The One That Held

Participants watched a wheel spin, saw it land on a number, and were then asked what percentage of United Nations member states were African. The wheel had nothing to do with Africa and everybody could see that. Their answers moved with it anyway.

Our Grade For This Claim

Applying the scheme from the first article in this series.

The existence and direction of the anchoring effect is Grade A. It has been recovered in large multi-laboratory replications running thousands of participants across many samples[1], and is described in the current negotiation literature as ranking amongst the most robust phenomena in human decision-making and in psychological science[2].

The size of any particular reported effect is Grade C. One source states the position exactly: the direction of the effect is robust; it is the size of any single reported number that deserves caution, and well-powered replications suggest the size is often smaller than originally reported[1].

The debiasing findings are Grade C. They rest on individual studies we did not obtain, and are reported here as a review article describes them.

This is the first Grade A claim in this series, and it is worth saying why that matters. The halving rule from the first article still applies to the numbers. It does not apply to the phenomenon.

A Note On Method

Everything here is verified to August 2026 against the primary literature or its published abstracts and reviews.

We did not obtain the full text of any primary study cited. We rely on peer-reviewed review articles and on a psychology reference resource, each identified.

The debiasing findings, the expert study and the correlation figure all come to us through review articles rather than from the original papers, and we name the intermediate source in each case.

One source is a psychology education website rather than a peer-reviewed publication[1]. We use it for its summary of the replication position and flag it.

We have not obtained effect sizes from the multi-laboratory replications and state none.

This article reviews decision research. It is not negotiation, legal or investment advice, and the conduct of any particular negotiation carries obligations this article does not address.

Why A Surviving Finding Matters

A note on the shape of this series so far. This section is our own.

The four preceding articles reported a field where roughly a third of replications reach significance, a management construct whose causal claim outruns its evidence, a consumer finding whose average effect is approximately zero, and a motivation literature where four meta-analyses disagree.

Read together, those could produce the wrong conclusion: that behavioural science has nothing usable in it.

Three reasons that conclusion would be wrong.

The scrutiny is itself the discipline working. A field that runs multi-laboratory replications of its own popular ideas and publishes null results is not one to write off.

Some findings survive that scrutiny, and anchoring is the clearest example we have encountered.

And the surviving findings are more valuable precisely because the field checked. A result that has been through large preregistered replication and come out intact is in a different evidentiary class from one that was simply never tested.

This article is about what to do with a finding of that kind, which is a different task from the previous four.

The Wheel Of Fortune

The founding demonstration, still the clearest.

Tversky and Kahneman (1974) asked participants to judge the percentage of African countries in the United Nations against either a low anchor of 10 percent or a high anchor of 65 percent. Those anchors came from the ostensible spin of a wheel with no relevance to judgments about Africa, and the anchors nevertheless influenced such judgments strongly: participants who received the high anchor estimated the percentage to be higher than those who received the low anchor[5].

The authors defined the anchoring effect as an insufficient adjustment from an initially obtained value[6], and showed that judgments of people are skewed when they are provided with an arbitrary number before the judgment[7].

Three features make this a strong demonstration, and this assessment is ours.

The anchor was visibly random. Participants watched the wheel. There was no plausible informational content.

The judgment was about an unrelated domain. A number produced by a wheel has no bearing on African representation at the UN.

And the effect appeared anyway, which rules out the most obvious deflationary explanation, that people were treating the anchor as a hint.

What The Large Replications Found

The evidence that puts this in a different category from most of what this series has covered.

A summary of the current position states that large multi-laboratory replications running thousands of participants across many samples have recovered clear anchoring effects, making it one of the more reproducible findings in the field, citing Klein and colleagues (2014)[1].

The negotiation literature treats this as settled. A 2025 review states that the anchoring effect ranks amongst the most robust phenomena in human decision-making, as well as psychological science and organizational behavior, citing the same multi-laboratory work alongside the original[2].

Another review describes it as one of the most well-known and well-documented decision heuristics, again citing Klein and colleagues[5].

Two observations, ours.

The convergence across independent review articles is itself informative. Three separate literatures, writing for different audiences, reach the same characterisation and cite the same replication.

And note what the replication was. Not a single well-run study, but a coordinated multi-site effort of exactly the kind the first article in this series described being used to demolish other findings. The same instrument that killed ego depletion confirmed anchoring.

Direction Robust, Size Uncertain

The qualification, stated by our source with unusual precision.

