Default Effects and Opt-Out Rules: What the Evidence Shows
Key takeaways · 15 min read
- In a 2003 online experiment, 82% of people stayed organ donors when donation was the default, against 42% when they had to opt in.
- Registering is not donating. Across six countries that switched to opt-out, deceased donors rose by a non-significant 7% while living donors fell by 29%.
- Pensions are where defaults clearly work: when one US firm enrolled new staff automatically, participation rose from 37% to 86%.
- The same inertia keeps people at whatever rate was chosen for them. Nearly two-thirds of those auto-enrolled stayed at the 3% default contribution.
Every form has an answer that applies if you do nothing. A box comes ticked or unticked, a pension starts or does not, a subscription renews or lapses. Behavioural scientists call this the default, and over two decades it has become one of their favourite tools, because changing it costs almost nothing and can shift what millions of people end up with.
Two examples carried the idea into government. One was a chart of European countries in which organ donor consent was either near-universal or rare, depending on whether citizens had to opt out or opt in. The other was a company pension plan whose membership more than doubled when new staff were enrolled unless they said no. Both are real findings. Only one of them translated cleanly into the outcome people cared about.
This article looks at what defaults change and what they leave alone. It separates a ticked box from an organ that is actually donated, and a pension that is joined from one that is big enough. It also asks the question the early enthusiasm skipped: whose interest does the default serve?
Why doing nothing matters
In standard economics a default should barely matter. If changing your answer takes a minute, people who prefer the other option will simply change it. In practice they often do not, and researchers have proposed three reasons. Changing takes effort, however small. The pre-set option can look like a recommendation from whoever designed the form. And once an option is in place, people judge alternatives against it, so moving away feels like a loss.
These channels are usually called ease, endorsement and endowment. They matter for this article because they point in different directions. If people stay because the default looks like advice, a default set by someone with their own agenda can mislead. If they stay because of effort, the people least able to make the effort are the ones most bound by whatever was chosen for them.
The famous donor chart
In 2003 Eric Johnson and Daniel Goldstein published a short paper in Science comparing European countries. Where citizens had to opt in to being a donor, the share who effectively consented ranged from about 4% to 28%. Where they had to opt out, it ranged from about 86% to almost everyone. Neighbouring countries with similar cultures, such as Germany and Austria, sat far apart.
The authors backed the chart with an online experiment. Americans were asked to imagine they had moved to a new state and to confirm or change their donor status. When donation was the default, 82% remained donors. When they had to opt in, 42% did. With no default at all, 79% chose to donate, suggesting that most people were willing and that the opt-in form was holding them back.
Saying yes to donation, by how the question was set
Share agreeing to be an organ donor in hypothetical online choices.
Sources: Johnson EJ, Goldstein DG, Science 302(5649):1338 (2003); Chandrashekar SP et al., Meta-Psychology (2022), direct replication.
The effect replicates, but smaller. A well-powered direct replication published in 2022 found 73.5% agreement under opt-out and 62.5% under opt-in. The difference was real, with an odds ratio of 1.67, but well below the original estimate of nearly six. That is a common pattern for famous findings, and it is still a sizeable nudge.
Notice what all these numbers measure, though. They count people saying yes on a form, or being counted as willing because they did not say no. None of them counts an organ reaching a patient.
Registration is not donation
Deceased donation is rare and complicated. A person has to die in a particular way, usually in intensive care, with organs still viable. Staff have to identify them, approach the family and organise retrieval. In nearly every country, opt-in or opt-out, the family is asked, and its objection usually ends the matter. A default set in law only bites when nobody knows what the person wanted and no relative is there to say.
A 2022 study in BMJ Open measured how often that happens. Using national data from Denmark, Germany, the Netherlands, Sweden, the UK and Wales, it found that the situation where the legal default alone decides the outcome made up between 0% and 4% of all retrieval opportunities. Everything else turns on what the deceased recorded, what the family says, and how well the hospital system works.
What the chart measured, and what counts
Two different outcomes that are often treated as one.
Opt-in countries: about 4% to 28%
Opt-out countries: about 86% to 99%
People counted as willing, 2003 data
Six European systems: 0% to 4% of cases
Family usually asked either way
Both the deceased’s and family’s wishes unknown
Sources: Johnson EJ, Goldstein DG, Science (2003), as summarised by Kalkstein DA et al., Behavioural Public Policy (2022); Molina-Pérez A et al., BMJ Open 12(9):e057107 (2022).
