Explain how an initial value, even an irrelevant one, pulls later judgements towards it, compare the two explanations of why, and evaluate anchoring research in legal, economic, health and everyday decisions.
Each is in the Psychology glossary, with every other term in the subject.
The tendency for a numerical estimate or decision to be drawn towards an initial value that a person has considered, even an irrelevant one, so the final judgement stays too close to it.
A systematic deviation from rationality or from an accepted norm in judgement, so that errors in thinking occur in a predictable direction rather than at random.
An explanation of anchoring in which people start from the anchor and adjust towards a plausible value but stop too early, at the edge of the plausible range nearest the anchor.
An explanation of anchoring in which considering an anchor activates knowledge consistent with it, making that knowledge more accessible in memory so that it dominates the final estimate.
Describe anchoring bias and explain why it counts as a cognitive bias rather than a sensible use of information.
Describe the standard anchoring paradigm and how the anchoring effect is measured.
Explain the insufficient adjustment and selective accessibility explanations of anchoring, and compare and contrast them.
Explain how uncertainty, expertise, incentives and the type of anchor affect the size of the effect.
Describe the aim, method, results and conclusions of Englich, Mussweiler and Strack (2006) on judicial sentencing.
Apply anchoring bias to decision-making in health, relationships, learning and economic settings.
Evaluate research on anchoring with reference to causality, measurement, bias, perspective and responsibility.
Anchoring bias is the tendency for a number that a person has just considered to pull a later estimate or decision towards it. It affects the price a buyer offers for a house, the salary a job applicant accepts, the length of a prison sentence and the dose a doctor judges to be safe, so it is one of the most studied examples of how decision-making departs from what a fully rational decision-maker would do.
Anchoring belongs to the cognitive approach, which explains behaviour through mental processes such as attention, memory and reasoning. It is one of the clearest examples of a cognitive bias, and research on it raises questions of causality (what exactly produces the pull of the anchor), measurement (how a mental process that people cannot report is detected) and bias in the research itself.
With a free account, Augment remembers where you stopped and what you have practiced.
Scope: An approach term in the cognitive approach. Questions can ask about anchoring bias with any example of behaviour, and it is applied in the learning and cognition context, where cognitive biases in decision-making are an area of study, and in scenarios from the other three contexts.
Many judgements in everyday life are estimates of uncertain quantities: how long a journey will take, what a second-hand car is worth, how many people will attend an event. People rarely calculate such values from first principles. Instead they start from a value that is already in mind and work from there. That starting value is the anchor.
The word “systematic” is what separates a bias from ordinary error. If a person’s estimates were simply inaccurate, they would be too high on some occasions and too low on others, and the errors would cancel out across many people. A bias pushes estimates the same way for most people under the same conditions, so it can be predicted and tested.
Three features show that anchoring is a bias and not a sensible use of information:
It is systematic. High anchors push estimates up and low anchors push them down, in a predictable direction.
It works with anchors that are obviously irrelevant. Numbers produced by a spinning wheel, a dice roll or the last digits of an identity number still shift judgements, so people cannot be treating the anchor as a useful clue.
It resists awareness. Warnings, rewards for accuracy and professional expertise reduce the effect only a little, and people usually deny that the anchor influenced them.
Anchoring is one of three judgement heuristics that Amos Tversky and Daniel Kahneman described in 1974, alongside availability and representativeness. A heuristic is a mental shortcut that gives a quick, usually adequate answer. Starting from a known value and adjusting is often efficient: a person estimating the boiling point of water at the top of a mountain does well to start at 100 °C. The bias is the systematic error that remains when the adjustment is not large enough.
Anchors differ in where they come from, and the difference matters because the two explanations of anchoring fit different kinds of anchor.
Type of anchor | What it is | Example |
|---|---|---|
Externally provided and informative | A value given by someone else that may carry real information | The asking price of a flat; the first offer in a salary negotiation |
Externally provided and irrelevant | A value that obviously carries no information about the answer | A number from a wheel of fortune, a dice roll, the last two digits of a phone number |
Self-generated | A value the person produces from their own knowledge, then adjusts from | Estimating when a historical event happened by starting from a nearby date one knows |
Incidental | A number that is merely present in the surroundings and is never compared with the answer | The number on a sports shirt or in the name of a restaurant |
The first two types are the ones studied most often, and they produce large and reliable effects. The evidence for incidental anchors is much weaker.
