Anchoring
A judgment bias where the first number we encounter — even an obviously irrelevant one — pulls our subsequent estimates toward it.
Tversky and Kahneman demonstrated anchoring in their 1974 Science paper with a now-classic experiment: participants spun a rigged wheel that landed on either 10 or 65, then estimated the percentage of African countries in the United Nations. Those who saw the higher anchor gave estimates around 45%; those who saw the lower one averaged 25%. Nobody believed the wheel was informative, but the number still moved the answer. Anchoring shows up in salary negotiations (the first number on the table sets the bargaining range), real-estate listings (the asking price shapes what buyers consider fair), restaurant menus (a high-priced entrée at the top makes the rest feel reasonable), and courtroom sentencing recommendations. The effect persists even when the anchor is randomly generated, when people are explicitly warned about it, and when they are paid for accuracy. The cleanest defense is to do the math before you see anyone else's number.
Frequently Asked Questions
Why does anchoring work even when the anchor is obviously irrelevant?
Because the brain anchors first and adjusts second — and the adjustment is almost always insufficient. Once a number is in working memory it becomes the starting point for the estimate, and the conscious correction step runs out of effort before it has moved you far enough away.
How do you avoid anchoring in negotiations?
Two ways: make a defensible first offer yourself so you set the anchor, or — if the other side moves first — explicitly ignore their number, write down what you think the fair price is from independent reasoning, and only then compare. Holding two anchors blunts the pull of either.
Is anchoring the same as priming?
They overlap but are not identical. Priming is the broader phenomenon of one stimulus making related concepts more accessible. Anchoring is a specific judgmental form: a numerical value you have just seen biases the magnitude of your next estimate, even when the value is irrelevant to the question.