Most behavioral findings cataloged in this hub collapsed under scrutiny. Numeric anchoring did not. From the Strack-Mussweiler 1997 mechanism paper through Englich’s work on courtroom sentencing, Galinsky on negotiation, Northcraft on real-estate appraisers, and the Many Labs 2 thirty-six-country confirmation, the anchoring effect is one of the most reliably replicating findings in social cognition. Here is the honest case for a behavioral mechanism that has survived where many sibling findings did not.
If you have been reading through this hub, you have watched a long parade of canonical behavioral-science findings get dismantled. Power posing did not survive Carney’s own recantation. Ego depletion collapsed under the Hagger 2016 multi-site replication. Money priming evaporated in preregistered tests. Bargh’s elderly priming has been quietly retired. The marshmallow test shrank to something much smaller than its original claim once SES was controlled. Most of the social-priming canon, the facial-feedback hypothesis, and the original stereotype-threat magnitudes have all walked back substantially in the post-2011 reckoning.
A reasonable reader might conclude that essentially all of behavioral economics is suspect. That conclusion is wrong, and the work that Thomas Mussweiler and Fritz Strack built around the anchoring effect from roughly 1997 through 2003 is one of the clearest counter-examples to it.
Because in the same two-decade window that has produced almost all of those replication failures, the numeric anchoring effect --- the proposition that a numeric value made salient just before a judgment systematically pulls that judgment toward itself, even when the value is transparently irrelevant or random --- has held up. It has held up in the original Tversky-Kahneman 1974 wheel-of-fortune paradigm. It has held up in Strack and Mussweiler’s 1997 mechanism work. It has held up when applied to judges’ sentencing decisions in real courtrooms (Englich and Mussweiler 2001, and Englich, Mussweiler, and Strack 2006). It has held up when applied to real-estate appraisers operating in their professional domain (Northcraft and Neale 1987). It has held up in salary negotiation (Galinsky and Mussweiler 2001). It has held up in medical diagnosis. It has held up in the Many Labs 2 cross-cultural mega-replication across 36 countries (Klein et al. 2018). It is one of the very small handful of social-cognition findings that survived the replication crisis with its effect-size estimate substantially intact and its underlying mechanism intact.
This is another anti-example article in a hub full of takedowns. It exists for the same three reasons every anti-example here does. First, calibration --- readers should leave this hub knowing that “behavioral economics is mostly broken” is the wrong inference; the more accurate inference is that the field produced a small number of robust, mechanism-grounded findings and a much larger number of fragile, contextually thin findings, and the failure mode was treating those two categories as if they were the same. Second, decision-usefulness --- for an executive evaluating which behavioral concepts to actually deploy in negotiation, pricing strategy, professional judgment, or organizational decision-making, anchoring is one of the highest-confidence inputs available. Third, intellectual honesty --- a hub that catalogs the failures owes readers the parts that worked.
So here is the case for the Mussweiler-Strack anchoring research program, with the legitimate critiques included.
What Tversky and Kahneman 1974 Originally Demonstrated, and Why It Wasn’t Enough
The foundational anchoring demonstration is a single experiment buried in Tversky, A., & Kahneman, D. (1974). “Judgment Under Uncertainty: Heuristics and Biases.” Science, 185(4157), 1124—1131. DOI: 10.1126/science.185.4157.1124, the same paper that gave the field its dominant analytical vocabulary for the next forty years.
The setup was the now-famous wheel-of-fortune procedure. Subjects watched the experimenter spin a wheel that was rigged to stop at either 10 or 65. After the wheel stopped, subjects were asked whether the percentage of African countries in the United Nations was higher or lower than the number the wheel had landed on, and then asked to estimate the actual percentage. The median estimate from the group that saw 10 was 25%. The median estimate from the group that saw 65 was 45%. A transparently random number, generated by a wheel in front of the subject’s eyes, with no possible relevance to U.N. composition, had moved the central tendency of subjects’ estimates by twenty percentage points.
Tversky and Kahneman called this the anchoring-and-adjustment heuristic. Their proposed mechanism was that subjects took the anchor as a starting point and then adjusted away from it, but adjusted insufficiently, leaving the final estimate biased toward the anchor.
The 1974 paper accomplished something important --- it identified the regularity. But it left a great deal unspecified. The proposed mechanism (insufficient adjustment) was speculative and turned out to be substantially wrong. The empirical demonstration was a single study in a single paradigm. The boundary conditions were unmapped. Whether the effect would generalize beyond toy laboratory tasks to consequential real-world judgments was unknown. Whether expert judges in their own domain would be susceptible, or whether expertise would protect against anchoring, was unknown. Whether the effect was an artifact of the specific U.N.-Africa task or a general feature of numeric judgment was unknown.
