Games · free, no sign-up
The anchoring game
Eight estimates, each after a higher-or-lower question about a number far from the truth. See whether your answers leaned toward the number you were given.
In short
Anchoring is the pull of a number you have just considered on an estimate you make next, even when that number carries no information. This game asks eight estimation questions and puts a comparison number in front of each. Half of those numbers are set far above the true value and half far below, so leaning toward them can only make your estimates worse.
How the game works
Each question is a quantity computed when the page builds the round: a count of dots, a product, a compound sum, a count of primes. You first say whether the truth is higher or lower than the comparison number, then type your best estimate. The procedure, a comparison question followed by an estimate, is the one Tversky and Kahneman used in 1974 and Jacowitz and Kahneman refined in 1995. The questions here are new each round, so nobody has seen them before.
What the result shows
The result compares your estimates after high numbers with your estimates after low ones, both measured against the truth on a log scale, so a guess twice too big counts the same as one half too small. If your high-number estimates run well above your low-number ones, the numbers pulled you. Anchoring is one of the effects that replicated clearly in the Many Labs project, which reran classic studies in dozens of laboratories at once.
Sources
- Tversky and Kahneman (1974), Judgment under uncertainty: Heuristics and biases, Science
- Jacowitz and Kahneman (1995), Measures of anchoring in estimation tasks, Personality and Social Psychology Bulletin
- Klein and colleagues (2014), Investigating variation in replicability: A Many Labs replication project, Social Psychology
Text, questions and code on this page are original to MyTestAtlas. The cited studies describe the classic procedure; none of their items is reproduced.