The authority fallacy and the “7-day NHS”

Prof Stephen Hawking with David AttenboroughEarlier today, the health minister Jeremy Hunt posted some tweets claiming that Professor Stephen Hawking was wrong in his assessment of the data regarding the “weekend effect” (the notion that people admitted to hospital over the weekend were more likely to die than those admitted during the week because fewer doctors, and in particular fewer consultants, are working). The ‘effect’ has been cited by the Tories and the right-wing press to support Hunt’s proposals for a “24-hour NHS”, while others have debunked the idea. Professor Hawking is to make a speech at the Royal Society of Medicine today criticising the plans and is accusing Jeremy Hunt of “cherry-picking” statistics to support his position. The social media response to Hunt has been to emphasise Hawking’s status as one of the world’s foremost scientists and Hunt’s as a relative nobody despite his powerful position. As obvious as it might seem that Hunt can’t argue with a famous scientist about numbers or data, it’s a classic logical fallacy, the “argument from authority”.

To put it simply, a ‘nobody’ and indeed a widely and rightly disliked politician can indeed be right and a scientist with a PhD and however many dozen peer-reviewed papers and books published can be wrong. This is particularly true when the issue is not the scientist’s particular kind of science. Professor Hawking is a theoretical physicist, not a statistician; his work has been mainly concerned with black holes and gravity, and doubtless there is a major mathematical element to all this (I’m no expert; I got an E in GCSE physics and dropped it thereafter) but it does not make him an expert in health statistics (his Wikipedia entry does not even mention statistics once) or indeed anything to do with healthcare except his own condition, and that could be said of anyone with his condition, PhD or no.

Citing someone’s status as “a scientist” is potentially a very dangerous use of this fallacy. As George Monbiot has noted, a lot of the material purporting to disprove man-made climate change is predicated on the scientific credentials of its authors, but most of them are in fact not climate scientists; the vast majority of those are agreed that it is real. In the early 2000s a number of women were languishing in prison for multiple child murders which were, in fact, natural deaths; a major plank of the prosecution was the insistence by the paediatrician, Professor Roy Meadow, that multiple cot deaths in one family just do not happen. In the case of Sally Clark, who had lost two children to cot death, he told the jury that the likelihood of this happening were 1 in 73 million; this was based on an elementary error in maths, the presumption that the two deaths (in which cot death affects one in every 8,543 babies born) were independent of each other, and thus you could just multiply the two probabilities together to get the 73 million figure. In the radio programme that exposed the case, the interviewer noted to his interviewees twice that Meadow was a distinguished paediatrician, and one of the interviewees responded that he was not a statistician. The same interviewer also made the point many people would have made when defending Meadow: will you believe a scientist and a knight, or an unemployed barman (the then-husband of one of the other wrongly-convicted mothers)?

Proving a fallacy does not, of course, prove that the entire argument is wrong. Others have countered the idea of a “weekend effect”, and doctors commonly work beyond their hours in the event of an emergency or, say, an operation overrunning the time it had been expected to take. The point is that it isn’t valid to argue that Hunt must be wrong and Hawking right because of Hawking’s stature in the world of theoretical physics or cosmology, because those are different disciplines from statistics, much as is paediatrics or any other form of medicine.

(On the particular subject of the NHS operating on weekends: I’ve known people who have been admitted to psychiatric wards on Friday evenings after the consultants have gone home and then found they are prevented from leaving the wards because the consultant has not approved it — which they should not have to in the case of informal patients, but this requirement is commonly, and illegally, imposed. There have also been cases where leave has been granted but not written up, and this is only discovered when the relative arrives to take the patient out and the consultant is at home. In such cases, consultants should at least be contactable at weekends or at least whenever a new patient is admitted, especially if it is planned; the patients they have power over do not, after all, get the weekend off. But if it really were dangerous to fall ill on a weekend, it would not have taken a politician to notice it; it would have been a public scandal going back years, and it has not been. Most of the public dissatisfaction around the NHS has been to do with under-funding and specific incidents of negligence and malpractice unrelated to the fact that many staff work weekdays and not, usually, weekends.)

Possibly Related Posts: