The Sunk Ride

English | 中文

A behavioral-economics simulator of shared-bike choices. The on-screen title is「共享单车实验模拟器」: twenty days of weather, distance, luggage, and time pressure, used to study overconfidence, payment-plan rationality, and the sunk-cost effect.

The Sunk Ride names the trap the experiment is built around: a prepaid package that is already spent, yet still pulls you onto the bike.

Repository: https://github.com/Lynnyyang/The-Sunk-Ride

---

What this is

A client-only web app. You set a participant and group, then generate 20 days of ride / no-ride decisions from a logistic-style probability. Results include a summary (predicted rides (E) vs actual rides, overconfidence (E-A), tokens, cash payoff) and a day-by-day table that can be exported as CSV.

This is simulated data for teaching and design, not field observations.

---

Design in brief

Item	Value
Package (plan A)	15 tokens
Pay-per-ride (plan B)	1 token per ride
Starting endowment	30 tokens
Token-to-cash	0.5
Break-even	15 rides
Days	20 scenarios (shuffled)

Groups

1. Standard package — sunk cost is not shown each ride (“this ride is free”).
2. Cost-visible package — each ride shows the amortized sunk cost (15/E).
3. Control — pay-per-ride only (plan B).

Hypotheses on the home page: H1 overconfidence; H2 purchase rationality; H3/H4 sunk cost.

Ride probability starts from (E/20), then weather, long distance, time pressure, heavy luggage, and a sunk-cost factor (plan A, groups 1–2) shift it.

---

Stack

Vite 5, TypeScript, React 18, Tailwind CSS, shadcn/ui. Dev server port 8080.

---

Run locally

git clone https://github.com/Lynnyyang/The-Sunk-Ride.git
cd The-Sunk-Ride
npm install
npm run dev

npm run build
npm run preview
npm run lint

No backend, no API keys.

---

Layout

src/
  pages/Index.tsx
  components/SimulationForm.tsx
  components/ResultsDisplay.tsx
  lib/experimentData.ts      # 20 scenarios, prices
  lib/simulationEngine.ts    # ride probability and payoffs
  types/experiment.ts

---

License

MIT