Comparisons between countries do show opt-out countries with more deceased donors on average. A 2006 panel study of 22 countries estimated rates 25% to 30% higher after controlling for road deaths, wealth and other factors, and a 2009 BMJ systematic review found similar associations. The review warned that this could not show the law itself caused the gap. Spain, the world leader, owes much of its success to hospital coordinators and investment rather than to its consent law.
When countries switched
The cleaner test is to watch countries that changed their law. A 2024 study in Public Health examined five, Argentina, Chile, Sweden, Uruguay and Wales, using donor rates before and after the switch. Averaged across them, moving from opt-in to opt-out did not raise deceased donation, and there was no sign of a gradual rise building after the change either.
A larger analysis in PNAS Nexus in 2025 compared six places that switched between 2005 and 2021, Slovakia, Argentina, Wales, England, the Netherlands and Scotland, with 18 that stayed opt-in. Deceased donors rose by 1.21 per million people a year, about 7%, a change too small to distinguish from chance. Living donors, people who give a kidney or part of a liver while alive, fell by 4.59 per million, or 29%, a statistically significant drop.
After the switch to opt-out
Average change in donors per million people a year, six adopters against 18 opt-in countries, 2000–2023.
Source: Güntürkün P et al., PNAS Nexus 4(10):pgaf311 (2025). Difference-in-differences with staggered adoption; bar length shows size, not direction.
The authors call this crowding out. In four further studies, they found opt-out framing went with less willingness to give a kidney to a stranger, as opposed to a relative. This is one study, and living donation also depends on transplant programmes, so the mechanism is not settled. But it shows that a default aimed at one behaviour can quietly reduce a related one.
Wales and England
Wales introduced deemed consent in December 2015, the first part of the UK to do so, while England stayed opt-in. That made a natural comparison. An early evaluation in BMJ Open found the Welsh consent rate in the first 18 months was 61.0%, up from a dip to 45.8% the year before. But consent in the rest of the UK had also risen, from 58.6% to 63.1%, so the Welsh increase could not be credited to the law. The number of deceased donors did not rise.
A longer analysis published in 2020 in Anaesthesia was more positive. From 2016 to 2018, 68% of Welsh families approached agreed to donation against 65% in England, and consent after brain death became significantly higher in Wales after about 33 months. No such change appeared for donation after circulatory death. The authors said the effect took around two years to emerge, which suggests the publicity and staff training surrounding the law may have mattered as much as its wording.
Wales after deemed consent
Selected results from evaluations of the 2015 Welsh law.
| Measure | Wales | Comparison |
|---|---|---|
| Family consent, first 18 months | 61.0% | Rest of UK rose 58.6% to 63.1% |
| Deceased donors, 21 months after vs before | 104 vs 101 | Difference not significant |
| Family consent, 2016–2018 | 68% | England 65% |
Sources: Noyes J et al., BMJ Open 9(4):e025159 (2019); Hawkes N, BMJ 359:j5659 (2017), reporting a Welsh Government evaluation; Madden S et al., Anaesthesia (2020), doi:10.1111/anae.15055.
England followed in May 2020, weeks into the pandemic, which makes any effect almost impossible to isolate. A 2024 analysis in PLoS One found English consent rates had been rising for a decade before the law and dropped during COVID-19, a fall the authors attributed to the pandemic. Even after the law, fewer than half of potential donors fell into the deemed-consent group. A 2025 conference abstract from NHS Blood and Transplant reported that 41.6% of the UK population had opted in on the register and 3.9% had opted out.
Where defaults clearly work
Pensions are the opposite case, because the choice on the form is the outcome. If you are enrolled, money leaves your pay every month. In 2001 Brigitte Madrian and Dennis Shea studied a large US company that began enrolling new hires in its 401(k) plan unless they opted out. Among staff with three to 15 months’ service, participation was 37% under the old opt-in rule and 86% under automatic enrolment. Nothing else about the plan had changed.
The United Kingdom made the idea national. From 2012, employers were obliged to enrol most staff automatically, phased in from the largest firms to the smallest. A study in International Tax and Public Finance used that staggered rollout to estimate the effect: participation among targeted private-sector employees rose by 36 percentage points, reaching 88% by 2015. The biggest gains were among people in their twenties, up 52 points from a base of 28%, and the lower paid.