The research program that answered all of these questions, with the mechanism worked out properly and the domain coverage expanded, is the body of work that Mussweiler and Strack and their collaborators produced between 1997 and roughly 2010. That program is what this article is actually about.
Strack and Mussweiler 1997 --- The Mechanism Paper
The first major reconception came from Strack, F., & Mussweiler, T. (1997). “Explaining the Enigmatic Anchoring Effect: Mechanisms of Selective Accessibility.” Journal of Personality and Social Psychology, 73(3), 437—446. DOI: 10.1037/0022-3514.73.3.437.
Strack and Mussweiler argued that the insufficient-adjustment story Tversky and Kahneman had proposed was almost certainly not the right mechanism, and they ran an experimental program to demonstrate the actual one.
Their proposal was the Selective Accessibility Model. When a subject is asked to compare a target value to an anchor (“is the height of the Brandenburg Gate higher or lower than 150 meters?”), the subject does not run a numerical adjustment computation. Instead, the comparative question triggers a hypothesis-testing process in which the subject searches memory for information consistent with the anchor being the correct answer. That selective search activates anchor-consistent information disproportionately, leaving it more accessible at the moment of the subsequent absolute judgment (“how tall, actually, is the Brandenburg Gate?”). The absolute judgment then draws on the now-biased memory pool, producing the anchored estimate. The mechanism is not arithmetic; it is semantic priming of anchor-consistent content.
The Strack-Mussweiler experiments tested this proposal in several ways. The most diagnostic was that they manipulated the relevance of the anchor to the target judgment independent of its magnitude. If insufficient adjustment were the mechanism, any anchor magnitude should produce a corresponding anchoring effect regardless of relevance. If selective accessibility were the mechanism, only anchors that plausibly retrieved relevant information about the target should produce the effect. The data favored the selective accessibility account: implausible anchors that did not plausibly cue target-relevant information produced much smaller anchoring effects than plausible anchors of equivalent numerical extremity.
The 1997 paper accomplished three things. It replaced a wrong mechanism with a right one. It embedded the anchoring effect in the broader literature on semantic priming and hypothesis-testing, giving it theoretical traction. And it predicted a series of boundary conditions (the effect should attenuate when the comparative question does not cue useful memory search; the effect should be sensitive to the semantic relationship between anchor and target) that subsequent research would test and largely confirm.
Mussweiler-Strack 2000 --- The Category And Exemplar Extension
Mussweiler, T., & Strack, F. (2000). “The Use of Category and Exemplar Knowledge in the Solution of Anchoring Tasks.” Journal of Personality and Social Psychology, 78(6), 1038—1052. DOI: 10.1037/0022-3514.78.6.1038 extended the selective accessibility framework by specifying what kinds of memory content get selectively activated.
The paper distinguished between category-level knowledge (general beliefs about a class of objects --- typical heights of buildings, typical prices of cars) and exemplar-level knowledge (specific instances --- the Brandenburg Gate, a particular car model). It showed that anchoring effects operate through both channels but in different ways. Comparative questions with anchors near the center of the relevant category distribution tend to activate category-level information; comparative questions with anchors near specific known exemplars tend to activate exemplar-level information. In both cases the absolute judgment is then biased toward the type of information selectively activated.
This refinement matters for two reasons. First, it explains why some anchoring effects are larger than others without invoking ad-hoc moderators --- the size of the effect depends predictably on whether the anchor activates a content pool that pulls the target judgment further or less far from the unbiased estimate. Second, it predicts which anchor positions will be most influential in applied settings: anchors positioned near plausible exemplar values activate exemplar memory in a way that propagates strongly to subsequent judgments.
The 2000 paper, together with Mussweiler’s 2003 theoretical synthesis (Mussweiler, T. (2003). “Comparison Processes in Social Judgment: Mechanisms and Consequences.” Psychological Review, 110(3), 472—489. DOI: 10.1037/0033-295X.110.3.472), turned anchoring from an isolated curiosity into a special case of a much broader theory of comparison-based judgment, where comparative processing systematically biases subsequent absolute judgments through the selective activation of similar versus dissimilar features. That broader theoretical embedding is much of why the anchoring literature has aged well --- the empirical findings rest on a mechanism that connects to a substantial body of work in social cognition on assimilation and contrast effects, and that mechanism has independent experimental support beyond the anchoring-task data.