Almost all anchoring research uses the same two-step procedure, which Tversky and Kahneman introduced. It is simple, quick to run and produces a large effect.
Comparative question. Participants judge whether a quantity is higher or lower than a given value, for example “Is the Mississippi River longer or shorter than 1,000 km?” (a low anchor) or “…than 6,000 km?” (a high anchor).
Absolute question. All participants then give their own estimate: “How long is the Mississippi River?”
Comparison of groups. The anchoring effect is the difference between the mean (or median) estimate of the high-anchor group and that of the low-anchor group.
Because the anchor is manipulated as the independent variable, the estimate is measured as the dependent variable and participants are randomly allocated to the low or high condition, the standard paradigm is a true experiment. The comparative question matters: it makes participants actively consider the anchor as a possible answer, which is the condition under which anchoring is strongest.
In the best-known demonstration, Tversky and Kahneman (1974) spun a wheel of fortune numbered 0–100 in front of participants. The wheel was rigged to stop at either 10 or 65. Participants first judged whether the percentage of African countries in the United Nations was higher or lower than the number, then estimated the percentage. The median estimate was 25% in the group whose wheel stopped at 10 and 45% in the group whose wheel stopped at 65, and paying participants for accuracy did not reduce the effect.
A second demonstration in the same paper did not use an anchor from outside. High-school students had five seconds to estimate the product 8 × 7 × 6 × 5 × 4 × 3 × 2 × 1 or the same numbers in ascending order. The median estimate was 2,250 for the descending sequence and 512 for the ascending one, although the correct answer is 40,320. Participants multiplied the first few numbers and extrapolated, so the first steps of the calculation acted as a self-generated anchor.
Strack and Mussweiler (1997) showed that anchors work even when they are absurd. Students judged whether Mahatma Gandhi died before or after the age of 9, or before or after the age of 140. No one believes either value, yet the group given 140 then estimated his age at death at 67 on average, compared with 50 for the group given 9.
YouTubeTwo explanations of why anchors shift estimates dominate the research. Each fits some kinds of anchor better than others, so the type of anchor largely decides which one applies.
Tversky and Kahneman (1974) proposed that people treat the anchor as a starting point and move away from it step by step until they reach a value that seems plausible. Adjusting takes effort, and people stop as soon as the answer feels acceptable. They therefore end at the edge of the range of plausible values that lies nearest the anchor, not in the middle of that range.
The figure shows why two groups who know the same facts end up with different estimates. Both groups accept roughly the same range of plausible answers, but because each one approaches that range from a different side, each stops at a different boundary. The gap between the groups is the anchoring effect.
Epley and Gilovich (2001, 2006) found the clearest support for this account with self-generated anchors. A person asked when George Washington was elected president of the United States often starts from 1776, the year of independence, and adjusts upwards, knowing the election came later. Estimates produced in this way stay closer to 1776 than the true answer of 1788. Adjustment shrank further when participants were distracted or were nodding their heads (an action associated with accepting a value) and grew when they were paid for accuracy, which fits the idea that adjustment is a deliberate, effortful process that stops once a plausible value is reached.
In terms of dual-processing theory, adjustment is a System 2 activity: slow, effortful and dependent on attention. When people are tired, rushed or distracted, System 2 does less work, adjustment is cut short and estimates stay closer to the anchor.
Insufficient adjustment has a clear weakness. It cannot explain why anchors that no one would ever use as a starting point still work. Nobody starts at 140 and adjusts downwards to estimate Gandhi’s age at death, yet the 140 anchor raised estimates by 17 years on average.
Strack and Mussweiler (1997) explained anchoring through memory. When people answer the comparative question, they implicitly test the hypothesis that the true value is close to the anchor. Like most hypothesis testing, this search is confirmatory: people bring to mind knowledge that fits the anchor. Asked whether Gandhi died before or after 140, a person thinks of evidence that he was old when he died, such as pictures of him as a thin, white-haired man. That knowledge is now more accessible in memory, so it dominates the later absolute estimate.
Selective accessibility makes predictions that insufficient adjustment does not, and these predictions have been tested.
Implausible anchors should still work, because considering them still activates relevant knowledge. The Gandhi study confirmed this.
The effect should be stronger when the comparative and absolute questions concern the same dimension. Strack and Mussweiler asked participants to compare the height of the Brandenburg Gate with an anchor. Later estimates of its height were anchored more strongly than estimates of its width, because the activated knowledge was about height.
