Here’s what I think it happening. When you run an Opt-In or Opt-Out system, you will have a donor registry with the names of those people who have opted in or opted out as appropriate. A number of Presumed-Consent countries have only an Opt-Out registry. I don’t have data for Portugal, Hungary or Poland. But France and Austria are like this, having no opt-in registry, just an opt-out one. Belgium has two registries, but its presumed-consent system is unusually strong. But Sweden has both an Opt-In and an Opt-Out registry. This presumably makes the choice situation more complicated and default-action type arguments wouldn’t be expected to apply so easily. So I think this may explain why Sweden has a lower rate of sign-up to be an organ donor than other presumed-consent countries—there’s more opportunity for choice at the point of decision, hence it’s easier not to sign up.

If we want to increase the supply of transplant organs in the United States, we could presume that people want to donate, rather than treating nondonation as the default. A study by social scientists Eric Johnson and Dan Goldstein showed that “presumed consent” could save thousands of lives annually.

Sounds like they’re reading even more into the study than Ariely does. Given your research, and your description of Johnson and Goldstein, it seems that Sunstein and Thaler’s claim is pretty indefensible.

Yes. In my book and some papers on this stuff I do an analysis for the U.S. case using the best available estimate of donor-evaluable deaths—that subset of all deaths where the cause and circumstance of death are consistent with the possibility of organ procurement. But in practice this number is very difficult to calculate and get a consistent time series on, even just for the United States. (A report from a while ago on this topic by the GAO found that the accurate collection of this kind of data at a national level would be too expensive, even for the Federal Government.)

As it turns out, at least in my experience the use of donor evaluable deaths vs all deaths vs the total population turns out not to make a very large difference to quantitative estimates of interest. (Which, I can tell you, was a little irritating to discover as a grad student, after I had gone to considerable trouble to calculate the in-principle better number. There is a lesson here.)

I think most of your French readers have just walked away in disgust. (Not that I blame them.)