The summary is worth quoting closely: the direction of the effect is robust; it is the size of any single reported number that deserves caution. And separately, that well-powered replications suggest anchoring's direction is robust, but its size is often smaller than originally reported[1].

Three consequences, ours.

This is exactly the pattern the first article in this series predicted from the replication projects: effects that are real but whose published magnitudes are inflated, with replication estimates running at roughly half.

Which means the halving rule applies here too, notwithstanding the Grade A. A specific percentage from a 1990s anchoring study should be treated as an upper bound.

And for a practitioner the distinction is workable. You can rely on the direction and should not rely on the coefficient. Knowing that a first offer pulls the settlement toward it is actionable. Knowing that it pulls it by a specific percentage is not, because that number is unreliable.

Our own view is that this is the single most useful epistemic habit available in this field: separate does it happen from how much, and grade them differently. Almost every finding in behavioural science is stronger on the first than the second.

The Estate Agents

The study that decides whether this is a laboratory curiosity.

Northcraft and Neale (1987) investigated whether subject matter expertise could reduce the strength of the anchoring effect. They compared students and real estate experts in a real-world setting, and found both subject types were impacted by the anchoring bias in the same way[4].

Another source describes the same finding in a wider list, noting that estate agents pricing real houses and judges sentencing on realistic case files were both swayed by numbers they knew to be arbitrary[1], the judicial work being attributed to Englich, Mussweiler and Strack (2006)[5].

Three observations, ours.

The experts were operating in their own domain of professional competence, on the task they are paid to perform, using their own valuation methods.

Expertise did not attenuate the effect. Not partially, on the description we have: the same way.

And the population generalises directly to the readers of this publication. An accountant valuing a business, a contractor pricing a job, a broker setting a listing price and a lawyer assessing a claim are all doing the same kind of task as the estate agents in that study.

And They Said It Had No Effect

The detail that makes this operationally important.

The same summary records that the experts were swayed by numbers they knew to be arbitrary while insisting the numbers had no effect on them[1].

Three consequences, ours, and this is the part we would emphasise to any professional reading this.

Introspection is not a detector. The people affected could not tell they were affected, and reported confidently that they were not. That is a specific and well-evidenced failure of self-knowledge on this particular bias.

Which means the sentence "I know what this is worth, the number they opened with doesn't affect me" is exactly what the affected experts said. It is not evidence of immunity. It is the observed response of people who were moved.

And it explains why the debiasing findings below matter so much. If you cannot detect the effect by looking inward, you cannot correct for it by trying harder. You need a procedure.

We would put this more strongly than we usually would. On the evidence available, professional confidence about being unaffected by an opening number carries approximately zero information about whether you were affected.

First Offers In Negotiation

The application, which is where this becomes a business question.

The mechanism as the negotiation literature describes it: when presented with a numerical anchor value, such as a first offer, before giving one's own estimate or proposal, individuals assimilate their proposal to the previously considered anchor, making a counteroffer close to the first offer[2].

A review states that negotiators across cultures and contexts cognitively anchor on the first offer, making counteroffers and agreeing to settlement prices that assimilate to the first offer across various circumstances, citing Galinsky and Mussweiler (2001)[5].

Another puts the practical conclusion directly: first offers significantly influence negotiation outcomes, and making the first offer appears to be an important success factor for negotiations[3].

Two observations, ours.

The effect operates on both parties. One source notes the figure makes offer-consistent information more accessible to both sides[1], which means the person who opens is also anchored by their own number.

And the finding is about settlement prices, not merely about counteroffers. The anchor survives the intervening exchange and shows up in where the deal lands.

What Point Four Nine Seven Means

The single most useful number in this literature, with our own working shown.

A meta-study by Orr and Guthrie (2005) reported a correlation of .497 between initial offers and final negotiation outcomes[3].

Squaring that gives approximately 0.247, so on this figure roughly 25 percent of the variance in where negotiations finish is associated with where they opened.

Three cautions before anyone uses that, and they are ours.

This is a correlation, not a causal coefficient. Negotiations over more valuable things will have both higher opening offers and higher settlements, which inflates the association without any anchoring occurring.

We did not obtain the meta-study and take the figure from a review article.

And the size caution above applies with full force: this is precisely the sort of specific coefficient the replication literature suggests treating as an upper bound.

What survives those cautions is still substantial. Even heavily discounted, the opening number is one of the largest identifiable influences on the outcome of a commercial negotiation, and it is entirely within one party's control.