Thinking of reasons against the anchor should weaken the effect, because it makes anchor-inconsistent knowledge accessible as well. Mussweiler, Strack and Pfeiffer (2000) asked German car experts to value a used car after a seller suggested a price. Experts who first listed arguments against the suggested price were anchored less than experts who did not.
Answering the absolute question should be faster after a plausible anchor, because the relevant knowledge has already been retrieved. Strack and Mussweiler (1997) found that absolute judgements took longer after implausible anchors, which can be rejected without retrieving much knowledge.
The model links anchoring to two other parts of the cognitive approach. The confirmatory search is a form of confirmation bias, and the knowledge that becomes accessible is organised in schemas: an anchor of 140 activates a schema for a very old man, which then guides retrieval.
YouTubeInsufficient adjustment | Selective accessibility | |
|---|---|---|
Core mechanism | Start at the anchor, adjust, stop too soon | Testing the anchor makes consistent knowledge more accessible |
Kind of processing | Effortful adjustment (System 2) that is cut short | Largely automatic activation of memory (System 1) |
Explains best | Self-generated anchors | Externally provided anchors, including absurd ones |
Key evidence | Epley and Gilovich: less adjustment under distraction, more with incentives | Strack and Mussweiler (1997); Mussweiler, Strack and Pfeiffer (2000) |
Main weakness | Cannot explain why absurd anchors work | Accessibility is inferred, not observed; some key tests have failed to replicate |
Both explanations share an assumption: people do not move far enough from the anchor, and they are unaware of how much it shaped their answer. They differ in where the bias enters. For insufficient adjustment it enters at the end of the judgement, when effortful correction stops; for selective accessibility it enters at the start, when the anchor shapes which evidence is retrieved. Many researchers now accept that both processes occur and that the type of anchor and the situation decide which one dominates.
Anchoring appears in almost every setting that has been tested, but its size varies. The factors that change it also help to show which mechanism is at work.
Anchoring is strongest when people know little about the quantity. Knowledge narrows the range of plausible answers, which leaves less room for the anchor to move the estimate. Wilson, Houston, Etling and Brekke (1996) found that participants who were more knowledgeable about a topic were less affected by an anchor. The same work showed that an anchor can bias judgement even without a comparative question, provided participants pay enough attention to the number, and that forewarning participants about the effect did not remove it.
Expertise protects much less than might be expected. Northcraft and Neale (1987) gave estate agents in Arizona a tour of a real house and a full information pack, which differed only in the listing price. Agents’ valuations rose with the listing price, much as students’ valuations did, yet few agents mentioned the listing price as something they had used. Experts in fields such as property, law and medicine still work with uncertain values, so they still rely on starting points.
Paying participants for accuracy did not reduce anchoring in Tversky and Kahneman’s wheel study, and forewarning does little. Incentives help mainly with self-generated anchors, where adjustment is effortful and more effort produces more adjustment. With externally provided anchors, the bias seems to arise before any effortful correction begins, which is what selective accessibility predicts.
Precise anchors tend to pull harder than round ones. Janiszewski and Uy (2008) found across five studies that participants adjusted less from a precise asking price for a beach house ($799,800) than from a round one ($800,000). They explained this through the scale people adjust along: a precise anchor is represented on a finer scale, so each mental step away from it is smaller. Anchors also work more strongly when they come from a source that seems relevant, although obviously random anchors still have an effect.
Some studies have claimed that numbers merely present in the environment, which participants never compare with the answer, can bias judgement. Critcher and Gilovich (2008) reported, for example, that people expected to pay more at a restaurant called “Studio 97” than at one called “Studio 17”. Shanks, Barbieri-Hermitte and Vadillo (2020) repeated this method in three studies and found no statistically significant incidental anchoring, while the standard comparative-question paradigm produced robust effects in the same studies. Anchoring therefore seems to require that people actually think about the anchor in relation to the judgement.
Anchoring matters because so many real decisions start from a number that someone else supplies. The same mechanism appears across very different settings, which is why it is a useful example of a cognitive process that explains behaviour in several contexts.
In a negotiation, the first offer acts as an anchor for everything that follows. Galinsky and Mussweiler (2001) found across three experiments that whichever side made the first offer, buyer or seller, obtained a better outcome, and that first offers strongly predicted final settlement prices. The advantage disappeared when the other negotiator focused on information inconsistent with the first offer, such as the opponent’s alternatives or their own target, which is the selective accessibility account applied to bargaining.
