The consent gap is 60 points. The donation gap is 2.3 per million.

The famous chart shows a 60-point consent gap. The same paper measured actual donations: 14.1 against 16.4 per million. Only one number travelled.

There is a bar chart you have almost certainly seen, even if you could not name it. Four short bars on the left, seven tall ones on the right. Denmark sits at 4.25 percent; Austria, 99.98. The caption says something about defaults, and the point lands instantly: change one checkbox, save thousands of lives.

The chart is real. It comes from Eric Johnson and Daniel Goldstein, first in Science in 2003, then in fuller form in Transplantation the following year. The numbers in it are accurate. What almost never travels with the chart is the second measurement the same authors made, on the same page, which is far less flattering to the idea.

What the tall bars are counting

The left-hand countries ask you to opt in. Denmark 4.25 percent, Germany 12, the United Kingdom 17.17, the Netherlands 27.5. The right-hand countries presume consent unless you opt out: Austria 99.98, Belgium 98, France 99.91, Poland 99.5, Portugal 99.64, Sweden 85.9. Hungary reaches 99.997, which should already tell you what is being measured.

It is called the effective consent rate, and it counts the people who did not fill in a form. In Hungary, three people in a hundred thousand actively removed themselves from the register. Everyone else is in the tall bar because they did nothing at all. That is not a country of enthusiasts. That is a country where the paperwork runs the other way.

Johnson and Goldstein knew this perfectly well — it is the entire argument of the paper. Their claim was never that Austrians are more generous than Danes. It was that most people have no settled preference about donation, and so the form of the question writes the answer. They supported it with an online experiment on 161 people: 42 percent agreed to donate under an opt-in default, 82 percent under opt-out. In a third condition with no default at all, forced to choose, 79 percent said yes — statistically indistinguishable from the opt-out group. The default created no willingness; it removed an obstacle between the willingness already there and the register.

The measurement that stayed behind

Then they asked the question that actually matters. Does any of this produce more transplants?

They took actual deceased donations per million population, ran a time series from 1991 to 2001, and controlled for transplant infrastructure, education, religion and each country's general propensity to donate. The default effect survived: opt-out countries came out at 16.4 donors per million against 14.1 for opt-in. Significant at p < 0.02. An increase of 16.3 percent.

Sixteen percent is worth having. Sixteen percent is not sixty percentage points. The gap in who is nominally registered and the gap in who is actually donated from are different by more than an order of magnitude, and only one of those two numbers made it into the popular version.

There is a small oddity here that shows up in the original and in none of the summaries. The published abstract of the 2004 paper states that "opt-in countries have much higher rates of apparent agreement with donation." Every figure and every line of the body says the opposite. It is a slip in the abstract of one of the most-cited papers in behavioural public policy, sitting there since December 2004.

Spain keeps getting cited for the wrong reason

Spain has the highest deceased donation rate in the world and has presumed consent. The inference writes itself, and it is wrong.

The Spanish law dates from 1979. The donation rate did nothing for a decade. It was 14.3 donors per million in 1989 — respectable, unremarkable. Then it climbed: 22.6 by 1993, the highest in the world at the time, and 35.1 by 2005.

What changed in 1989 was not the law but the staffing: the Organización Nacional de Trasplantes, and specifically the network of hospital transplant coordinators it built — intensive-care doctors, inside the hospital, who already know the staff and are there when a family is told the worst news of their lives. The coordinator identifies potential donors that the system would otherwise never see, and handles the conversation.

The part that undercuts the whole default story: Spanish coordinators ask the family anyway. The presumed-consent law is, in day-to-day practice, close to a dead letter. Spain runs an opt-in system staffed as if it were the most important job in the hospital, and beats every opt-out country in Europe.

Five countries that actually switched

Austria and Denmark differ in a hundred ways, and only one of them is the checkbox. The cleaner test is to watch a single country cross the line.

In 2024, a group at the Max Planck Institute for Human Development — Dallacker, Appelius, Brandmaier, Morais and Hertwig — did exactly that, in Public Health. They found five countries that moved from opt-in to opt-out and looked at donation rates within each one, before and after: Argentina, Chile, Sweden, Uruguay, Wales.

The switch did not raise donation rates. Their conclusion is not that defaults are powerless but that a default arriving on its own, without money for intensive care capacity, without public campaigns, and without work on what families say in the room, does not move the number that counts.

Wales is the case worth sitting with, because it was supposed to be the demonstration. The soft opt-out system took full effect on 1 December 2015. England followed on 20 May 2020. By the second quarter of 2023, family consent rates in England, Scotland and Wales had all fallen compared with the same quarter of 2019 — and Wales, the earliest adopter, had fallen furthest, from 63.6 percent to 56.3.

I want to be careful with that last figure, because a pandemic sits directly across the middle of it and the researchers who published it said so. You cannot read a clean verdict out of 2020 to 2023 data about anything that happened in an intensive care unit. What you can say is that four years after England changed the law, nobody is pointing at a jump.

Why the family is the whole thing

The mechanism is not mysterious once you look at where a donation actually gets lost. Almost every opt-out country in Europe operates a soft version: the register says yes, and the clinicians still ask the next of kin, and if the next of kin says no, it stops. Legally the family may have no veto. In practice no hospital in Europe is going to proceed over a grieving mother's objection, and none should have to.

So the register moves from 17 percent to 99 percent, and the decision stays exactly where it was — with a handful of relatives in a side room at three in the morning, being asked something they have never discussed. Whether that conversation goes well depends on who is having it, how much training they have had, and whether the person they loved ever said anything out loud.

Which is the one thing the default cannot do for you. A presumed-consent law tells the state what to assume. It tells your family nothing.

What I take from it

I don't think the 2003 paper is wrong, and I am suspicious of the current fashion for treating nudge research as uniformly discredited. The finding replicates. Defaults do move registration, dramatically, and 16 percent more donors is not a rounding error.

What went wrong is the compression. A careful paper with two numbers in it got quoted with one, and the one that travelled was the one that was sixty points wide and measured inaction. A decade of policy arguments were conducted at that number. Wales and England changed their laws partly on the strength of it.

What I do not know is how much of the Spanish result would transfer. Spain has a public health service, a specific hospital culture, and a coordinator role that took fifteen years to mature. Every country that has tried to copy the model has copied the org chart, and the results have been mixed. I haven't found a convincing account of which part is load-bearing, and I don't think anyone has.

For anything built around getting people to help each other — which is what we do here — the lesson is uncomfortable and cheap to ignore. Making the yes easy is real work and it works. It is also the easy half. The other half is someone showing up, knowing what to say, at the worst moment of somebody's life. Nobody has found a checkbox for that.

Sources: Johnson EJ, Goldstein D, "Do defaults save lives?", Science 302 (2003), 1338–1339; Johnson EJ, Goldstein D, "Defaults and Donation Decisions", Transplantation 78:12 (27 December 2004), 1713–1716 — all consent percentages, the 161-person experiment, and the 14.1→16.4 donors-per-million figure are taken from the text and figures of this paper. Dallacker M, Appelius L, Brandmaier AM, Morais AS, Hertwig R, "Opt-out defaults do not increase organ donation rates", Public Health 236 (2024), 436–440. Spanish donation rates and the role of the Organización Nacional de Trasplantes: American Journal of Transplantation, "How Spain Reached 40 Deceased Organ Donors per Million Population". UK consent rates April–June 2019 and April–June 2023: NHS Blood and Transplant data as reported in the analysis of donation trends in England, Scotland and Wales under COVID-19 and opt-out legislation. Checked 13 September 2026.