Ask somebody whether the tallest redwood tree is taller or shorter than 1,000 feet.
Then ask them to estimate the height of the tallest redwood.
Ask another person whether it is taller or shorter than 180 feet.
Then ask the same estimate.
The two groups will tend to produce different answers even though the first number was not evidence about the tree.
The first number entered the reasoning process and pulled later judgment toward itself.
This is anchoring.
It is one of the most famous findings in behavioural decision research because it reveals something uncomfortable:
A number can influence judgment before it earns the right to matter.
Quick Read
Anchoring is the tendency for an initial value or reference point to exert disproportionate influence on later estimates and decisions.
The American Psychological Association defines anchoring bias as giving excessive weight to an initial value or starting point when forming later judgments.
Daniel Kahneman’s Nobel lecture discusses anchoring as part of the classic heuristics-and-biases research conducted with Amos Tversky.
Anchors can come from:
- the first price mentioned;
- a previous year’s budget;
- an initial forecast;
- a list price;
- a salary history;
- a target;
- a random number;
- a colleague’s estimate;
- a suggested answer;
- a headline statistic.
The central question is:
How much of this estimate came from evidence—and how much came from the first number that happened to enter the room?
The One-Sentence Answer
Anchoring works when an initial reference value narrows the range of later adjustment, causing judgments to remain closer to the starting point than the available evidence alone would justify.
The Anchoring Chain
initial value appears → attention and plausible range shift → evidence is interpreted relative to anchor → adjustment occurs → adjustment stops too early → final judgment remains pulled toward start
The important point is not that people never adjust.
They often do.
The problem is that the adjustment is frequently insufficient.
Anchoring Is Not a Default
A default determines what happens if no active choice is made.
An anchor influences the active choice itself.
If a pension contribution automatically starts at 5%, that is a default.
If a person is asked whether 5% is too high or too low and then chooses 6%, the initial 5% may also act as an anchor.
One mechanism acts through inaction.
The other acts through reference dependence in judgment.
See How The World Works | Defaults.
Anchoring Is Not Framing
Framing changes how equivalent information is described.
Anchoring changes judgment through a starting reference value.
The two can appear together.
A “$1,000 regular price, now $600” label frames $600 as a discount and anchors the buyer on $1,000.
But the mechanisms remain distinguishable.
The Classic Wheel-of-Fortune Experiment
In one famous Tversky and Kahneman experiment, participants first encountered a number generated by a wheel of fortune, then answered a numerical general-knowledge question.
The random number influenced later estimates even though it contained no relevant evidence.
This showed that anchors do not need to be informative to exert influence.
Irrelevant numbers can become psychologically sticky simply because they are available when the estimate is constructed.
Why Anchors Work
Researchers have proposed several mechanisms.
One is insufficient adjustment: people begin from a starting value and move away, but stop before reaching the estimate they would have made independently.
Another is selective accessibility: the anchor makes anchor-consistent information easier to retrieve.
If asked whether Gandhi was older than 120 when he died, a person may begin considering evidence consistent with great age.
If asked whether he was older than 35, different information becomes salient.
Anchoring therefore may change not only the arithmetic starting point but also which evidence the mind searches for.
The First Price in a Negotiation
A seller says $100,000.
The buyer planned to offer $60,000.
After hearing $100,000, $75,000 begins to feel like a reasonable compromise.
The first number has changed the bargaining field.
This is why first offers can matter.
But the anchor is not the whole negotiation.
Outside options, deadlines, liquidity and bargaining power still determine what is feasible.
See How The World Works | Bargaining.
List Prices Are Anchors
“Was $499. Now $299.”
The $499 number tells the customer what comparison to make.
If the product was never realistically worth $499, the anchor can still make $299 feel inexpensive.
This is why reference pricing is regulated in many jurisdictions: a fictional comparison price can mislead consumers even when the final price itself is clearly displayed.
Anchoring and Price Discrimination
Pricing systems can use anchors to influence perceived value while using segmentation to extract different willingness to pay.
A premium tier can make the middle tier look affordable.
A high list price can make a personalised discount appear generous.
Anchoring changes perception.
Price discrimination changes the price architecture across buyers.
See How The World Works | Price Discrimination.
Budget Anchoring
Organisations often build next year’s budget from this year’s budget.
Last year becomes the anchor.
“Add 3%.”
“Cut 5%.”
The process is efficient.
It can also preserve obsolete allocations.
A programme that should disappear receives 95% of last year’s funding because last year’s existence became the starting point.
Zero-based budgeting tries to weaken this anchor by requiring activities to justify themselves from a lower starting assumption.
Forecast Anchoring
A team forecasts revenue at $50 million.
New evidence arrives.
The market weakens badly.
The revised forecast becomes $47 million.
Did the team update enough?
Or did $50 million remain the centre of gravity simply because it was the first formal number?
Forecasting systems need deliberate re-anchoring when the world changes regime.