Ariely, Loewenstein and Prelec (2003) asked students to write down the last two digits of their social security number, consider whether they would pay that many dollars for items such as wine and chocolates, and then bid in a real auction. Students with high digits bid considerably more than students with low digits, yet their relative valuations were sensible: everyone valued the better wine more highly than the ordinary one. The authors called this pattern “coherent arbitrariness”: absolute values are anchored by an arbitrary number while comparisons between items stay consistent. Retail “was £80, now £40” labels and the high-priced item placed first on a menu use the same principle.
In medicine, anchoring describes a clinician fixing on an early piece of information, such as the first diagnosis suggested at triage, and failing to adjust enough when later symptoms do not fit. Graber, Franklin and Gordon (2005) analysed 100 cases of diagnostic error in internal medicine and found that cognitive factors contributed to 74% of them. The most common was premature closure, settling on an early diagnosis without considering alternatives, which is closely related to anchoring. Anchoring also affects patients: a person who is told that a treatment has “up to” a 90% success rate may judge their own chances against that figure rather than against the average.
In education, a teacher who sees a student’s predicted grade before marking may anchor on it, so that marks cluster around the prediction. Students also anchor when they plan: an estimate of how long revision will take is often anchored on the time a friend mentioned or on the length of a previous task.
In relationships, first impressions can act as anchors for later judgements of a person, and opening positions anchor disputes over shared money or chores. In group decisions, the first figure suggested in a meeting, such as a proposed budget, tends to shape the range that the group discusses, so the person who speaks first has disproportionate influence.
Context | Example of anchoring | Consequence |
|---|---|---|
Health and well-being | A doctor fixes on the first diagnosis suggested and under-weights later symptoms | Delayed or missed diagnosis |
Human development | Parents judge a child’s progress against the first assessment the child received | Expectations that lag behind real change |
Human relationships | The first offer in a negotiation over a shared purchase sets the range for the final agreement | The person who speaks first gains an advantage |
Learning and cognition | A teacher marks an essay after seeing the student’s predicted grade | Marks pulled towards the prediction |
Law and economics | A prosecutor’s sentencing demand or a listing price anchors a judge or a buyer | Unequal sentences; higher prices |
Anchoring is a cognitive explanation because it locates the cause of a behaviour (a biased estimate) in mental processes: hypothesis testing, retrieval from memory and effortful adjustment. It shows several assumptions of the cognitive approach at work. Mental processes can be studied scientifically through their effects on measurable behaviour; people are active processors of information who use shortcuts when time, information or motivation is limited; and the shortcuts produce errors that are systematic enough to predict.
Anchoring also fits the idea of the cognitive miser: a decision-maker who saves effort by using a heuristic and corrects only when there is a reason to do so. In dual-processing terms, the anchor is taken up quickly and automatically, and correcting for it requires effortful System 2 processing, which is often incomplete.
A purely cognitive account leaves out factors from the other approaches. Biological factors such as fatigue and alcohol, which reduce effortful processing, make adjustment shorter. Sociocultural factors decide which anchors people meet and how they treat them: bargaining norms in markets, sentencing guidelines in a legal system and the conventions of a shop’s price labels all shape the numbers that become anchors. Whether culture changes the strength of anchoring has been tested directly, by running the same anchoring tasks in samples from many countries.
Most early anchoring studies used trivia questions and student participants, which left open whether the effect matters when experts make decisions with serious consequences. Birte Englich, Thomas Mussweiler and Fritz Strack studied one of the most consequential numerical decisions in society: the length of a prison sentence.
The researchers aimed to find out whether experienced legal professionals’ sentencing decisions are anchored by a sentencing demand even when the demand comes from an irrelevant source or is openly determined at random, and to test whether the effect works through selective accessibility.
The paper reports four experiments with German legal professionals, mainly judges and prosecutors with courtroom experience. In each, participants read a realistic case file about a woman who had been caught shoplifting again, after earlier convictions, and gave the sentence they would impose. The source of the anchor varied across studies:
Study 1: the sentencing demand came from a journalist’s question in a telephone call (an irrelevant source).
Study 2: participants were told explicitly that the prosecutor’s demand had been determined at random.
Study 3: participants threw a pair of dice themselves and were told that the total was the prosecutor’s demand in months. The dice were loaded so that half the participants rolled a low total (3) and half a high total (9).
