Behavioral task
Balloon Analogue Risk Task (BART)
Twenty balloons, $0.05 a pump, and a burst point you cannot see. What you do here is measured against what you said about yourself on the risk questionnaire.
- Balloons
- 20
- Time
- 3–5 min
- Max pumps
- 64
- Account
- Not needed
Before you start
- You get 20 balloons. Each pump adds $0.05to that balloon’s temporary bank.
- Collect banks the money and moves to the next balloon.
- Every balloon bursts at some point. If it bursts, that balloon’s bank is lost — the money you already collected is safe.
- The burst point is different for every balloon and you are not told what it is. That is the whole task.
The money is not real and there is no score to beat. Playing the way you actually would is the only thing that makes the result mean anything.
What this measures
Every questionnaire on this site has the same ceiling: it asks you to report on yourself. That report is assembled from memory and shaped by who you understand yourself to be, and the decisions it asks about are usually ones you made a while ago under conditions you no longer fully recall. It is useful information. It is not a record of behavior.
This task removes the self-report step. Each pump is a decision between a small certain gain and an unknown chance of losing the balloon’s whole bank, made with no information about where the burst point sits. There is no way to answer it as the person you would like to be — you either pump or you collect, twenty times over, and the pattern that emerges is what you did rather than what you would say.
What makes the result worth reading is the pair. Self-reported and behavioral risk measures correlate weakly, around r = .2 to .3 across large samples, which means the two genuinely capture different things and a difference between them is the expected outcome rather than a contradiction. The direction of that difference is the interesting part: it tells you whether your account of yourself runs bolder or more careful than your play, and neither measure alone can show you that.
Where this version departs from the published task
The structure follows Lejuez et al. (2002): pump-by-pump inflation, a hidden burst point drawn uniformly, a bank that clears on collection and is lost on a burst, and scoring on the average pumps of balloons you banked rather than all balloons. Burst rounds are excluded from that average for a specific reason — a burst cuts the decision short, so the round never reveals where you would have stopped, and including it would push the score down hardest for exactly the boldest players.
Two parameters differ. This version uses 64 maximum pumps rather than 128, and 20 balloons rather than 30, so a session fits in a few minutes in a browser. The money is also simulated. Both changes matter for interpretation: real incentives change risk behavior, and different parameters produce different averages.
That is why no percentile appears anywhere in your result. The published norms belong to the published parameters, and reporting them against these ones would be a fabricated comparison. Your average is placed against this task’s own risk-neutral optimum instead — with a uniform burst point, expected earnings peak at exactly half the maximum, so that is a real anchor derived from the task rather than a borrowed one.
Common questions
What is the Balloon Analogue Risk Task?
A behavioral measure of risk taking, introduced by Lejuez and colleagues in 2002. You inflate a balloon one pump at a time; each pump adds money to a temporary bank, and you can collect it at any point. Every balloon bursts eventually and a burst loses that balloon's bank. Because the burst point is hidden and varies per balloon, each pump is a real choice between a certain small gain and an unknown loss — which is what makes it a measure of behavior rather than of self-image.
How is this different from a risk tolerance questionnaire?
A questionnaire asks you to summarise how you behave. This asks you to behave. The distinction matters because the two do not agree closely: across large samples, self-reported and behavioral risk measures correlate at roughly r = .2 to .3. Neither is the true one — a questionnaire reaches decisions this task cannot simulate, like career and money choices, while the task captures what you do when a loss is immediate and the odds are opaque.
Is my score compared against other people?
No, and deliberately so. Published norms exist for the original 128-pump version with its own payout structure and balloon count; this version uses 64 pumps and 20 balloons, so quoting those norms here would be inventing a comparison. Instead your average is placed against this task's own risk-neutral optimum — the stopping point that would earn the most on average, which for a uniformly distributed burst point is exactly half the maximum.
Does the money mean anything?
No. It is not real and there is no leaderboard, which does weaken the task relative to a lab study paying real cash — incentives change behavior and it would be dishonest not to say so. What it still captures is how you respond to gain and loss framing under uncertainty, and if you play as you actually would rather than trying to hit a number, the pattern in your session is real.
Can I take it more than once?
Yes, and the burst sequence is different each time. Be aware that a second session is not independent of the first: once you know roughly how far balloons go, you are no longer estimating blind. The first session is the cleanest measure, and later ones say more about how you update on experience than about your baseline appetite for risk.
Pair it with the questionnaire
Risk Persona separates financial risk appetite from physical thrill-seeking and impulse control. Its financial score is the one this task is set against, and having both is what produces the comparison.
Take Risk Persona →