Forecasts Can Become Social Anchors
Once a forecast is published, people coordinate around it.
Budgets are built.
Targets are set.
Contracts are signed.
The forecast stops being merely an estimate and becomes a reference point inside the organisation.
Changing it now creates political and operational consequences.
Anchoring becomes institutional.
Historical Prices Anchor Markets
A homeowner remembers that the neighbour sold for $2 million.
The market has fallen.
The owner still refuses $1.7 million because $2 million has become the reference point.
This can slow transaction volume because sellers adjust more slowly than market-clearing prices.
Anchors can therefore create price stickiness.
Salary History
A candidate previously earned $60,000.
The new role may be worth $90,000 based on responsibilities and market rates.
If negotiation begins from the old salary, the candidate can remain anchored to a number produced by a different job in a different context.
This is one reason some jurisdictions restrict salary-history questions.
A historical number can reproduce historical inequality when it becomes the reference point for a new decision.
Sentencing and Legal Judgment
Research on legal decision-making has examined how suggested sentences and numeric demands can influence later judgments.
This matters because experts are not automatically immune.
Professional knowledge can reduce error without eliminating anchoring completely.
High-stakes systems therefore benefit from structured independent assessment before exposure to potentially contaminating anchors.
Medical Estimates
An initial diagnosis, probability or test result can anchor later reasoning.
Once a case is labelled “migraine,” later symptoms can be interpreted through that frame even when new evidence points elsewhere.
This is not purely numerical anchoring, but the same starting-point dependence appears in diagnostic reasoning.
Good clinical practice uses differential diagnosis and deliberate reconsideration to reopen the problem.
Anchoring in Education
A teacher sees a student’s previous grade before reading a new essay.
The grade can become an anchor.
A parent hears “average student” and interprets later performance relative to that label.
A student scores 55 on the first test and begins to treat 55 as identity rather than evidence from one assessment.
Anchors can become ceilings.
This is why fresh work sometimes deserves blind marking or independent review.
The First Mock Examination
A student scores 42 in an early mock exam.
The number is useful.
It locates current performance.
It becomes dangerous when every later target is adjusted around 42.
“Maybe 50 is realistic.”
But the 42 may have come from missing one major topic that can be repaired quickly.
A diagnostic system should explain the mark, not worship it.
Anchoring and Legibility
Legible systems generate official numbers.
Scores.
Rankings.
Benchmarks.
Budgets.
Once formalised, these values can anchor future decisions because they look authoritative.
See How The World Works | Legibility.
Anchoring and Winner’s Curse
An acquisition team establishes an early valuation of $500 million.
Competitive bidding begins.
Winning should cause a downward informational update if value is uncertain and common.
But the original $500 million estimate can anchor the team and prevent sufficient adjustment.
Anchoring and Winner’s Curse can therefore reinforce one another.
See How The World Works | Winner’s Curse.
Anchoring and Common Knowledge
An anchor becomes more powerful when everybody knows everybody else is using it.
A central forecast becomes the basis for budgets.
A salary benchmark becomes the market reference.
A published property valuation becomes the number every negotiator cites.
The anchor stops being merely cognitive and becomes coordinative.
See How The World Works | Common Knowledge.
Anchoring and Path Dependence
The first budget affects the second.
The second affects the third.
A starting number can therefore become embedded in institutional history.
The organisation forgets why the original number existed.
It remains because every later adjustment starts from it.
Anchoring can become path dependence expressed numerically.
See How The World Works | Path Dependence.
Anchoring and Bargaining Power
Anchors influence perceptions of reasonable outcomes.
Bargaining power determines which outcomes are actually enforceable.
A weak party can set an ambitious anchor and still lose if the other side has a much stronger outside option.
Do not confuse cognitive influence with structural power.
Anchoring and Price Discrimination
Sellers can present a high reference price before offering a lower targeted price.
The segmentation changes who receives which price.
The anchor changes how each price feels.
These mechanisms can work together but should be analysed separately.
Expertise Helps Less Than People Expect
Experts know more.
That gives them better evidence for adjustment.
But experts still reason from available starting points.
Anchoring research across domains has repeatedly shown that expertise does not guarantee immunity.
Good professional systems therefore use process, not confidence, to reduce anchoring.
Debiasing: Estimate Before Exposure
The strongest practical defence is simple.
Make an independent estimate before seeing the anchor.
Before hearing the seller’s price, estimate market value.
Before seeing the manager’s forecast, calculate your own.
Before reading previous grades, mark the new work.
Independent first-pass judgment preserves information that would otherwise be contaminated by the starting point.
Use Several Anchors
If one anchor narrows thought, deliberately introduce alternatives.
Best case.
Base case.
Worst case.
Comparable transaction A.
Comparable transaction B.
Cost-based valuation.
Income-based valuation.
Several independent reference points reduce dependence on one arbitrary start.
Reason From Components
Instead of asking, “Is $10 million reasonable?” rebuild the estimate.