Study 4: after a high or low demand, participants completed a task that measured how accessible arguments for a harsher sentence (incriminating arguments) and for a milder one were in their minds.
The design in each study was an independent samples design with the anchor (low or high) as the independent variable. The dependent variable was the sentence in months, a precise numerical measure. Participants also reported their professional experience, which allowed the researchers to test whether experience reduced the effect.
Sentences were pulled towards the anchor in every study. In Study 3, participants who had rolled a 9 gave longer sentences, about 8 months on average, than those who had rolled a 3, about 5 months. Experience and expertise did not reduce the effect: experienced judges were anchored as much as less experienced ones. In Study 4, incriminating arguments were more accessible after a high anchor than after a low one.
Study | Source of the anchor | Main finding |
|---|---|---|
1 | A journalist’s question | Sentences assimilated towards the irrelevant demand |
2 | A demand described as random | Sentences still assimilated towards the demand |
3 | Loaded dice thrown by the participant (3 or 9) | About 5 months after a 3; about 8 months after a 9 |
4 | High or low demand, then an accessibility task | Incriminating arguments more accessible after a high anchor |
Sources: Englich, B., Mussweiler, T. and Strack, F. (2006). Playing dice with criminal sentences: the influence of irrelevant anchors on experts’ judicial decision making. _Personality and Social Psychology Bulletin, 32(2), 188–200. Study 3 means as reported in Kahneman (2011), Thinking, Fast and Slow._
Judicial decisions, which should depend only on the facts of a case, are biased by numbers that the decision-maker knows to be irrelevant. Expertise does not protect against the bias. Study 4 supports the selective accessibility model: a high demand made reasons for a harsh sentence come to mind more easily, and those reasons then shaped the sentence.
Each study manipulated the anchor and allocated participants to conditions at random, so the difference in sentences can be attributed to the anchor. Study 3 is especially strong: participants rolled the dice themselves, so they knew the demand was random, yet it still affected them. Study 4 goes further than most anchoring studies by testing a mechanism as well as an effect, although accessibility was measured indirectly and the study shows that accessibility and sentencing change together rather than proving that one causes the other.
Using practising judges and prosecutors gives the findings much more relevance to real courts than trivia questions with students.
However, participants read a short case file and decided alone, without hearing witnesses or the defence, with no time pressure from a real court list and no consequences for anyone. Real sentencing involves more information and sometimes several judges, which could either dilute the anchor or strengthen it through fatigue. The case was a minor offence, so the size of the effect for serious crimes is unknown, and the German legal system, in which the prosecutor proposes a sentence, differs from adversarial systems such as those in England or the United States.
The studies used deception: loaded dice and fictitious case files. The deception was mild, participants were professionals who were not harmed, and the cost of deceiving them is small compared with the benefit of showing a source of unfairness in the justice system. The findings also raise a question of responsibility for how they are applied. If a prosecutor’s demand anchors a judge, then a prosecutor who asks for an inflated sentence can lengthen it, which supports reforms such as sentencing guidelines and training judges to consider arguments against a demanded sentence.
Anchoring is one of the most robust findings in cognitive psychology. Its mechanism and its size in everyday decisions are much less certain.
Almost all anchoring research uses true experiments with random allocation, so the claim that anchors cause shifts in estimates is very well supported.
Claims about why are weaker. Selective accessibility and insufficient adjustment are inferred from patterns of results, such as stronger effects on the same dimension or longer response times after implausible anchors, and the patterns are sometimes consistent with both. Harris and colleagues (2019) ran four experiments using a test that had been proposed as a signature of selective accessibility and found no evidence for it, and a fifth experiment using a different method also failed. They concluded that anchoring itself is remarkably robust but that the theoretical basis of some earlier tests of selective accessibility is shaky.
People generally deny that an anchor affected them, as Northcraft and Neale’s estate agents did, so self-report is useless for measuring anchoring. Researchers instead compare the estimates of groups given different anchors. This is a clean measure of the effect at group level, but it says nothing about any one person: it cannot show whether a particular judge was anchored, only that judges as a group were. Mechanisms are measured even more indirectly, through response times and accessibility tasks.
