The Default Effect
Eight real-world policy scenarios across four domains. Each presented twice — once under an opt-in default, once under an opt-out default. Same population, same choice, only the pre-checked box differs. You predict take-up on a 0-100 slider. WIZ measures the gap between your predictions and forty years of choice-architecture data.
“The same default produces nearly opposite distributions across populations whose stated preferences are almost identical.”Johnson & Goldstein (2003) Science vol 302
Johnson & Goldstein (2003) compared 11 European countries with effectively identical demographics and almost identical attitudes toward organ donation. Effective consent rates split almost perfectly along default-policy lines. The four explicit-consent countries averaged 15%. The seven presumed-consent countries averaged 97.4%. Germany sat at 12%; neighboring Austria sat at 99.98%. The single mechanical fact of which box was pre-checked produced a gap of 80-plus percentage points on a decision people insist is among the most personal they will ever make.
Madrian & Shea (2001) QJE vol 116 reported 401(k) participation jumping from 49% to 86% when a US firm switched from opt-in to auto-enrollment. Pichert & Katsikopoulos (2008) JEP vol 28 reported green-electricity retention going from ~1% under conventional-default to ~94% under green-default in adjacent German utilities. Chapman Li Colby & Yoon (2010) JAMA vol 304 reported a 12-point flu-shot gap on a single email rewrite.
Meta-analytic status. Jachimowicz Duncan Weber & Johnson (2019) Behavioural Public Policy vol 3 pooled 58 studies across organ donation, retirement, environmental, marketing, and end-of-life domains and reported a mean effect of d=0.68 (medium to large). The effect replicates across decades, continents, and domains.
Lay prediction. Johnson Bellman & Lohse (2002) MIS Quarterly vol 14 and Smith Goldstein & Johnson (2013) JMR vol 50 documented that lay subjects systematically under-estimate default-effect magnitudes by 20-40 points. Sunstein (2013) Yale Law Journal vol 122 names the meta-bias: people believe their own choices are principled and stable, see other people as nudge-able. The double projection is the engine of the bias.
The exercise is not a test of stated values. The values are mostly real. The exercise measures how much weight you give to the box that has been pre-checked on the form. The 40-year answer is: nearly all of it, in the domains where defaults are the policy.