The Applied Extension: Anchoring In Real Judgment Domains
The strongest evidence that anchoring is not a laboratory artifact came from a series of papers that took the effect into real professional judgment domains, where expertise should have protected against it and the financial or social stakes were real.
Northcraft, G. B., & Neale, M. A. (1987). “Experts, Amateurs, and Real Estate: An Anchoring-and-Adjustment Perspective on Property Pricing Decisions.” Organizational Behavior and Human Decision Processes, 39(1), 84—97. ran the most-cited expertise test. They took real-estate agents to an actual house in Tucson, gave them the standard real-estate-agent inspection materials including a ten-page packet on comparable sales, and asked them to appraise the property’s market value. The experimental manipulation was the listing price shown in the packet. Different groups of agents saw different listed prices, randomly varied around a central value, with all other information held constant.
The agents’ appraised values tracked the listed prices substantially. Agents who saw a higher listing price appraised the property higher. Agents who saw a lower listing price appraised it lower. The spread in appraisals tracked the spread in listing prices, with anchoring effects in the range of $11,000 on a roughly $150,000 property. Importantly, when asked afterward to rank the factors that had influenced their appraisal, the agents ranked the listing price near the bottom of the list. They were systematically anchored on a price they explicitly denied being influenced by, in a professional judgment domain where their expertise was real.
Englich, B., & Mussweiler, T. (2001). “Sentencing Under Uncertainty: Anchoring Effects in the Courtroom.” Journal of Applied Social Psychology, 31(7), 1535—1551. DOI: 10.1111/j.1559-1816.2001.tb02687.x took the effect into criminal sentencing. Englich and Mussweiler presented experienced German trial judges with a realistic case file and asked them to recommend a sentence. Before the recommendation, the judges were exposed to a sentencing demand that was experimentally manipulated. Judges exposed to a higher demand recommended substantially longer sentences than judges exposed to a lower demand, with the cases otherwise identical. The follow-up paper, Englich, B., Mussweiler, T., & 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, took the demonstration further: judges exposed to a sentencing recommendation that was explicitly generated by throwing dice in the courtroom still gave substantially different sentences depending on which random number had been rolled. The judges knew the number was meaningless. The number still moved their decisions by several months of recommended imprisonment.
Galinsky, A. D., & Mussweiler, T. (2001). “First Offers as Anchors: The Role of Perspective-Taking and Negotiator Focus.” Journal of Personality and Social Psychology, 81(4), 657—669 took the effect into negotiation. Across multiple negotiation experiments, the side that made the first offer obtained a more favorable settlement on average than the side that responded, with the size of the first-offer-anchoring advantage substantial and reliable. The paper also identified moderators (perspective-taking and the focal information used by the responder) that meaningfully attenuate the effect, giving negotiators specific tools for defending against an anchored first offer rather than just identifying the problem.
Other domain extensions through the same era documented anchoring in medical diagnosis (where prior clinical hypotheses anchored subsequent test interpretation), charitable giving (where suggested-donation amounts anchored actual gift sizes), willingness-to-pay elicitation (where reference values in survey instruments substantially shaped reported willingness-to-pay even with explicit warnings about the reference being arbitrary), and several other applied judgment domains. By the mid-2000s the empirical record on anchoring covered laboratory tasks, professional judgment, legal decisions, consumer behavior, negotiation, and a number of other real-stakes domains, with consistent direction and broadly consistent effect sizes across all of them.
Many Labs 2 --- The Cross-Cultural Confirmation
The decisive test for whether anchoring would survive the replication crisis was the Many Labs 2 project: Klein, R. A., et al. (2018). “Many Labs 2: Investigating Variation in Replicability Across Samples and Settings.” Advances in Methods and Practices in Psychological Science, 1(4), 443—490. DOI: 10.1177/2515245918810225.
Many Labs 2 was a coordinated, preregistered, large-N replication project that took 28 prominent social-cognition findings and tested each one in 125 samples across 36 countries, with approximately 15,000 participants per finding. The project was designed specifically to identify which findings replicated robustly across populations and which did not.
Most of the replicated effects collapsed or substantially weakened. Some classical effects disappeared entirely. A small subset replicated with effect sizes close to the original.
