is terrible but also give us loads more of it please’”) provide a good counter-example for Tom’s notion of sound utilitarian decision-making. As he puts it, “If a [hypothetical] vaccine saves 90% of the population but it kills one in a million people who take it, pure Kantian ethics would say, ‘Don’t administer that vaccine because by doing so you’re allowing people to die through your action’. But you can’t think in those terms when you’re looking at society as a whole: on the balance of probability this will save many more lives than it ends, therefore we should use the vaccine.” In other areas, as with vaccine priority lists and lockdowns, the answers that statistics and ethics provide are perhaps more difficult to get one’s head around. But as complex as they may be, these academic disciplines provide us with the necessary tools for scrutinising the presumptions behind and justifications for decisions imposed on us from above. Understanding these disciplines is the first step to demanding better solutions to complex and evolving problems.
Comments