Experiment E from "Neglect at Your Own Risk: Evidence on Risk-Taking Prevalence and Motives from the Field"

All participants complete the same insurance education, comprehension item, incentive explanation, plan menu, post-choice probability questions, liquidity question, and incentivized random-outcome procedure. The three conditions vary only in how the same objective damage distribution is described immediately before plan choice.

1. Separate damage-outcome probabilities condition: 96% chance of no damage, 1% chance of minor damage costing $750, and 3% chance of severe damage costing at least $2,500.

2. Any-damage and conditional-severity condition: 4% chance of any damage of at least $750 and, conditional on any damage, a 75% chance that damage is severe and costs at least $2,500.

3. Any-damage and severe-damage probabilities condition: 4% chance of any damage of at least $750 and 3% chance of severe damage costing at least $2,500.

Participants choose among a Basic Plan ($616 premium, $2,000 deductible), Medium Plan ($900 premium, $400 deductible), and Premium Plan ($1,020 premium, $250 deductible). Their plan choice and a randomly selected damage outcome determine a bonus between $0.00 and $1.54. The regular participation payment is $1.00.

2026-07-26

2026-08-31

An overinsurance indicator equal to 1 when the participant chooses the Medium or Premium Plan and 0 when the participant chooses the Basic Plan. The central comparison is the any-damage and conditional-severity condition versus the separate damage-outcome probabilities condition. The directional prediction is that the conditional-severity representation increases overinsurance.

This is an individually randomized online survey experiment. Participants see one condition and make one incentivized insurance-plan choice. The separate damage-outcome probabilities and any-damage plus conditional-severity conditions provide mathematically equivalent information about the same damage distribution. The any-damage plus severe-damage probabilities condition is a secondary diagnostic arm.

Not available

Qualtrics assigns participants using a 12-cell evenly presented randomizer: five cells assign the separate damage-outcome probabilities condition, five assign the any-damage and conditional-severity condition, and two assign the any-damage and severe-damage probabilities condition. The intended allocation is therefore 5:5:2.

Individual

No

600 individuals

600 individuals

The target analytic sample is 600 eligible completed participants after the preregistered exclusions: approximately 250 in the separate damage-outcome probabilities condition, 250 in the any-damage and conditional-severity condition, and 100 in the any-damage and severe-damage probabilities condition. Recruitment may continue until the target analytic sample is reached, subject to budget and platform availability.

Experiment_E_analysis_plan.docx