Phenomena Related to Anchoring

Anchoring is a pervasive judgment phenomenon in which a salient but uninformative number biases subsequent numerical estimates. To clarify discussion, the term is often used in three distinct ways: the anchoring procedure (exposing a decision maker to an irrelevant number), the anchoring effect or result (assimilation of the estimate toward the anchor), and the underlying psychological process that produces the effect. In a standard paradigm, people first compare the target value to the anchor (e.g., “Is the percentage of African countries in the UN more or less than 10%?”) and then produce their own estimate. Anchors can be completely arbitrary—generated by a wheel of fortune or a phone number—and still exert a powerful pull, demonstrating that the effect is a genuine bias rather than a rational use of information or a demand characteristic. Even when we deliberately start with a known number and adjust in the direction we believe is correct, that adjustment is typically insufficient: a person from Chicago guessing the population of Milwaukee might anchor on Chicago’s three million and adjust down to one million, while a person from Green Bay might anchor on Green Bay’s one hundred thousand and adjust up to three hundred thousand, when the actual population is about 590,000.

Beyond numbers, the same cognitive tendency can appear as a broader fixation on a single piece of information—focusing so intently on one initial fact or idea that other relevant data and alternative possibilities are ignored. In problem-solving, for example, this narrow focus can cause people to overlook creative solutions simply because they are anchored to the first course of action that comes to mind.

Necessary Conditions for Anchoring

Not every uninformative number produces an anchoring effect. The bias depends on features of the anchor, the target, and the judgment task. For an anchor to influence judgment it must be attended to, it must be sufficiently salient, and the task must require a numerical response on a compatible scale. When the anchor is expressed in a different metric or domain than the target, compatibility effects severely reduce or eliminate anchoring (e.g., monetary anchors affect monetary judgments but not life‑expectancy estimates). Furthermore, the effect is most robust when the judgment task involves some uncertainty and the decision maker’s knowledge is limited or ambiguous.

Causes of Anchoring

Several psychological mechanisms contribute to anchoring, which can operate at different stages of the judgment process: retrieval of information from memory, integration of information into an overall evaluation, and expression of the evaluation on an external scale.

Retrieval Stage: Selective Accessibility and Priming

The most prominent account is the Selective Accessibility Model, which posits that when people compare the target to the anchor, they automatically search memory for information consistent with the hypothesis that the target’s value equals the anchor. This biased recruitment of anchor‑consistent knowledge increases the availability of reasons why the target resembles the anchor, while neglecting reasons why it differs. This process is a form of associative activation: the anchor primes similar features, making them more accessible to System 1 and thereby shaping the data on which deliberate System 2 judgments are based. Importantly, people are unaware of this biased retrieval and cannot control it, which is why absurd anchors still have an effect—such as the finding that people’s estimates of the year Attila the Hun sacked Europe are pulled toward the last three digits of their own phone number, even though the two numbers are plainly unrelated.

Integration and Expression Stages

Some approaches emphasize that the anchor may be given weight during information integration—for instance, as an initial estimate that is insufficiently adjusted away from (the classic “anchoring and adjustment” heuristic). Alternatively, the anchor may influence how an internal judgment is mapped onto a response scale, making certain ranges of the scale more accessible. These mechanisms are not mutually exclusive; anchoring likely results from a combination of processes, with the retrieval‑stage priming effect being the most empirically supported, although adjustment heuristics also play a role in many everyday judgments where people self‑generate an initial value and then correct.

Applications of Anchoring

Anchoring influences a vast array of real‑world judgments. In legal settings, mock jurors’ liability decisions and damage awards are swayed by the plaintiff’s requested compensation, even when the request is implausibly high or low. Real estate agents’ appraisals of properties are biased by listing prices, despite their expertise. Teachers’ grading is influenced by students’ prior performance records. In surveys, anchoring distorts contingent valuation estimates (e.g., willingness to pay to save wildlife) and even self‑reports of well‑known quantities like monthly spending. Negotiation outcomes strongly depend on the first offer, which serves as an anchor anchoring the negotiation range. A classic negotiation tactic exemplifies this: a seller starts with an extremely high price, shifting the buyer’s perception of value, then gradually lowers it so the final price appears more attractive. Group decisions, consumer evaluations of product bundles, and predictions of future performance all display anchoring effects.

In financial markets, investors often anchor on the price at which they bought a stock, judging subsequent decisions by that reference rather than the underlying business value. Similarly, a stock that has fallen from a previous high can seem like a bargain because of anchoring on its peak price, even when the company’s fundamentals have deteriorated.

