How The Anchoring Effect Changes Decisions

Ask two groups to estimate the percentage of African countries in the United Nations, but first have each group spin a wheel rigged to land on either 10 or 65, and their final estimates will differ by roughly 20 percentage points, clustering higher for whichever group saw the larger number. The wheel's result has nothing to do with the actual answer, and participants knew the wheel was random, yet it still pulled their guesses toward it. This is the anchoring effect: an initial number, even an explicitly meaningless one, measurably distorts a subsequent judgment. It shows up in salary negotiations, home appraisals, courtroom verdicts, and the price tag at a department store, and understanding the mechanism behind it is one of the most practically useful things a person can take from behavioral psychology.

The original experiment that identified the effect

Psychologists Amos Tversky and Daniel Kahneman first formally documented anchoring in their 1974 paper "Judgment Under Uncertainty: Heuristics and Biases," published in the journal Science. The paper cataloged several mental shortcuts, or heuristics, the mind uses to make judgments under uncertainty, and anchoring was one of three core biases they described alongside representativeness and availability. This body of work later contributed to Kahneman winning the Nobel Memorial Prize in Economic Sciences in 2002; Tversky had died in 1996 and was therefore ineligible, as the Nobel is not awarded posthumously.

The wheel-of-fortune experiment worked like this. Participants watched a wheel marked zero to 100 that was secretly rigged to stop at only one of two values, 10 or 65. After the spin, participants were asked two questions in sequence: first, whether the percentage of African nations belonging to the United Nations was higher or lower than the number the wheel landed on, and second, what they thought the actual percentage was. This two-step format, a comparative judgment followed by an absolute estimate, became the template most anchoring studies still use today.

The results were striking given how obviously irrelevant the wheel's number was to a factual question about UN membership. The group that saw the wheel land on 65 gave a median estimate of 45 percent. The group that saw it land on 10 gave a median estimate of 25 percent. That 20-point gap was produced entirely by a random number generator, and participants had been told, and could plainly see, that the wheel's outcome was rigged and arbitrary. Knowing the anchor was meaningless did not stop it from working.

Tversky and Kahneman ran variations to test whether the effect held under different conditions, including versions where participants generated their own anchor by doing a quick multiplication problem under time pressure rather than seeing a number chosen for them. In one such variant, participants asked to rapidly estimate the product of 8 x 7 x 6 x 5 x 4 x 3 x 2 x 1 gave a substantially higher median estimate than participants asked to estimate 1 x 2 x 3 x 4 x 5 x 6 x 7 x 8, the same calculation presented in reverse order, because the first few numbers multiplied in each sequence set a different initial anchor even though the correct final answer, 40,320, was identical either way.

Why an irrelevant number can still shift a judgment

The dominant explanation for decades was what researchers call insufficient adjustment. The idea is straightforward: when a person is handed a starting reference point, or generates one themselves, they don't ignore it and build a fresh estimate from scratch. Instead, they adjust away from that starting point until the answer feels plausible, and stop adjusting too soon. The final number ends up closer to the anchor than an unbiased estimate would be, simply because the mental effort of adjusting further outweighs the felt need to keep correcting.

A second, more recent explanation, developed by Thomas Mussweiler and Fritz Strack in the late 1990s, is called selective accessibility. Under this account, when someone is exposed to an anchor, their mind briefly and automatically tests the hypothesis that the anchor is correct. That test activates memories, associations, and pieces of knowledge that are consistent with the anchor being true. Once the person moves on to forming their actual judgment, that temporarily activated, anchor-consistent information is what's most readily available to them, which biases the final answer toward the anchor even though the person is no longer consciously thinking about the number itself.

The two explanations aren't mutually exclusive, and current research suggests both mechanisms probably operate depending on the type of anchor involved. Self-generated anchors, like the multiplication problem above, appear to rely more on insufficient adjustment, since the person is doing active mental math and stopping early. Externally provided anchors, like a listing price or an opposing counsel's number, appear to rely more on selective accessibility, since there's no explicit adjustment process to interrupt.

Anchoring effects have been documented even when the anchor is explicitly and obviously irrelevant, as in the spinning wheel experiment, and even among trained professionals working within their area of expertise. A widely cited study by Gregory Northcraft and Margaret Neale, published in Organizational Behavior and Human Decision Processes in 1987, found that experienced real estate agents' property valuations were measurably shifted by an arbitrary listing price anchor, despite the agents insisting the listing price played no role in their independent professional judgment. When later asked which factors had influenced their appraisal, only about 19 percent of the agents mentioned the listing price at all, compared with 37 percent of the amateur participants in the same study, even though both groups were equally anchored by the number.

Does anchoring still work if you tell people to ignore the anchor?

Only partially. Studies that explicitly warn participants about anchoring bias and instruct them to disregard the initial number find the effect shrinks but rarely disappears. This pattern is one reason researchers believe anchoring operates at least partly outside full conscious control, rather than being a simple failure of willpower that harder effort could fix.