Early anchoring research relied on university students in the United States, Israel and Germany, which raises the problem of WEIRD samples (Western, educated, industrialised, rich and democratic). The Many Labs project addressed this. Klein and colleagues (2014) ran 13 classic effects, including anchoring, in 36 samples from 12 countries with 6,344 participants, both online and in laboratories. Anchoring replicated consistently and was among the largest effects tested, and effect sizes depended much more on the effect than on the country or setting. Because the procedure was standardised and registered in advance and the results were published whatever they showed, the project also guards against publication bias.
Two cautions remain. Some anchoring questions in Many Labs referred to American facts, such as distances between US cities, so participants elsewhere were more uncertain and may have been more anchored as a result. Sampling many countries is also not the same as sampling many cultures, since most samples were still university students.
The heuristics-and-biases research programme judges decisions against mathematical norms, in which an irrelevant number should have no effect at all. From another perspective, using an anchor is usually reasonable. In everyday life the numbers people are given, such as asking prices, recommended doses and previous sentences for the same crime, normally carry information, so a strategy of starting from them is efficient. Laboratory studies make the strategy look irrational by using anchors that are deliberately meaningless. Many researchers combine the two views, treating anchoring as an adaptive shortcut in most environments and as a source of predictable error where someone controls the anchor.
Knowledge of anchoring can be used to manipulate as well as to protect. Retailers use inflated “original” prices and negotiators open with extreme offers. On the other hand, the same knowledge supports blind marking in schools, structured sentencing guidelines and training in the consider-the-opposite strategy for judges, doctors and negotiators. Research on anchoring usually involves mild deception, such as rigged wheels and loaded dice, which is generally judged acceptable because participants face no harm and are debriefed.
Concept | How it applies to anchoring | Example |
|---|---|---|
Bias | Anchoring is itself a cognitive bias; research on it can show sampling and publication bias | WEIRD student samples; Many Labs as a correction |
Causality | Random allocation supports causal claims about the anchor, less so about the mechanism | Englich et al. (2006) Study 3 vs Study 4 |
Change | Debiasing reduces but rarely removes the effect | Car experts listing arguments against a suggested price |
Measurement | The effect is measured as a group difference; people cannot report it themselves | Estate agents denying use of the listing price |
Perspective | Normative versus ecological views of whether anchoring is irrational | Asking prices as useful information |
Responsibility | Mild deception in studies; ethical use of the findings | Loaded dice; sentencing reform |
Anchoring bias is an approach term, so Paper 1 Section A can ask about it directly, and Paper 1 Section B can set a scenario in any of the four contexts and ask how anchoring explains a decision in it. Anchoring research is also strong evidence in a Paper 1 Section C essay on cognitive biases in the learning and cognition context, and the anchoring experiment is a common class practical for Paper 2 Section A.
Anchoring bias is the tendency for an estimate or decision to be drawn towards an initial value that has been considered, even when that value is irrelevant.
It is studied with a comparative question (higher or lower than the anchor?) followed by an absolute estimate, in a true experiment with random allocation.
Tversky and Kahneman (1974) found median estimates of 25% and 45% after a rigged wheel stopped at 10 or 65; Strack and Mussweiler (1997) showed that even an absurd anchor (Gandhi dying at 140) raises estimates.
Insufficient adjustment explains anchoring as stopping too early when adjusting from a starting point; it fits self-generated anchors best.
Selective accessibility explains anchoring as a confirmatory search that makes anchor-consistent knowledge more accessible; it explains absurd and externally provided anchors.
Expertise, incentives and warnings reduce anchoring only a little; knowledge and considering the opposite reduce it more. Incidental anchors produce weak, unreliable effects.
Englich, Mussweiler and Strack (2006) showed that experienced legal professionals’ sentences were anchored by demands they knew to be random, including dice they threw themselves.
The effect replicates across 12 countries (Many Labs), but its mechanism is less certain, its size in everyday decisions is debated, and it may be adaptive when anchors carry information.
I can describe anchoring bias and explain why it is a systematic bias rather than random error.
I can describe the standard anchoring paradigm and how the anchoring effect is calculated.
I can explain insufficient adjustment and selective accessibility and say which type of anchor each explains best.
I can explain how uncertainty, expertise, incentives and incidental anchors change the size of the effect.
I can describe the aim, method, results and conclusions of Englich, Mussweiler and Strack (2006).
I can apply anchoring bias to a scenario in any of the four contexts, naming the anchor and the mechanism.
I can evaluate anchoring research with reference to causality, measurement, bias, perspective and responsibility.
I can explain anchoring bias with one example of behaviour, showing how the example fits the definition.