Even the order of survey questions can anchor responses on unrelated topics. When college students are asked “How often are you dating?” before “How happy are you?”, the correlation between the two answers jumps from .11 to .62, because the dating question serves as an anchor, causing students to evaluate their overall happiness through the lens of their dating frequency. Anchors can also be deliberately built into choice environments as nudges. In taxicabs, payment screens that present preset tip options of 15%, 20%, and 25% raise average tips compared to no defaults, partly because people anchor on the suggested percentages and often select the middle option. Higher‑default screens (20%, 25%, 30%) further increase average tips, although they also slightly raise the number of customers who give nothing—a phenomenon known as reactance, where people rebel against what they perceive as an overly aggressive nudge.

The phenomenon is also a core theoretical mechanism for other judgment biases, including egocentric biases, overconfidence, hindsight bias, preference reversals, and framing effects. In each case, salient information draws disproportionate attention to compatible features, shifting subsequent evaluations.

Relation to Other Judgmental Phenomena

The anchoring process—biased recruitment of hypothesis‑consistent evidence—is analogous to confirmation bias. Just as people seek and interpret information to confirm their current hypothesis, anchoring leads decision makers to generate reasons why the target value is similar to the anchor while overlooking disconfirming evidence. The reduction of anchoring when people are explicitly instructed to consider opposite arguments mirrors debiasing techniques used to reduce overconfidence and hindsight bias.

Overconfidence itself is often attributed to a failure to consider why one’s chosen answer might be wrong; prompting people to list counter‑reasons reduces overconfidence. Similarly, self‑generated anchors produce less bias than experimenter‑provided anchors, presumably because generation involves considering alternative values. Hindsight bias—the tendency to see outcomes as more predictable after knowing them—can be understood as anchoring on the known outcome, which selectively retrieves outcome‑consistent facts. Having people explain how alternative outcomes could have occurred reduces the bias, just as considering counter‑anchor arguments reduces anchoring.

Shafir’s “accept/reject” framing effect shows that the decision goal itself can act as an anchor, shifting attention toward positive features when accepting and toward negative features when rejecting. Preference reversals, where the weight given to attributes changes with the response mode (e.g., pricing vs. rating), also share an underlying attentional compatibility mechanism with anchoring: the response scale, like an anchor, directs attention toward compatible features.

Anchoring is sometimes confused with reference point effects, but they are distinct. Reference points alter the perception of gains and losses, change risk preferences, and produce contrast effects (e.g., a lower reference point enhances perceived value). Anchoring, by contrast, is an associative bias that produces assimilation effects—judgments are pulled toward the anchor. The former is rooted in psychophysical processes, the latter in priming and selective accessibility.

Practical Implications and Debiasing

The automaticity of anchoring makes people highly suggestible. Anchors influence judgments even when they are obviously random, and people confidently deny any influence. Because System 2 operates on data retrieved by System 1, it cannot easily resist the biased sampling of anchor‑consistent memories. This insight has wide‑ranging implications: marketing tactics like arbitrary purchase limits (“Limit of 12 per person”) increase sales by anchoring the quantity; initial offers in negotiations heavily shape final agreements. Even well‑intentioned policies, such as capping damage awards, may inadvertently raise the median award because the cap serves as an anchor. However, reactance reveals a boundary condition: if an anchor—such as a high preset tip option—feels manipulative, some people will reject it entirely, so the anchor’s strength must be calibrated to avoid provoking a backlash.

Awareness of the bias is a necessary first step, but it is rarely sufficient on its own. Defenses against anchoring require deliberate activation of System 2. In negotiations, focusing attention on the opponent’s minimum acceptable offer or the costs of impasse reduces the anchor’s pull. A general strategy of “thinking the opposite”—systematically generating reasons that contradict the anchor—can mitigate the biased recruitment of thoughts. Another powerful approach is to evaluate decisions from a zero base: reassess the value or situation as if the anchor did not exist, focusing on fundamental facts and what you ultimately want to achieve. For instance, an investor should judge a stock on its current business fundamentals rather than on the price originally paid or a past peak. More broadly, deliberately expanding the pool of information considered disrupts the narrow fixation that anchors produce.

Consider an everyday problem: a driver who gets a flat tire often anchors on the demanding task of changing the tire alone on the roadside. That initial fixation blocks out other possibilities—noticing that the tire has only a slow leak and could be inflated at a nearby station to reach a repair shop, calling a friend for help, or using a roadside assistance service. By consciously stepping back and allowing more information into the decision, a much wider set of solutions becomes visible, and the anchor loses its grip. In all domains, when stakes are high, recognizing that any salient number or idea may be anchoring your thinking—and then deliberately seeking out disconfirming evidence, alternative perspectives, and baseline values—is essential for making sound judgments.