Trap/Cognitive Bias/No. 0037

Anchoring Bias

Anchoring bias, also called the anchoring effect, is a tendency for an early number to pull later estimates toward it, even when irrelevant. Described by Amos Tversky and Daniel Kahneman in 1974, it affects judgments even when people know the anchor is arbitrary.

Also called Anchoring Effect

a trap: easy to walk into

Evidence: Well established
Read: 6 min
Links: 9 connections
Useful when: Estimating time and cost · Forecasting · Money and investing · Negotiating

01You've seen this when…

in life
You see a used car listed for $18,000. Before checking comparable sales, you start working out how much less than $18,000 you should offer.
at work
The kickoff slide says six weeks. The team discovers dependencies across seven departments, but everyone’s revised estimate still lands around six to eight weeks.
out in the world
A city proposes a $100 annual parking permit. The hearing fills with suggestions for $80 or $120, before anyone estimates what the program needs to raise.

02The idea

Once a number enters a discussion, it can become the starting point for everyone else’s judgment. Later estimates lean toward it, even when it came from a guess, a bargaining tactic or something unrelated to the question.

That starting point is an anchor. Anchoring bias is the pull it exerts beyond what its information deserves. A seller’s $18,000 price can make $16,000 seem cautious, even if comparable cars sell for $13,000. You’ve moved away from the asking price, but it still organizes your thinking.

Your own information and estimates can also supply anchors. Last year’s spending, an early forecast or a number you generate yourself can do the same job. The influence can arise through consideration alone: merely judging whether a quantity is above or below a supplied number can affect your eventual estimate.

The useful test isn’t whether you used a starting point. It’s whether your answer would change substantially if someone supplied a different starting point with no better evidence.

03Why it happens

Different mechanisms can produce anchoring effects. Two well-studied routes can produce similar-looking results:

The number changes what comes to mind. When you consider whether a house could be worth a high asking price, you retrieve features that could justify it. The renovated kitchen comes to mind, and you consider how the location and garden could support that price. A low price makes different features easier to retrieve. Researchers call this selective accessibility. It resembles confirmation bias. Considering the anchor can steer your search regardless of whether you believe it.
Adjusting takes effort, and uncertainty gives the starting point more room. With some anchors, especially ones you generate yourself, you start with a known value and move away from it. You may stop once the answer feels plausible, before reaching the best-supported estimate. This is the classic anchoring-and-adjustment account. When you know little about the quantity, many answers seem possible. An initial number supplies structure. Relevant knowledge can reduce its influence, but professional experience doesn’t guarantee protection.

These routes matter because the remedy differs. More time may help with deliberate adjustment. When an anchor has already shaped which evidence you notice, its influence may persist even with more time.

04A worked example

In a 1987 study, Gregory Northcraft and Margaret Neale asked students and real estate agents to value a house. Participants inspected the house and received property information with an asking price that the researchers varied while keeping the property itself the same.

What it looks like An informed valuation. The agents have professional experience, and they receive information about a property they can examine directly. The asking price appears to be just one input among several.

What’s actually going on Changing the asking price shifts valuations, including those made by the agents. The house’s value stays the same as the number in the materials rises. Yet that number influences judgments made with substantial relevant information. The effect persists even among experts.

What would have helped A better procedure would separate the independent appraisal from the seller’s request. Estimate value from comparable sales and property features first, then record that estimate before inspecting the asking price. This is a practical safeguard whose effectiveness the study leaves untested. The study demonstrates the vulnerability; a guaranteed cure would require separate evidence.

05How to spot it

Every revision stays near the original guess. New information changes the explanation more than the estimate.
The debate stays focused on adjustments. People discuss adding 10% or subtracting two weeks while leaving the starting figure’s source unexamined.
The first number has a weak pedigree. It came from a seller, an outdated budget, a placeholder or someone’s quick impression.
Different starting figures produce different answers. Two groups given the same facts reach estimates that track the numbers they saw first.
Discounting replaces comparison. An offer feels attractive based on its discount from the asking price, regardless of how it compares with realistic alternatives.

06What to do instead

Estimate before exposure when you can. Write down your forecast, valuation or budget before seeing another person’s figure. In a team, have people submit estimates independently before discussion. Once they’ve all seen the same anchor, averaging their answers can leave its influence in place.
Build from relevant comparisons. Find observed outcomes, including actual sale prices and completed project durations. Use reference-class forecasting to replace an unsupported starting number with an estimate grounded in those outcomes. Choose comparable cases for their relevance to the estimate, whether they support or challenge your preferred answer.
Make the contrary case concrete. List facts that would make the anchor too high or too low. The consider-the-opposite strategy has reduced anchoring in some experiments. Changing which evidence you examine is more useful than simply reminding yourself to be unbiased.
Set negotiating limits from alternatives. Use market evidence to set a target and walk-away point that reflect your needs and your best alternative. Treat the other party’s opening offer as a proposal to evaluate against those limits.
Check the number’s origin. Identify whether you’re looking at a measured quantity, a guess, a negotiating position or a recycled assumption. The evidence stays weak even when presented confidently.

None of these makes you immune. The strongest practical move is often procedural: keep the questionable number out of the estimation process until you’ve built an independent basis.

07When it isn’t a bias

A starting number can supply useful information. Yesterday’s exchange rate is relevant to today’s rate. A recent sale of an almost identical apartment deserves weight. Using those figures can be appropriate. The bias is giving a number more influence than its reliability and relevance warrant.

Anchoring also differs from reference dependence. Reference dependence concerns how an outcome feels relative to a baseline: a salary can feel like a gain or a loss. Anchoring concerns the numerical estimate itself: what salary you think the role should pay. Both can occur together.

The framing effect changes judgments through different presentations of equivalent information. Anchoring works through a supplied value that pulls another value toward it.

Arbitrary numbers don’t always work, and the effect’s size varies with the task, knowledge and context. The familiar demonstration is not evidence that any number can control anyone’s judgment.

08Roots

Amos Tversky and Daniel Kahneman used a wheel of fortune to expose a surprisingly portable influence. The wheel displayed numbers from 0 to 100 but was arranged to stop at either 10 or 65. Participants then considered whether the percentage of United Nations members that were African countries was higher or lower than that number, before giving their own estimate. The median answers were 25% and 45%, respectively.

A wheel spin couldn’t supply evidence about UN membership. That was the point. In their 1974 article in Science, Tversky and Kahneman presented anchoring and adjustment alongside other shortcuts people use when judging uncertain quantities. Later researchers carried the question into property valuations, bargaining and forecasting, and investigated why an irrelevant number could survive apparently thoughtful reasoning.

One plausible but speculative account is that starting from an available value is usually economical: everyday estimates often build on useful prior information. The trouble comes when that habit accepts a poor starting point. This hypothesis concerns the tendency’s usefulness in ordinary environments; an evolutionary explanation would require further evidence.

09How solid is this?

Numerical anchoring is well supported by laboratory experiments and applied studies, including judgments by professionals. Its size and mechanisms vary across tasks; debiasing methods help in some settings but do not reliably eliminate it.

10Connections

Often confused with Reference Dependence
Countered by Outside View vs. Inside View, Reference-Class Forecasting, Consider-the-Opposite Strategy, BATNA
Part of Cognitive Bias
See also Confirmation Bias, Framing Effect, Default Effect

11Origin and sources

Described by Amos Tversky and Daniel Kahneman in their 1974 Science article, “Judgment under Uncertainty: Heuristics and Biases.”

Suggest an edit· Updated 2026-10-02