What That Is Worth To A Business

The magnitude, computed by us on illustrative volumes.

Consider what a one percent shift in average settlement is worth annually.

A firm doing 24 negotiations a year averaging $45,000: roughly $10,800.

A firm doing 60 at $18,000: roughly $10,800.

A firm doing 8 at $220,000: roughly $17,600.

Two observations, ours.

These are illustrative arithmetic on assumed volumes, not predictions of what anchoring will deliver. We are not claiming a one percent improvement is available; we are showing what one would be worth if it were.

And the point of the calculation is the comparison with the cost of the intervention. Preparing a defensible opening number and a supporting rationale takes hours. There is no capital outlay, no system, no headcount and no ongoing cost.

Very few behavioural interventions have that profile. Most of what this series has covered requires changing a culture, a compensation scheme or a product range. This one requires deciding what number to say first, and deciding it before the meeting rather than during it.

The Cost You Cannot See

An asymmetry this series keeps encountering, in a new setting. This section is our own analysis.

A business that lets the other side open first, and settles, has a completed deal at a known price. Everything about that transaction is visible.

What is not visible is where it would have landed had the business opened. There is no counterfactual settlement to compare against.

Three consequences.

The cost of losing the anchor is structurally unobservable, which means it is never in anyone's numbers and never triggers a review.

A negotiator who consistently lets the other side go first will accumulate a record of completed deals and no evidence of underperformance.

And this is the same shape as the error-reporting problem in the psychological safety article and the deferral problem in the choice overload article. The failure produces no artefact.

Our own observation is that this recurs often enough across unrelated literatures to be worth naming as a general principle: the behavioural errors that persist are disproportionately the ones that leave no trace in a ledger.

Two Competing Explanations

Why the effect happens, which turns out to determine what fixes it.

Tversky and Kahneman proposed insufficient adjustment: a person starts from the anchor and adjusts toward their own estimate, stopping too soon[6].

Later research argued for the selective accessibility account, in which information is recalled and interpreted selectively to fit the anchor, attributed to Chapman and Johnson (1999), Mussweiler and Strack (1999, 2001) and Strack and Mussweiler (1997)[6].

The distinction matters, and this framing is ours.

Under insufficient adjustment, the anchor is a starting point you fail to travel far enough from. The problem is one of effort and stopping rules.

Under selective accessibility, the anchor changes what comes to mind. Encountering a high number makes reasons the true value might be high more available, so your own independent-feeling estimate is built from a biased sample of considerations.

Those predict different remedies. The first suggests trying harder to adjust. The second suggests that trying harder will not help, because the raw material of your reasoning has already been selected.

The Test That Distinguishes Them

An elegant experiment that discriminates between the two accounts.

Strack and Mussweiler (1997) showed that anchors on a different scale from the estimation question produced a weaker anchoring effect. Their example compared an anchor concerning the height of the Brandenburg Gate against an estimation question about its width[6].

Two observations, ours.

Under pure insufficient adjustment, the scale should not much matter. A number is a starting point regardless of what it measures.

Under selective accessibility, it matters a great deal. Considering the height of a structure makes height-related knowledge accessible, which is only partly useful for judging width, so the effect should attenuate. That is what was found.

We are describing a single study through a review article and did not obtain it. But the result is worth understanding because it explains the debiasing evidence below, which otherwise looks arbitrary.

Our own summary: the weight of this evidence suggests anchoring is substantially a problem of what comes to mind, not merely a problem of arithmetic laziness. That is why the fixes that work are the ones that change what comes to mind.

Why The Mechanism Decides The Fix

The bridge between theory and practice, and it is ours.

If anchoring works by selectively making anchor-consistent information accessible, then three predictions follow.

Warnings should fail. Telling someone a number is arbitrary does not un-retrieve the considerations it has already brought to mind.

Motivation should fail. Paying someone to be accurate does not give them access to a different set of considerations.

Only interventions that generate anchor-inconsistent information should work, and they should need to do so actively rather than by instruction.

The remarkable thing, set out in the next two sections, is that the published evidence matches all three predictions.

Three Things That Do Not Work

The negative findings, which are more useful than they look.

Expertise does not work. Real estate experts and students were affected in the same way[4].

Financial incentives do not work. Contingent rewards did not seem to change the effect of anchoring, attributed to Diaz and colleagues (1999)[4].