Where anchoring shows up in negotiation and pricing

Negotiation research has consistently found that the first number named in a negotiation, whether it's a salary figure, a home price, or a settlement amount, pulls the final agreed-upon outcome toward it more strongly than either party's independently formed sense of fair value. This is part of why negotiation training frequently coaches people to make the first offer, when circumstances allow, rather than waiting for the other side to name a number first: making the first offer lets you set the anchor on favorable terms rather than reacting to someone else's.

Retail and e-commerce pricing exploits the identical mechanism. A price tag showing a high "original" price crossed out beside a lower "sale" price uses the crossed-out figure as an anchor, which makes the sale price feel like a bigger discount than the same sale price shown on its own, without a reference point. This pattern is well documented in consumer psychology research on reference-price framing, and it's one reason retailers so rarely list a sale price without also showing what it's being discounted from.

Salary negotiation follows the same logic in a way that has real financial consequences over a career. A candidate who names a number first, grounded in solid market research, sets the range the rest of the conversation gets adjusted around. A candidate who waits to be given a number first is effectively letting the employer's opening figure become the anchor their own counteroffer gets pulled toward, even if that opening figure was set strategically low.

Where anchoring shows up in courtrooms and legal judgments

Anchoring's reach extends into contexts where the stakes are considerably higher than a discount on a jacket. Research on jury damage awards has repeatedly found that the dollar amount a plaintiff's attorney requests strongly predicts the amount jurors ultimately award, largely independent of the strength of the underlying evidence. A plaintiff's attorney who asks for a higher number tends to walk away with a higher award than one who asks for less, even when the two cases are comparable on the facts.

Judicial sentencing shows the same pattern with professionals whose entire job is supposed to insulate them from exactly this kind of bias. In a study by Birte Englich, Thomas Mussweiler, and Fritz Strack published in Personality and Social Psychology Bulletin in 2006, experienced criminal court judges were given a fictional case and then exposed to a sentencing recommendation that both the researchers and the judges themselves knew was generated randomly, in one version literally by rolling dice. Judges who were exposed to a higher random number recommended meaningfully longer sentences than judges exposed to a lower one, despite reviewing identical case files and despite knowing the number's origin was arbitrary.

Separately, prosecutorial sentencing demands function as a related real-world anchor: Englich and Mussweiler's earlier 2001 research found that the specific sentence length a prosecutor requests tends to anchor the sentence a judge ultimately hands down, even though the prosecutor's request is, legally speaking, only a recommendation rather than a binding figure. Defense attorneys have increasingly built trial strategy around this finding, since simply arguing that a plaintiff's number is "too high" without offering a competing anchor of their own tends to be a weak countermeasure, according to jury simulation research published in the Iowa Law Review.

Comparing anchoring's documented strength across contexts

Anchoring isn't a fixed-size effect; how strongly it bends a judgment depends heavily on the domain, how much relevant expertise the judge has, and how ambiguous the underlying question is. The table below summarizes documented anchoring effects across three well-studied real-world contexts.

Context	Who is affected	Documented anchor source	Approximate effect
Numeric estimation (lab)	General public	Random wheel spin (Tversky & Kahneman, 1974)	~20-point shift in median estimate between low- and high-anchor groups
Real estate appraisal	Both amateurs and professional agents	Property listing price (Northcraft & Neale, 1987)	Professionals and amateurs equally anchored, despite agents denying influence
Legal sentencing	Experienced criminal court judges	Randomly generated number, e.g. dice roll (Englich, Mussweiler & Strack, 2006)	Higher random anchor produced meaningfully longer recommended sentences

Anchoring measurably shifted judgments in all three domains, and the effect size did not shrink meaningfully when the people being tested were domain experts rather than novices.

Why professionals aren't immune, and why that matters

The consistent finding across the real estate and judicial research is that expertise does not reliably protect against anchoring, and it may not protect against it at all. This runs counter to the intuitive assumption that training and experience should let professionals recognize and discount an irrelevant number. In the Northcraft and Neale study, agents were, if anything, more confident that they hadn't been influenced than the amateurs were, even though both groups showed statistically indistinguishable anchoring effects on their final appraisals.

One plausible reason is that anchoring operates through automatic, largely unconscious cognitive processes, the selective accessibility mechanism described above, rather than through a deliberate weighing of the anchor as one input among many. Because the bias doesn't feel like being persuaded, from the inside, it's much harder to catch yourself doing it than it would be to catch yourself being talked into something through an argument you consciously evaluated and accepted.

This has a practical implication for anyone making a numeric judgment that matters: appraisers, underwriters, hiring managers setting salary bands, and analysts forecasting a budget line item all operate in professional roles where an anchoring number, a prior year's figure, a competitor's estimate, a client's stated budget, is often present in the room before an independent judgment gets formed. The research suggests that simply being an expert in the domain doesn't neutralize the anchor's pull.

What actually reduces anchoring's influence

Given that awareness alone only partially works, researchers have looked for interventions that do better. The most consistently effective strategy documented in the literature is generating an independent estimate from a different reference point, or a different starting calculation entirely, before being exposed to any anchor at all. This works because it denies the selective-accessibility process a foothold: there's no anchor yet for the mind to search for supporting evidence around.

