A default gets described like it's one thing: whatever happens if nobody touches the form. It isn't. Whether a default exists at all, which direction it points, when it takes effect, and who it was actually computed for are four separate design decisions. Each one has its own real study behind it, and each one lands its cost on a different person.
The stakes are rarely trivial. The same lever, in its different settings, has decided how much of a paycheck gets saved for retirement, who ends up on an organ donor registry, and what a dying patient's advance directive says about their own care.
Five real settings of that lever follow, from a form that forces a real choice to one computed for the individual reading it. One of the field's most famous results, the country-level organ donation gap, has also just had its causal story complicated by a newer, better-designed study, worth reading before repeating the older number as settled.
The story in four parts
Switching a 401(k) or an organ-donor registry from opt-in to opt-out lifts participation by tens of percentage points, sometimes past 70, without changing what anyone is actually being offered.
The organ-donation gap came from comparing countries that already differ in a dozen ways. A newer study that tracked five countries before and after they actually switched found most of the promised lift wasn't there.
Forcing a real choice, with no default at all, moves behaviour too, and wording that forced choice around what's lost by picking wrong moves it further still.
A U.S. federal rule now lets a retirement plan's default fund shift by the saver's own birth year, a default built for the individual instead of guessed for everyone in the plan at once.
The first of those four decisions, whether a default exists at all and which direction it points, is the one most people mean when they say “the default effect.” Laid out as a line, it runs from a form that refuses to let anyone stay passive, to a default nobody has looked at again in years.
The middle two settings are where almost every famous default-effect finding lives, covered next. The two ends get discussed less, and turn out to be more interesting: one because it removes the lever entirely, the other because it's usually invisible until someone goes looking for how long a default has been sitting there unexamined.
Eric Johnson, Steven Bellman and Gerald Lohse ran two real online experiments that changed nothing but which way a checkbox pointed. Opted-in participants agreed to further contact 48.2% of the time. The same offer, worded identically, opted-out by default: 96.3% agreement. A third group, given no default at all and forced to tick a box either way, landed around 70%, meaning the default wasn't simply nudging people towards the choice they'd have made anyway. It was moving the decision itself. Default Effect covers the psychology behind why: implied endorsement, the effort of acting against whatever's pre-selected, and a mild aversion to losing the option currently held.
Brigitte Madrian and Dennis Shea studied one employer's 401(k) plan before and after it switched new hires from opt-in to automatic enrolment. Participation in the first six months went from roughly 37% to roughly 86%. Nobody's actual investment options changed. Only what happened if they did nothing did.
Germany requires an explicit opt-in to become an organ donor. Austria presumes consent unless a citizen opts out. Roughly 12% of Germans are registered donors, against a figure approaching 99% in Austria, a gap Eric Johnson and Daniel Goldstein's cross-country comparison put on the map in 2003.
A gap this size, between two countries that share a border, a language family, and a healthcare system, looks like the default doing all the work. It also looks exactly like the trap this site's own Not Testing Is Still a Bet describes: a comparison, not an experiment, carrying every other difference between Germany and Austria along with it, transplant infrastructure, family-consent practices, public trust in the medical system, none of it held constant.
A newer study finally separated those confounds from the default itself, the only way that's really possible: by tracking countries that actually switched. Martina Dallacker and colleagues followed deceased-donor rates in five countries, Argentina, Chile, Sweden, Uruguay and Wales, before and after each one moved from opt-in to opt-out. Averaged across all five, the switch produced no discernible change in the rate of new donors. Living-donor rates, meanwhile, fell significantly after the same switch, a real cost the cross-country comparison never had a way to show.
A parallel finding shows up in an entirely different domain. Raj Chetty and colleagues studied Danish retirement accounts that came in two versions: a subsidy requiring an active decision to open the account, and an automatic employer contribution requiring no decision at all. Each dollar spent on the subsidy raised total household saving by about one cent, because most people who claimed it simply moved money that was already accumulating somewhere else in their finances. The automatic employer contribution, deducted before it was ever received as spendable income, produced real new saving instead.