Warnings do not work. Automated warnings about values that are too high in a decision support system did not reduce anchoring effects, attributed to George and colleagues (2000)[4].

Three consequences, ours.

The standard organisational responses to a known bias are training, incentives and system warnings. On this evidence, all three are the wrong tools for this particular problem.

The incentive finding is especially worth noting given the previous article in this series. Paying people to be accurate did not make them accurate here, which is a further caution against treating incentives as a general-purpose lever.

And the warning finding has a specific implication for anyone building a pricing or quoting system: a flag that says "this value looks high" does not fix the judgment that produced it.

We report all three through a single review article and did not obtain the underlying studies.

Three Things That Do

The positive findings, and the practical core of this article.

The same review reports that certain strategies did reduce the effect. Thinking about other alternatives to the negotiation, the opponent's reservation price, or one's own target reduced the anchoring effect[4].

Relatedly, it notes that the mechanism involves attending to information that is inconsistent with the anchor, citing Galinsky and Mussweiler (2001)[4].

Set the three out precisely.

Your alternatives. What happens if this deal does not close. What else you would do with the capacity, the capital or the time.

Their reservation price. The worst outcome the other side would still accept, reasoned from their position rather than yours.

Your own target. The number you decided you wanted, before the conversation started.

Two observations, ours.

Each of these is a concrete cognitive task with an output, not an instruction to be careful. You can write down the answer.

And each generates information the anchor did not supply, which is exactly what the selective accessibility account predicts should be required.

The Common Thread

What the working interventions share, and it is ours.

All three require you to generate a number of your own, from your own reasoning, before or independently of theirs.

Three consequences follow.

The timing matters. Once the other side's number has been said, it has already done its work on what comes to mind. Preparing your target after hearing their offer is doing the exercise on contaminated inputs.

Which means the practical instruction is not resist the anchor. It is arrive with your own, worked out in advance, written down.

And that is the same act as making the first offer, approached from the defensive side. A party who has done the preparation is both better protected if the other side opens and better positioned to open themselves.

Our own view is that this reconciles the two halves of the literature neatly. The advice to make the first offer and the advice to consider your alternatives and target are not separate techniques. They are the same preparation, and the preparation is the intervention.

A Protocol You Can Run

Our own construction from the evidence above, offered as a procedure rather than as a finding.

Before any negotiation of consequence, write down three numbers. Your target, your walk-away, and your best estimate of theirs.

Write down what happens if this does not close. Specifically, not as a feeling. What you do with the capacity or the capital instead.

Write down two reasons the other side's likely opening number would be wrong. This is the anchor-inconsistent information the evidence points to, generated in advance.

Date the document and do not revise it during the negotiation. Its value is that it was produced before contamination; editing it afterwards destroys the only property that makes it useful.

Open, where you reasonably can. The literature describes making the first offer as an important success factor.

If they open first, do not counter in the same conversation. Return to your prepared document before responding.

Two things about this protocol, and both are ours.

It costs under an hour and requires nothing but a document.

And it is testable in your own business, which the first article in this series argued is stronger evidence for you than any literature. Run it on half your negotiations and compare, and you will learn something about your firm that no published study can tell you.

When Not To Go First

The limits of the advice, because unqualified rules are the failure mode this series is trying to avoid.

Three situations where opening is not obviously right, and these are our own reasoning rather than findings.

When you are badly informed about value and they are well informed. An opening number reveals your estimate, and if it is far below what they would have paid, you have capped yourself.

When the relationship is the asset and an aggressive number would damage it more than the price difference is worth.

When there is no established range at all, where an opening figure may read as arbitrary rather than as a position.

We note that the literature we obtained does not systematically address when not to make the first offer, and one source refers to variables moderating the effect of first offers as an active research area[3] without our being able to report what those moderators are.

The honest statement is therefore: the general finding favours opening, the boundary conditions exist, and we did not obtain them.

The Part Worth Being Careful About

A closing note, and it applies to us. This section is our own.

Everything above describes a reliable way to move another party's judgment using a number, on a bias they cannot detect in themselves and that expertise does not protect against.

That is a capability, and it is worth being clear-eyed about what kind.

Three distinctions we would draw.

Opening with a number you can justify is ordinary commercial conduct. You are entitled to state your position first and to have thought about it.

Opening with a number designed to exploit the counterparty's inability to detect anchoring, in a context where they cannot reasonably inform themselves, is a different act, and the fact that it works is not an argument that it is acceptable.