Automatic enrolment and pension membership
Share of workers in a workplace pension, before and after enrolment became automatic.
Sources: Madrian BC, Shea DF, Quarterly Journal of Economics 116(4) (2001); Cribb J, Emmerson C, International Tax and Public Finance (2019), doi:10.1007/s10797-019-09565-6; DWP, workplace pension participation and savings trends 2009 to 2025 (30 July 2026).
Government figures published in July 2026 put participation among eligible employees in Great Britain at 90% in 2025, about 22.6 million people. Around 11% to 12% of new savers opt out. The gaps that remain are telling: in firms with fewer than five staff, only 55% of eligible employees were saving, a sign that a default still depends on someone setting it up properly.
When the default is not on your side
The inertia that pulls people into a pension also holds them at whatever settings came with it. In the Madrian and Shea firm, the default was 3% of pay in a money market fund. Almost two-thirds of the auto-enrolled cohort contributed exactly 3%, and fewer chose higher rates than before, so some people saved less than they would have under the old system. A study of three firms found that about 80% initially accepted both the default rate and the default fund, and half still did three years later.
Stickiness is not spread evenly. A working paper using data from ten large US employers estimated that a low-income employee was 7.2 percentage points more likely than a high-income one to stay at the default even when they wanted a different rate. Ireland’s new scheme starts workers at 1.5% of pay. Whether people treat the starting rate as a floor or a recommendation will matter far more than how many stay enrolled.
The same mechanics work for sellers. A subscription that renews unless you cancel is a default, and it earns most from people who forget. In September 2025 the US Federal Trade Commission settled its case against Amazon over Prime sign-up and cancellation for $2.5 billion, a $1 billion civil penalty plus $1.5 billion in refunds. The agency alleged that people were enrolled without clear consent and that leaving was made deliberately hard.
That is the balancing question for any default: who chose it, and what do they gain if you stay? A pension default set by law to help savers and a renewal setting chosen by a seller exploit the same habit. Researchers have also noted that defaults are less effective when people distrust whoever set them, which suggests the endorsement channel can be used up.
How big is the effect
The best overall estimate comes from a 2019 meta-analysis in Behavioural Public Policy by Jon Jachimowicz and colleagues, including Johnson. It pooled 58 default studies with 73,675 participants. On average, making an option the default raised its uptake by 0.68 standard deviations, a medium effect. In the studies with yes-or-no outcomes, that worked out at about 27 percentage points.
The average hides a very wide spread. Of the 58 results, 46 found a significant positive effect, ten found none and two found that the default backfired. Defaults worked best for consumer choices and less well for environmental ones, where people may hold firmer views. They were stronger when the default looked like a recommendation or like the existing state of affairs.
Do defaults work? 58 results
Direction of the effect in studies pooled by a 2019 meta-analysis.
Source: Jachimowicz JM, Duncan S, Weber EU, Johnson EJ, Behavioural Public Policy (2019), doi:10.1017/bpp.2018.43. Pooled n = 73,675; average d = 0.68.
Two limits matter. Most studies measure a choice made in the moment, often hypothetically, rather than what happens later. And the authors found heterogeneity so high, an I-squared of about 98%, that anyone setting a new default cannot know in advance whether it will produce a large effect, a small one or none.
What the evidence suggests
Defaults reliably change what people tick, and they change outcomes when the tick is the outcome, as with joining a pension. Where a long chain of events separates the form from the result, as with organ donation, switching to opt-out has produced little measurable gain in deceased donors and may have cost some living ones. Defaults also fix people at settings someone else picked, and stickiness is strongest among those with the least time or money. The useful question about any default is not whether it works, but for whom.
Questions people ask
Do opt-out laws increase organ donation?
The best before-and-after studies find little or no significant rise in deceased donors. Countries with high donation rates tend to have strong hospital systems as well as their laws.
Can my family still refuse under opt-out?
In practice, in most countries, yes. Families are asked, and their objection usually stops donation, which is one reason the law changes outcomes so little.
Does automatic enrolment work?
For membership, very well: participation roughly doubled in the US and UK studies. Whether the default contribution rate is enough for retirement is a separate question.
Why do people stick with defaults?
Changing takes effort, the default can look like advice, and people judge other options against what they already have.
Are defaults manipulation?