A related technique, sometimes called "consider the opposite," asks a person to deliberately generate reasons the anchor might be wrong, or reasons the true answer could be considerably different from the anchor, before finalizing a judgment. Because this actively counters the automatic search for anchor-consistent information, some studies have found it reduces the anchoring effect more than a simple instruction to "try to ignore the number."

Neither strategy eliminates anchoring completely in the research literature, which is itself a useful data point. The honest limitation here is that no fully reliable debiasing technique has been documented; even trained researchers who study anchoring for a living remain susceptible to it in contexts outside their specific expertise. The practical takeaway from the research isn't that anchoring can be switched off, but that its size can be reduced by structuring a decision process, deliberately forming an independent estimate first, before an anchor enters the picture, a habit connected to the broader challenge of noticing one's own overconfidence explored in why confirmation bias is hard to notice.

Common misconceptions about anchoring

A frequent misunderstanding is that anchoring only works on people who aren't paying attention, or who don't know the trick exists. The research directly contradicts this: Tversky and Kahneman's original participants knew the wheel was rigged, and Englich, Mussweiler, and Strack's judges knew the sentencing number came from a dice roll. Awareness of the mechanism reduces its strength somewhat but does not require ignorance to function in the first place.

Another common misconception is that anchoring is the same thing as simple persuasion, where someone hears an argument, weighs it, and consciously decides to be convinced. Anchoring instead operates largely below the level of deliberate reasoning, more like a perceptual illusion than a rhetorical argument, which is part of why telling someone "don't let that number influence you" is a much weaker countermeasure than it would be against an actual persuasive argument.

A third misconception is that anchoring is primarily about numbers in a narrow, mathematical sense, relevant only to price tags and estimates. In practice, anchoring effects have been documented for judgments as varied as probability estimates, moral evaluations, and first impressions of a person's competence, anywhere a person is asked to place something on a scale after being exposed to an initial reference point, a pattern connected to how readily the brain latches onto structure in why the brain notices patterns everywhere.

The takeaway

The anchoring effect shows that an arbitrary starting number, even one a person knows to be random, irrelevant, or explicitly meaningless, can measurably pull a subsequent judgment toward it, largely because people adjust away from an initial reference point too conservatively rather than discarding it outright, and because exposure to the anchor briefly biases what information comes to mind. This isn't a narrow lab curiosity: it shows up in salary negotiations, real estate appraisals, retail discount pricing, and jury damage awards, and it measurably affects trained judges handing down criminal sentences just as much as it affects untrained members of the public. Even explicit warnings to ignore the anchor only partially reduce its pull, which is why the most reliable documented countermeasure is forming an independent judgment before ever being exposed to the number in the first place, a discipline worth building into any decision where the stakes of being subtly nudged, rather than persuaded, are high, and one closely related to the broader difficulty of noticing what the mind filters out entirely, discussed in how attention blindness hides obvious things.

---

Read next:

• Why confirmation bias is hard to notice
• How attention blindness hides obvious things
• Why the brain notices patterns everywhere

FAQ

Q: What is the anchoring effect in psychology?

It's a cognitive bias where an initial number or reference point, even an arbitrary or explicitly irrelevant one, measurably influences a person's subsequent numerical estimate or judgment, pulling the final answer toward the anchor. It was first formally documented by Amos Tversky and Daniel Kahneman in 1974.

Q: Who first documented the anchoring effect?

Amos Tversky and Daniel Kahneman documented it in their influential 1974 paper "Judgment Under Uncertainty: Heuristics and Biases," published in Science, using an experiment in which a rigged spinning wheel's random number measurably shifted participants' estimates on an unrelated factual question about UN membership.

Q: Does anchoring still work even when people know the number is random?

Yes. In Tversky and Kahneman's original experiment, participants saw a wheel they knew was rigged to produce a random number, yet their estimates were still pulled significantly toward that number. Judges in later research showed the same pattern when told a sentencing anchor came from a dice roll.

Q: Can experts and professionals be affected by anchoring too?

Yes. A 1987 study by Gregory Northcraft and Margaret Neale found experienced real estate agents' property valuations were measurably influenced by an arbitrary listing price anchor, even though the agents believed their professional judgment was unaffected by it. A 2006 study found experienced criminal court judges' sentencing recommendations shifted based on a randomly generated anchor number as well.

Q: How does anchoring affect legal cases and jury verdicts?

Research on jury damage awards has found that the dollar amount a plaintiff's attorney requests strongly predicts the final award amount, largely independent of the underlying evidence's strength. Separate research found prosecutorial sentencing demands similarly anchor the sentences judges hand down.

Q: How can someone reduce the influence of anchoring on their decisions?

Research suggests that simply being warned about anchoring only partially reduces its effect. The most effective documented strategy is generating an independent estimate from a different reference point before being exposed to any anchor at all, sometimes combined with deliberately considering reasons the anchor could be wrong.

Q: Why does an anchor influence a judgment even when it's known to be irrelevant?

Two mechanisms are thought to be involved: insufficient adjustment, where people move away from a starting reference point but stop adjusting too soon, and selective accessibility, where exposure to the anchor briefly activates anchor-consistent information in memory that then biases the final judgment.