And the asymmetry is largest where the other party is least resourced, which in most professional contexts means an individual dealing with a firm.

A firm advising on this subject, including this one, should notice that it is describing a technique it is itself positioned to use on clients. We think the honest position is that the protocol above is most valuable as protection, and that its use as a weapon against an unrepresented counterparty is a choice rather than a consequence of the research.

The Limits Of This Analysis

Several caveats matter. This article reviews decision research and is not negotiation, legal or investment advice. Everything is verified to August 2026. We did not obtain the full text of any primary study cited, including Tversky and Kahneman (1974), Klein and colleagues (2014), Galinsky and Mussweiler (2001), Northcraft and Neale (1987), Orr and Guthrie (2005), Strack and Mussweiler (1997), Diaz and colleagues (1999), George and colleagues (2000), or Englich and colleagues (2006). Every figure and finding reported here reaches us through peer-reviewed review articles or a psychology education resource, and we name the intermediate source in each case. We obtained no effect sizes from the multi-laboratory replications and state none. The correlation of .497 is a correlation, not a causal coefficient, and is subject to the confound that negotiations over more valuable subjects will show both higher openings and higher settlements. One principal source is a psychology education website rather than a peer-reviewed publication. The debiasing findings, both positive and negative, come to us through a single review article, and we did not obtain any of the underlying studies. We did not obtain the boundary conditions under which making the first offer is disadvantageous, which one source identifies as an active research area. Our arithmetic on what a one percent shift is worth uses assumed volumes and is illustrative only; it is not a prediction of what anchoring can deliver. The mechanism analysis, the invisible-cost observation, the six-step protocol, the three situations for not opening and the closing ethical discussion are our own reasoning, not findings from the literature.

Related Reading

No. 6 Correcting The Corrections A statistical correction applied for fifty years turned out to be wrong. Structured interviews now rank above cognitive ability, and the author calls... Behavioural Finance Four Meta-Analyses, Two Answers Where paying people for accuracy also failed to produce it. Behavioural Finance The Jam Study Is Not A Rule The other side of the ledger, where the famous finding did not generalise.

Frequently Asked Questions

Is anchoring actually real, given everything else has been questioned?

Yes, on the strongest evidence available in this field. Large multi-laboratory replications running thousands of participants recovered clear anchoring effects, and independent review articles across psychology, decision-making and organisational behaviour describe it as among the most robust findings in the discipline.

Does experience protect me?

On the evidence, no. Real estate experts pricing actual properties were affected the same way as students. More pointedly, they were swayed by numbers they knew were arbitrary while insisting the numbers had no effect on them, which means confidence about being unaffected carries no information.

Should I make the first offer?

The literature describes making the first offer as an important success factor, with one meta-study reporting a correlation of .497 between initial offers and final outcomes. That is a correlation rather than a causal coefficient, and we did not obtain the boundary conditions under which opening is disadvantageous.

What actually reduces the effect?

Three things are reported to: thinking about your alternatives to the negotiation, about the other side's reservation price, and about your own target. What is reported not to work: expertise, contingent financial rewards, and automated warnings that a value looks too high.

Why do warnings fail when those three work?

The likely reason is the mechanism. If anchoring works by making anchor-consistent information more accessible, a warning does not un-retrieve what has already come to mind. The three that work all generate anchor-inconsistent information actively, which is what the selective accessibility account predicts is needed.

What is the minimum useful action?

Write down your target, your walk-away and your estimate of theirs before the conversation, and do not revise it during. Its entire value is that it was produced before the other number arrived. That takes under an hour and requires nothing but a document.

