How do we all know what works? In an unsure world, filled with untested remedies, disputed insurance policies and untried innovations, it must be a urgent query. For an impoverished younger tailor named Franz Reichelt, the reply was: we discover out the exhausting means.
In February 1912, on a chilly Sunday morning, Reichelt persuaded journalists, a digicam crew and a small crowd to look at him reveal his new wearable parachute. The official story was that Reichelt would throw a dummy from the primary ground of the Eiffel Tower, greater than 50 metres up. Reichelt, nonetheless, had tipped off journalists that he would make the soar himself. In any other case, one suspects, there would have been fewer spectators.
The parachute is, because it occurs, one thing of a working joke in evidence-based policymaking. Within the Christmas 2003 version of the British Medical Journal, obstetrician Gordon CS Smith and public well being researcher Jill Pell printed a scientific evaluation of “Parachute use to forestall loss of life and main trauma associated to gravitational problem”. They famous, with mock sorrow, that “we had been unable to determine any randomised managed trials of parachute intervention” and known as for “essentially the most radical protagonists of evidence-based drugs” to organise and take part in a rigorous randomised trial of the parachute.
In case the joke has handed you by, the argument being lampooned is a criticism by advocates of evidence-based policymaking. Many well-established medical practices — together with concepts in policing, legal justice, schooling, economics and past — have by no means been severely examined by the gold commonplace for proof: a double-blind, randomised managed trial (RCT). RCT advocates counsel that it is a scandal. Is it?
Go away that query hanging for a second, as a result of 15 years later — once more within the BMJ Christmas subject — a group of researchers detonated an evidential bombshell. That they had run the long-missing randomised trial of the parachute — and located that parachutes had been no higher at stopping main harm or loss of life than an empty backpack. Anybody shocked at this end result want solely seek the advice of the small print: no person within the trial had suffered any noteworthy harm in any respect after leaping from an plane, and it’s because they did so “at considerably decrease altitude . . . and decrease velocity”. To be extra particular, they jumped out of “small stationary plane on the bottom”. One other hilarious satire from the BMJ.
There are severe classes right here, for many who can cease sniggering. The primary is that, sure, there isn’t any want for a RCT of a know-how — such because the parachute — whose advantages are indeniable. The second is that RCTs themselves don’t essentially show something: the researchers counsel “cautious extrapolation to high-altitude jumps”.
It’s putting how incautious extrapolation could be, notably in press releases and breathless media summaries. A small research of nursery college kids drawing photos is generalised to be a completely fledged concept of human motivation. Massive claims about the advantages or dangers of a meals additive teeter on the questionable basis of analysis on laboratory mice.
Randomisation is neither the one path to information nor a assure of everlasting reality. However it’s a excellent concept — which can assist to elucidate why there have been severe randomised trials, even of parachutes. As Andrew Leigh explains in Randomistas (2018), “experiments on parachute efficacy and security are widespread”. A few of them use crash dummies, whereas others examine particulars of method or gear: paratroopers are six occasions extra more likely to undergo an ankle sprain in the event that they soar with out an ankle brace. So there.
One vital purpose to make use of some type of randomisation is that it strips away every kind of potential confounding. Do small class sizes assist college students to thrive? You gained’t discover that out by merely evaluating an overcrowded, cash-strapped inner-city academy with a lavishly funded public college: their class sizes might differ, however so does the whole lot else. If you wish to study studying, run a randomised experiment.
The second purpose, nonetheless under-appreciated, is that consultants know lower than they wish to assume. Docs embraced bloodletting for two,000 years earlier than belatedly acknowledging that it’s not often an acceptable therapy. (The Lancet is known as after a bloodletting software.) Extra not too long ago, as Helen Pearson notes in Past Perception (2026), “In Eighties Britain . . . it was commonplace apply to shave pregnant girls’s pubic hair after they had been admitted to hospital and to present them enemas.” Episiotomies, too — cuts within the opening of the vagina — had been “fairly routine”. None of this was evidence-based. It simply appeared to docs that it might be a jolly good concept.
In the summertime of 2020, when the Restoration trial revealed that an inexpensive steroid, dexamethasone, was massively helpful to Covid-19 sufferers present process mechanical air flow, one physician instructed me that this so-called discovery was easy widespread sense. However Sir Martin Landray, one of many leaders of the trial, recollects the letters that consultants had despatched to him and to the regulators warning that dexamethasone — which suppresses the immune system — was so harmful to Covid sufferers that it was unethical even to check it. When some consultants assume it’s a parachute, and others assume it’s bloodletting, it’s a good suggestion to run the trial.
What occurs once you want a choice urgently, and gold-standard proof doesn’t exist? One method is to do what Landray and his colleagues did: transfer quick. By introducing a easy medical trial alongside routine care, they had been capable of get the trial working inside 10 days and produce life-saving outcomes inside 100.
However what if there isn’t any trial, and no prospect of 1? Then, argues Pearson, we want “aluminium commonplace” proof. Be as clear as potential about your speculation, break it down into a series of causal hyperlinks, and get the perfect information you possibly can — for and in opposition to — to check every hyperlink within the chain. Gold is best, however aluminium is reasonable and light-weight and generally adequate for the job.
Franz Reichelt was a greater publicist than parachute designer. His leap from the Eiffel Tower was captured for the ages on Pathé-Journal newsreel. I’d advise in opposition to watching it. He died of a shattered cranium and backbone, and — maybe — of a coronary arrest mid-fall. Wikipedia merely lists his explanation for loss of life as “parachute failure”. The self-belief of a pioneer is a robust factor. Usually, when the consultants suggest untested options, they don’t seem to be those that suffer the implications.
Written for and first printed within the Monetary Instances on 24 June 2026.
Loyal readers may benefit from the ebook that began all of it, The Undercover Economist.
I’ve arrange a storefront on Bookshop within the United States and the United Kingdom. Hyperlinks to Bookshop and Amazon might generate referral charges.
