Changelog · how the belief was versioned

Risk Compensation

Risk compensation is the tendency for people to adjust their behavior toward greater risk-taking when a safety measure makes them feel more protected, partially offsetting the measure's intended benefit.

# README · how it works

The proposed mechanism is a perception-action feedback loop: a safety measure lowers perceived risk (or raises perceived margin of safety), and people 'spend' part of that margin by behaving more boldly (driving faster, following closer, foregoing condoms), so realized risk reverts partway toward its prior level. The strong 'risk homeostasis' version (Wilde 1982) posits an internally regulated TARGET level of risk that people actively defend, implying near-complete offset. The weaker 'risk compensation' version only claims partial, variable adjustment. Whether compensation appears is thought to depend on whether the safety measure is visible/salient to the user, perceived as effective, motivationally relevant, and under the person's behavioral control.

v0.1Sam Peltzman · 1975initial release

+First shipped: The Effects of Automobile Safety Regulation. When a safety intervention (seat belts, helmets, ABS, PrEP, masks) lowers perceived risk, people may take more risk in response, eroding some of the safety gain; the strong form (risk homeostasis) claims people hold a fixed target risk level so benefits are fully offset. The evidence is genuinely domain-dependent: there is reasonable support for partial offsetting in road safety (e.g., redistribution of harm toward pedestrians/cyclists, ABS behavioral adaptation), but multiple systematic reviews find little or no behavioral compensation for bicycle helmets, COVID-19 masks, or PrEP. The original strong "complete offset" and homeostasis claims are widely regarded as overstated and poorly supported.

v1.1Sam Peltzman · 1975breaking change20%

−Peltzman auto-safety regulation analysis. Argued offsetting driver behavior was 'virtually complete,' so mandated safety devices did not reduce total highway deaths and shifted fatality risk toward pedestrians and cyclists.retracted / failed · National US accident/fatality time-series data

v1.2Sagberg, Fosser, Saetermo (Sagberg et al.) · 1997no stat

~Munich/Norwegian taxi ABS & airbag field study. Found evidence of behavioural adaptation consistent with risk compensation for ABS-equipped vehicles (e.g., reduced safety margins); cited alongside the earlier Aschenbrenner & Biehl Munich taxi ABS study reporting slightly higher crash rates for ABS cabs.effect revised down · Taxi-driver fleet (field study)

v1.3Maryam Esmaeilikia, Machar Rana, et al. (Esmaeilikia et al.) · 2019+23

+Esmaeilikia bicycle-helmet systematic review. Eighteen studies found no supportive evidence that helmet use was positively associated with risky behaviour; many found helmet use associated with SAFER cycling. Only two studies (from one research group) supported the hypothesis.added support · ~23 studies reviewed

v2.0Michael W. Traeger, Sophia E. Schroeder, Edwina J. Wright, et al. · 201895% · +17

+Traeger PrEP sexual-behavior meta-analysis. Found increased STI diagnoses among PrEP users (notably rectal STIs), some evidence consistent with behavioral risk compensation, though condom-use data were too heterogeneous to pool.added support · 17 studies, 6,671 participants

v1.5Julia L. Marcus, David V. Glidden, Kenneth H. Mayer, et al. · 201395% · +2408

+iPrEx trial risk-compensation analysis. No evidence of sexual risk compensation; no increase in condomless receptive anal intercourse among those believing they received active drug.added support · 2,408 participants (of 2,499 enrolled)

beta

Release stability — mixed. Strongly domain-dependent. Partial offsetting is reasonably supported in road safety (Peltzman-type redistribution of harm; ABS behavioral adaptation in taxi fleets), but the strong 'risk homeostasis / complete offset' claim is widely rejected and Peltzman's original empirics are now considered weak (Peltzman himself called the early work 'primitive'). Multiple systematic reviews find little or no risk compensation for bicycle helmets (Esmaeilikia 2019), COVID-19 masks, and PrEP behavior in RCT settings (iPrEx 2013); however, observational PrEP studies (Traeger 2018) show increased STIs consistent with some compensation. The honest reading: the phenomenon is real but partial, inconsistent, often absent, and frequently overstated.

Issues filed in production

#001 open · 1975
US 1960s automobile safety mandates and pedestrian/cyclist fatalities
Peltzman (1975) documented that after federally mandated auto safety devices, occupant fatality reductions were accompanied by increased non-occupant (pedestrian and cyclist) deaths, the empirical basis of the 'Peltzman effect.' The magnitude remains disputed but the redistribution-of-harm pattern is the canonical documented case.
#002 open · 1997
Munich/Norwegian taxi anti-lock brakes (ABS) field studies
Field studies of taxi fleets equipped with ABS found behavioural adaptation (reduced safety margins, comparable or slightly higher crash rates for ABS-equipped cabs), a frequently cited real-world instance of risk compensation to a vehicle safety technology.
#003 open · 2018
Increased STIs among MSM PrEP users
A meta-analysis of 17 studies (6,671 participants) found higher STI diagnoses among MSM taking HIV PrEP (any STI OR 1.24; rectal chlamydia OR 1.59), interpreted by some as behavioral risk compensation following reduced perceived HIV risk, though RCT-embedded analyses (iPrEx) found no such effect.

Recommended migration

Demand domain-specific behavioral evidence before accepting a compensation claim, and distinguish partial offset (common, still leaves net benefit) from complete offset (rare, mostly unsupported). Check whether adopters self-select on safety attitudes (confound) and whether the measure is even visible to the user; use outcome data (injuries/deaths/infections), not just intuition, to size the net effect.

Systematic reviews of bicycle helmets (Esmaeilikia 2019) and COVID-19 masks found little/no compensation, and the RCT-embedded iPrEx analysis found none for PrEP, while net safety benefits persisted; even where offset exists (auto safety, ABS) it is partial, not complete.

Catch it in the act

Watch for the argument 'making X safer will just make people reckless, so it won't help' used to oppose a safety measure. The claim is plausible but empirically inconsistent: ask which domain, whether the safety measure is salient to the user, and whether there is actual behavioral data showing offset (often there is not, or the safety-conscious adopters behave MORE cautiously). Treat 'complete offset' or 'no net benefit' claims as the strong, usually-unsupported homeostasis form.

Dependencies

Peltzman EffectchildThe specific economic instantiation of risk compensation in safety regulation (offsetting driver behavior), named after Sam Peltzman's 1975 study.
Risk Homeostasis TheorysiblingWilde's strong form claiming a defended fixed target risk level and near-complete offset; risk compensation is the weaker, partial version. Often conflated.
Moral Hazardeasily-confusedEconomic concept where insulation from consequences (e.g., insurance) increases risk-taking; mechanistically related but framed around incentives and contracts rather than perceived physical safety.
Optimism Biasmechanistically-linkedUnderestimating one's own risk can drive the boldness that compensation predicts, but is a distinct perceptual bias.
Safety in numbers / behavioral adaptationsiblingBroader 'behavioral adaptation' to technology change encompasses risk compensation as one direction; adaptation can also increase safety.

Open the same case on your own draft.