References

1. Simply Psychology. Anchoring Bias and Adjustment Heuristic in Psychology, on large multi-laboratory replications running thousands of participants across many samples having recovered clear anchoring effects, making it one of the more reproducible findings in the field, citing Klein and colleagues (2014); on the direction of the effect being robust while the size of any single reported number deserves caution; on well-powered replications suggesting the direction is robust but the size often smaller than originally reported; on estate agents pricing real houses and judges sentencing on realistic case files both being swayed by numbers they knew to be arbitrary while insisting the numbers had no effect on them; on an aggressive opening offer dragging the final agreement towards it because the figure makes offer-consistent information more accessible to both sides, citing Galinsky and Mussweiler (2001); and on Tversky and Kahneman (1974) having first identified the effect, which persists even when the anchor is completely random. Note: a psychology education website, not a peer-reviewed publication; we did not obtain any of the studies it describes. simplypsychology.org
2. Organizational Behavior and Human Decision Processes. (2025). The power and peril of first offers in negotiations: a conceptual, meta-analytic, and experimental synthesis, on the anchoring effect ranking amongst the most robust phenomena in human decision-making, as well as in psychological science and organizational behavior, citing Chapman and Johnson (1999), Tversky and Kahneman (1974), Kahneman (1992) and Klein and colleagues (2014); and on individuals presented with a numerical anchor value such as a first offer assimilating their own proposal to the previously considered anchor. Note: a peer-reviewed review; we obtained the opening section only. sciencedirect.com
3. Negotiation and Conflict Management Research. Toward a Process Model of First Offers and Anchoring in Negotiations, on first offers significantly influencing negotiation outcomes, citing Galinsky and Mussweiler (2001), Kristensen and Gärling (2000a) and Ochs and Roth (1989); on a meta-study by Orr and Guthrie (2005) reporting a correlation of .497 between initial offers and final negotiation outcomes; on making the first offer appearing to be an important success factor; and on variables moderating the effect of first offers being an area of investigation. Note: a peer-reviewed article; we did not obtain the Orr and Guthrie meta-study itself. ncmr.lps.library.cmu.edu
4. Negotiation and Conflict Management Research, same article, on Northcraft and Neale (1987) having compared students and real estate experts in a real-world setting and found both impacted by the anchoring bias in the same way; on contingent rewards not seeming to change the effect of anchoring, citing Diaz and colleagues (1999); on automated warnings about values that are too high in a decision support system not reducing anchoring effects, citing George and colleagues (2000); on attending to information inconsistent with the anchor, citing Galinsky and Mussweiler (2001); and on thinking about other alternatives to the negotiation, the opponent's reservation price, or one's own target having reduced the anchoring effect. Note: all debiasing findings reach us through this single review article and we did not obtain any of the underlying studies. ncmr.lps.library.cmu.edu
5. Organizational Behavior and Human Decision Processes. (2018). "I was going to offer $10,000 but...": The effects of phantom anchors in negotiation, on Tversky and Kahneman (1974) having asked participants to judge the percentage of African countries in the United Nations against a low anchor of 10 percent or a high anchor of 65 percent, the anchors coming from the ostensible spin of a wheel with no relevance to the judgment yet influencing it strongly; on anchoring being one of the most well-known and well-documented decision heuristics, citing Klein and colleagues (2014); on anchors influencing decisions in domains including purchasing quantity, auction bidding and investment choice; on negotiators across cultures and contexts anchoring on the first offer and agreeing to settlement prices that assimilate to it, citing Galinsky and Mussweiler (2001); and on judicial verdicts, citing Englich, Mussweiler and Strack (2006). Note: a peer-reviewed article; we obtained portions. sciencedirect.com
6. Negotiation and Conflict Management Research, same article, on Tversky and Kahneman (1974) defining the anchoring effect as an insufficient adjustment from an initially obtained value; and on later research having argued for the selective accessibility perspective, in which information is recalled and interpreted selectively to fit the anchor, citing Chapman and Johnson (1999), Mussweiler and Strack (1999, 2001) and Strack and Mussweiler (1997), with Strack and Mussweiler (1997) showing that anchors on a different scale from the estimation question, being the height rather than the width of the Brandenburg Gate, had a weaker anchoring effect. Note: we did not obtain Strack and Mussweiler (1997) and report it through this review. ncmr.lps.library.cmu.edu
7. ResearchGate record for Galinsky, A. D., & Mussweiler, T., First Offers as Anchors: The Role of Perspective-Taking and Negotiator Focus, on anchoring having been among the cognitive biases highlighted by Tversky and Kahneman (1974); on anchoring affecting salary negotiation outcomes, criminal sentencing, economic decisions and the comprehension of other people; and on Tversky and Kahneman having shown that judgments are skewed when people are provided with an arbitrary number before the judgment. Note: a publisher record; we did not obtain the paper. researchgate.net

This article reviews decision research and is not negotiation, legal or investment advice. No primary study cited was obtained; every finding reaches this article through review articles or a psychology education resource, each identified. No effect sizes from the multi-laboratory replications are reported. The .497 figure is a correlation rather than a causal coefficient and is subject to a stated confound. The boundary conditions under which making the first offer is disadvantageous were not obtained. The protocol and ethical discussion are the authors' own.