How many units?
At what price?
With what margin?
Over what horizon?
Component reasoning reduces the anchor’s ability to define the total directly.
Use Outside Views
Inside-view estimates begin from the details of the current case.
Outside views ask what happened in comparable cases.
If your construction project is estimated at 18 months, what is the distribution of completion times for similar projects?
If your startup forecast assumes rapid growth, how often do comparable startups achieve it?
Reference classes introduce evidence that can pull judgment away from an internally generated anchor.
Re-Anchor After Regime Change
A pandemic.
A war.
A new technology.
A regulatory break.
A structural demand collapse.
When the generating process changes, yesterday’s average can become a dangerous anchor.
The model should be rebuilt rather than incrementally adjusted forever.
The Anchoring Audit
- Identify the first number. What entered the reasoning process first?
- Ask whether it contains evidence. Is the anchor relevant or arbitrary?
- Make an independent estimate. Can you calculate before seeing other views?
- Decompose the problem. Build from components rather than total reference values.
- Use multiple comparables. Avoid dependence on one reference case.
- Check the outside view. What do comparable cases usually do?
- Ask what evidence would justify moving far away. Do not adjust timidly by habit.
- Check institutional anchors. Previous budgets, targets, scores and forecasts can be sticky.
- Check negotiation structure. Is a first offer shaping perceived fairness?
- Check price presentation. Is a list price functioning mainly as a reference point?
- Check common knowledge. Has the anchor become a coordination point for everyone?
- Check regime change. Does the original reference still belong to the same world?
- Use blind review where useful. Hide prior scores or estimates until the fresh judgment is recorded.
- Document the update. Why did the new evidence move the estimate by this amount?
When the Anchoring Lens Fails
The lens fails when every use of a prior value is treated as bias.
Past data can be genuinely informative.
Bayesian updating should begin from prior information when the prior is relevant.
The problem is not starting somewhere.
Every estimate starts somewhere.
The problem is allowing a weak or irrelevant start to retain more influence than the evidence justifies.
Anchoring Is Not Bayesian Updating
A rational prior changes when evidence arrives according to how informative the evidence is.
Anchoring describes insufficient or distorted movement caused by the reference point itself.
A prior earns weight from evidence.
An arbitrary anchor has no such entitlement.
A Better Question Than “What Is a Reasonable Number?”
Ask:
If I had never seen the first number, what evidence would I use to construct this estimate from the ground up?
How Anchoring Connects to the Rest of the World
- Bargaining: first offers can influence perceived negotiation ranges.
- Price discrimination: reference prices change how differentiated prices feel.
- Winner’s curse: initial valuations can block the downward update that winning should trigger.
- Legibility: official scores and benchmarks become institutional anchors.
- Defaults: both use starting points but through different mechanisms.
- Path dependence: repeated adjustment from old values embeds history.
- Common knowledge: public anchors can become coordination points.
- Measurement: the first measured value can shape interpretation of later measurements.
- Information: anchor-consistent evidence can become more cognitively accessible.
- Time inconsistency: pre-committed estimates and limits can protect later judgment from competitive escalation.
Frequently Asked Questions
What is anchoring bias?
Anchoring bias is the tendency to give disproportionate weight to an initial number or reference point when making a later estimate or decision.
Can an irrelevant number cause anchoring?
Yes. Classic experiments found that arbitrary starting numbers can influence later numerical judgments even when participants know the numbers are unrelated to the question.
Does expertise eliminate anchoring?
No. Expertise can improve the evidence available for adjustment, but professional judgments can still be influenced by starting values. Independent assessment and structured review help.
How can I reduce anchoring?
Estimate independently before exposure, decompose the problem, use several reference points, compare with outside-view data, document the reason for updates and rebuild the model after major regime changes.
Research Basis and Further Reading
- American Psychological Association, Anchoring Bias, for the formal behavioural definition.
- Daniel Kahneman, Nobel lecture, on the heuristics-and-biases programme developed with Amos Tversky.
- Amos Tversky and Daniel Kahneman’s classic work on judgment under uncertainty established anchoring and adjustment as a foundational heuristic.
What to Read Next on eduKateSG
- How The World Works | Bargaining — why first offers influence perception but outside options determine structural power.
- How The World Works | Winner’s Curse — why winning should sometimes force a downward valuation update.
- How The World Works | Legibility — why official numbers become powerful reference points.
- How The World Works | Path Dependence — how yesterday’s starting point can remain inside today’s decision.
The Larger Idea
The first number is often not just a number.
It creates a centre.
Once there is a centre, every later estimate feels like movement away from something rather than construction from nothing.
That saves cognitive effort.
It also creates inertia.
The discipline is not to become anchorless.
Reasoning without any starting point is impossible.
The discipline is to ask whether the starting point earned its influence.
Anchoring is what happens when the first number quietly changes the problem from “What does the evidence say?” to “How far away from this number am I willing to move?”