Active choosing
Requiring people to make a deliberate choice rather than defaulting them.
What it means
Active choosing is a choice-architecture approach that removes the default and requires people to make an explicit decision, so they must deliberately opt in or out rather than drifting into a pre-set option. The mechanism is that by forcing engagement it neutralizes the inertia and implied-endorsement effects that make defaults so powerful, eliciting people's genuine preferences instead of their path of least resistance. It is most appropriate when preferences are heterogeneous so no single default would suit everyone, when the decision is consequential enough to warrant attention, and when a default might feel manipulative or paternalistic. There are trade-offs and boundary conditions: forcing a choice imposes cognitive cost and can backfire when people are uninformed, overwhelmed, or simply unready to decide, in which case a well-chosen default may serve them better. A refinement, 'simplified active choosing,' offers a recommended option while still requiring an explicit selection. It matters in retirement enrollment, organ donation, privacy settings, and consent design, where it is often promoted as a more autonomy-respecting middle path between hard defaults and pure free choice.
What the evidence shows
The anchor study is Carroll and colleagues (2009): an employer made new hires return a form stating whether they wanted to enroll in the 401(k). Enrollment rose roughly 28 percentage points over standard opt-in, and the savings distribution reached after three months would have taken about thirty months to reach otherwise. Keller and colleagues' field trials tested the enhanced variant — opt-in against a loss-framed forced choice, with no neutral active-choice arm — and roughly doubled sign-ups to automatic refills (32 percent against 15.7 percent by phone; 21.9 percent against 12.3 percent on the web). Disenrollment was equivocal by the authors' own account — identical to opt-in in the phone study, significantly higher in the web study (15.2 percent against 10.2 percent) — but net of dropout the enrolled population still grew 16.7 percent against 2.3 percent, so the gain largely stuck. Their lab work is where the unenhanced version earns its keep: the basic active-choice condition made people tick "I will get a flu shot this Fall" or "I will not get a flu shot this Fall," and a second study, using reminder wording, found anticipated regret partially mediated the effect — people price a branch a skipped opt-in link lets them ignore. The record is not uniform, though. A randomized trial of the same enhanced variant in Malta's colorectal screening program (n = 8,349) found no overall gain, only a subgroup effect among men. Read together, the doubling results all come from the enhanced variant; the field evidence for plain, neutral active choosing rests on Carroll and on Beshears below.
There is almost always an implicit default
Pure active choosing is rarer than the label suggests. Someone always fails to answer, and whatever happens to them is a default, however unadvertised. Beshears and colleagues (2021) studied a scheme making a choice between prescription home delivery and pharmacy pick-up a condition of insurance eligibility: 42 percent chose delivery, 39 percent chose pick-up, and 19 percent chose nothing and got pick-up without the subsidy. The sorting was partly rational — those with the smallest subsidies at stake skipped most — but the non-choosers also skewed younger, lower-paid and less adherent to their medications historically, so silence carried signal and noise both. The lesson holds anyway: dropping the visible default shifts the cost of inattention from the designer, who used to pick one and own it, onto whoever does not reply.
Limits and caveats
Carroll and colleagues' model states the boundary condition precisely: an active decision beats a well-chosen default only when the propensity to procrastinate is strong and preferences are genuinely heterogeneous. Where one option suits nearly everyone, the default wins, and defaults pull hard on average (d = 0.68, 95% CI [0.53, 0.83], across 58 studies in Jachimowicz and colleagues' meta-analysis, though with wide variation and a few nulls and reversals), so dropping one forfeits a lot. Sunstein's sharper objection is that requiring a choice is itself paternalistic: some people know they are uninformed and would rather delegate, and forcing them overrides that preference. Carroll and colleagues grant the point in their own abstract — financial illiteracy, they note, favors default enrollment over active decision enrollment. "Enhanced" active choosing goes further still, since loading the loss frame onto one branch buys compliance by surrendering the neutrality that justified removing the default.
Using it in practice
Bolt the question to a step the person must complete anyway, because a standalone prompt gets dismissed. Decide the fallback for non-responders before launch, write it down, and watch who lands there: if the people skipping are the ones it matters most to, the design has failed. Keep the branches symmetrical unless you will defend the tilt in public. When preferences are not actually heterogeneous, admit it and set a default. Simplified active choosing is the usual middle path, and its recommendation should be labelled rather than smuggled in through the wording of the options.
Examples
A form that won't submit until you actively select a retirement-contribution rate.
Instead of a pre-ticked box, an app refuses to open until you tap either allow tracking or don't allow. Nobody drifts into sharing, and no option is quietly recommended.
Made to pick an energy tariff from nothing, many customers stall and stay on the worst rate. Naming a recommended tariff they must still select actively gets them decided without pretending they had a view.
A benefits provider asks the refill question inside checkout, a step the member has to finish anyway, instead of emailing it as a standalone prompt. The same question in an email is a question most people never open.
A cloud console refuses to create a database until the engineer picks a backup schedule. No workload silently inherits a cadence nobody reviewed, because no single cadence suits every workload.
First described in Sunstein; Carroll et al. (2009).
Key references
- Stoffel, S. T., Bombagi, M., Kerrison, R. S., von Wagner, C., & Herrmann, B. (2022). Testing enhanced active choice to optimize acceptance and participation in a population-based colorectal cancer screening program in Malta. Behavioral Medicine, 48(3), 141-146. doi.org/10.1080/08964289.2020.1828254
- Beshears, J., Choi, J. J., Laibson, D., & Madrian, B. C. (2021). Active choice, implicit defaults, and the incentive to choose. Organizational Behavior and Human Decision Processes, 163, 6-16. doi.org/10.1016/j.obhdp.2019.02.001
- Jachimowicz, J. M., Duncan, S., Weber, E. U., & Johnson, E. J. (2019). When and why defaults influence decisions: a meta-analysis of default effects. Behavioural Public Policy, 3(2), 159-186. doi.org/10.1017/bpp.2018.43
- Sunstein, C. R. (2014). Choosing not to choose. Duke Law Journal, 64(1), 1-52. scholarship.law.duke.edu/dlj/vol64/iss1/1/
- Keller, P. A., Harlam, B., Loewenstein, G., & Volpp, K. G. (2011). Enhanced active choice: A new method to motivate behavior change. Journal of Consumer Psychology, 21(4), 376-383. doi.org/10.1016/j.jcps.2011.06.003
- Carroll, G. D., Choi, J. J., Laibson, D., Madrian, B. C., & Metrick, A. (2009). Optimal defaults and active decisions. Quarterly Journal of Economics, 124(4), 1639-1674. doi.org/10.1162/qjec.2009.124.4.1639