These aren't the same fight. Forcing an active decision can raise participation rates above opt-in's, because opt-in still leaves the very first form to procrastination. But raising participation isn't the same as creating new behaviour: an active decision is comparatively easy to satisfy on paper and then quietly offset elsewhere, in a way a payroll deduction someone never held in their hand isn't. On that second question, whether the resulting behaviour is genuinely new, opt-out doesn't just beat opt-in. For money fungible enough to move around, it beats a merely-forced choice too.
The one point on this line where inertia has nowhere to sit is a design that never installs a default at all. Gabriel Carroll, James Choi, David Laibson, Brigitte Madrian and Andrew Metrick found that mandating an active decision, new hires had to tick a real box before their onboarding paperwork would let them proceed, raised initial 401(k) enrolment by 28 percentage points over a standard opt-in process. Nobody was defaulted into anything. Procrastination simply had no default left to hide behind.
Punam Keller, Bryan Harlam, George Loewenstein and Kevin Volpp asked employees to actively check a box either accepting or declining a flu shot, with nothing pre-selected. 62% chose the shot, against 42% under a standard opt-in form offered to a comparable group.
A third group in the same study made the identical forced choice, but the decline option was worded around what it gave up: acknowledging a higher chance of getting the flu. Agreement rose again, to 75%, without a default ever entering the design.
“Enhanced” doesn't mean a default quietly returned. Nothing is pre-selected in either version. The second version borrows the exact mechanism this site's own Loss Aversion article names: a described loss carries more weight than an equivalent, unstated gain, even when the choice stays entirely in the decision-maker's hands.
An opt-out default doesn't have to take effect the moment someone signs up. Smart Defaults, this site's write-up of Richard Thaler and Shlomo Benartzi's Save More Tomorrow plan, covers what happens when the same lever is timed against a raise that hasn't landed yet instead of a paycheck someone's already counting on: an increase in savings that never has to feel like a cut, because there's nothing yet to feel a cut from.
The opposite question, whether a default ever gets revisited once it's set, is where a default can do quiet, compounding damage. Default Effect covers Brot-Goldberg and colleagues' study of Medicare Part D beneficiaries randomly assigned a starting drug plan: years later, most were still on it regardless of how badly it actually fit their prescriptions, a persistence that tracked passivity, not satisfaction.
One design answers that specific failure without ever asking the saver to act again. Since the Pension Protection Act of 2006, U.S. retirement plans have been permitted to set a target-date fund, keyed to a saver's expected retirement year, as the default investment for anyone who never actively chooses one. The fund keeps shifting its own mix from riskier to more conservative holdings as that date approaches, on a schedule set the day the saver was defaulted in, so the default keeps adjusting to something like the saver's real situation without a second decision from them.
The target-date fund above is “smart” in a specific, technical sense that has nothing to do with timing. N. Craig Smith, Daniel Goldstein and Eric Johnson use the same two words, smart defaults, for something different from this site's own Smart Defaults principle: not a default timed against a future gain, but one computed from data about the specific person receiving it, rather than copied from whatever setting works best for the average person in the pool. A target-date fund keyed to birth year is a mild version of this: one input, age, personalises an otherwise identical default across an entire workforce.
Smith, Goldstein and Johnson's own account of why any default carries this much force names three separate causes: it reads as an implicit recommendation from whoever set it, it exploits the same cognitive shortcuts and effort-avoidance covered in Default Effect, and it removes a real cost, the time and attention choosing would take, that a forced choice doesn't. Craig McKenzie, Michael Liersch and Stacey Finkelstein's experiments on the first of those three found people genuinely read a policy default as advice: told that a retirement plan's default contribution rate had been set high, participants inferred the plan's designers thought a high rate was the right one, and shifted their own stated preference towards it. Personalising a default doesn't remove that inference. A default visibly built around one person looks less like an administrative shortcut and more like advice meant specifically for them, aimed by whoever built the model that guessed it.