Every choice has a default, so one cannot be avoided. The concern arises when the default serves whoever set it more than the person choosing.
The short version
- In a 2003 online experiment, 82% of people stayed organ donors when donation was the default, against 42% when they had to opt in.
- Registering is not donating. Across six countries that switched to opt-out, deceased donors rose by a non-significant 7% while living donors fell by 29%.
- Pensions are where defaults clearly work: when one US firm enrolled new staff automatically, participation rose from 37% to 86%.
- The same inertia keeps people at whatever rate was chosen for them. Nearly two-thirds of those auto-enrolled stayed at the 3% default contribution.
- A meta-analysis of 58 studies found defaults shift choices by about 27 points on average, but results range from large to none. Ask who set the default and why.
This article summarises published research for general information. It is not financial, legal or medical advice, and it does not recommend any pension, provider or product. For decisions about your own pension contributions or organ donation wishes, check the official guidance where you live and consider speaking to a qualified, regulated adviser.
Further reading: Jachimowicz et al., Behavioural Public Policy (2019), for the meta-analysis of default effects. Madrian and Shea, Quarterly Journal of Economics (2001), for the pension study that started it. Güntürkün et al., PNAS Nexus (2025), for what happened when countries switched to opt-out.
- The Elements of Choice, Eric J. Johnson (2021). The researcher behind the donor chart on how the design of a choice shapes the answer. Clear and practical; as a co-author of the key studies, he is not a neutral judge of them.
- Nudge: The Final Edition, Richard H. Thaler and Cass R. Sunstein (2021). The updated version of the book that made defaults a policy tool, now with a reconsidered take on organ donation. Readable; it is a case for nudging rather than a balanced review.
- Sludge, Cass R. Sunstein (2021). A short book on the friction that keeps people stuck, from cancelling subscriptions to claiming benefits. Useful on the harmful side of defaults; light on hard numbers.
Sources
- Johnson EJ, Goldstein DG. Do defaults save lives? Science 302(5649):1338 (2003), doi:10.1126/science.1091721.
- Chandrashekar SP et al. Defaults versus framing: replications of Johnson and Goldstein (2003) and Johnson, Bellman and Lohse (2002). Meta-Psychology (2022).
- Kalkstein DA et al. Defaults are not a panacea. Behavioural Public Policy (2022), doi:10.1017/bpp.2022.24.
- Molina-Pérez A, Rodríguez-Arias D, Delgado J. BMJ Open 12(9):e057107 (2022).
- Abadie A, Gay S. The impact of presumed consent legislation on cadaveric organ donation. J Health Econ (2006); NBER w10604.
- Rithalia A et al. BMJ 338:a3162 (2009).
- Dallacker M et al. Opt-out defaults do not increase organ donation rates. Public Health (2024), doi:10.1016/j.puhe.2024.08.009.
- Güntürkün P et al. Crowding-out effects of opt-out defaults. PNAS Nexus 4(10):pgaf311 (2025), doi:10.1093/pnasnexus/pgaf311.
- Noyes J et al. BMJ Open 9(4):e025159 (2019).
- Hawkes N. Welsh opt-out law fails to increase organ donations. BMJ 359:j5659 (2017).
- Madden S et al. Anaesthesia (2020), doi:10.1111/anae.15055.
- O’Neill S et al. PLoS One (2024), doi:10.1371/journal.pone.0306541.
- Gardiner D et al. UK opt-out legislation: impact on organ donor registrations and consent (conference abstract). Transplantation (2025).
- Madrian BC, Shea DF. Q J Econ 116(4):1149–1187 (2001), doi:10.1162/003355301753265543.
- Choi JJ, Laibson D, Madrian BC, Metrick A. For better or for worse: default effects and 401(k) savings behavior. NBER w8651 (2001).
- Beshears J et al. Who is easier to nudge? Working paper (2016).
- Cribb J, Emmerson C. Int Tax Public Finance (2019), doi:10.1007/s10797-019-09565-6.
- Jachimowicz JM, Duncan S, Weber EU, Johnson EJ. Behav Public Policy (2019), doi:10.1017/bpp.2018.43.
- DWP. Workplace pension participation and savings trends of employees: 2009 to 2025 (30 July 2026).
- US Federal Trade Commission, Amazon Prime settlement, 25 September 2025.
- The Irish Times, 29 September 2026.