Anchoring was one of the strongest replicators in the entire dataset. The Many Labs 2 anchoring protocols, derived from Jacowitz and Kahneman 1995 (which used the Strack-Mussweiler-style paradigm), produced anchoring effects of substantial and consistent magnitude across nearly every sample. The effect was present in nearly every country tested. The cross-cultural variation in effect size was modest relative to most other findings tested. The effect did not require any specific cultural background, language, or demographic context to manifest. By the standard the replication-crisis literature has been holding social-cognition findings to, anchoring passed with one of the strongest marks of any classical effect.
This is not a small data point. It is essentially the definitive cross-cultural test of whether the anchoring effect is a real, robust feature of human cognition or a contextually narrow demonstration that happens to have been over-replicated in WEIRD (Western, Educated, Industrialized, Rich, Democratic) student samples. The answer the data give is that it is real and robust, and the effect-size estimate from the meta-analytic Many Labs 2 result is the number to use rather than the typically larger effect sizes reported in individual original studies (which were subject to the usual publication-bias-driven inflation).
Why It Survived Where Many Sibling Findings Failed
The honest question for any anti-example article in this hub is: what is structurally different about this finding versus the much larger catalog of findings that did not survive? Three things, in the anchoring case.
First, the empirical base is genuinely large and diverse. The anchoring literature contains hundreds of studies across multiple paradigms, multiple expertise levels, multiple stakes contexts, and (after Many Labs 2) multiple national populations. The size and diversity of the empirical base means the finding does not rest on any single experimental setup. By contrast, several of the collapsed findings in this hub --- power posing, money priming, several of the social-priming effects --- rested heavily on a small number of original studies in a single laboratory in a single paradigm, and the cross-paradigm generalization claims were assertions rather than empirical demonstrations.
Second, the mechanism is specified at a level that connects to other independently-evidenced cognitive phenomena. The Selective Accessibility Model says that anchoring works through semantic priming of anchor-consistent content during a hypothesis-testing comparison process. That mechanism predicts, and the data confirm, things like: anchoring effects should be modulated by the semantic relevance of anchor to target (yes), the effect should be attenuated when the comparative question does not cue useful memory retrieval (yes), the effect should be linked to broader assimilation phenomena in social judgment (yes), and the effect should persist even when the anchor is explicitly random because the comparison process happens automatically (yes). Most of the collapsed findings in this hub had mechanism stories that, in retrospect, did not predict their own boundary conditions and did not connect to independently-evidenced cognitive phenomena. The mechanism was assertion-shaped rather than prediction-shaped. Anchoring’s was prediction-shaped, and the predictions held up.
Third, the applied demonstrations were not contrived. The Englich judges, the Northcraft real-estate appraisers, the Galinsky negotiators --- these were not college sophomores in a windowless room being asked about hypothetical scenarios with no consequence. These were domain experts making decisions inside their own professional context, and the effect was demonstrated to operate on them. The applied-replication record is what made anchoring credible as an account of how real judgment works, not as a curiosity of laboratory artifact. Many of the collapsed findings in this hub had nothing comparable --- the canonical demonstrations were entirely within the convenience-sample lab paradigm, the field demonstrations were thin, and the assumption that the effect would generalize from lab to applied setting turned out to be unwarranted when it was tested directly.
The combination --- broad empirical base, well-specified mechanism that makes correct predictions, applied demonstrations in real professional judgment --- is what a robust behavioral-science finding looks like. It is rare. Anchoring is one of the small handful in the broader behavioral economics literature that meets all three conditions.
What Replicates And What Doesn’t In The Anchoring Literature
Now for the honest part, because every anti-example article in this hub owes one. Not every claim ever made under the anchoring banner has held up.
What has clearly replicated:
The core anchoring effect itself --- numeric values made salient before a related judgment systematically pull that judgment toward themselves --- across hundreds of studies, multiple paradigms, multiple expertise levels, and the Many Labs 2 cross-cultural test.
The Selective Accessibility Model as the dominant mechanism, including its key predictions about semantic relevance modulation and the role of comparative-question structure.
The applied demonstrations in legal sentencing (Englich), real-estate appraisal (Northcraft), salary negotiation (Galinsky), and several other professional judgment domains.
The robustness of the effect to subjects’ awareness of the anchor’s randomness --- the dice-throwing demonstration in Englich, Mussweiler, and Strack 2006 is a particularly striking version of this.
What has been more conditional or contested:
The originally-claimed magnitude of anchoring effects, especially in the early 1970s-1980s literature, was inflated by the standard publication-bias mechanisms. The Many Labs 2 effect-size estimate is the better number to use, and it is smaller than what the marketing-textbook version of anchoring tends to imply, though still substantial and operationally meaningful.
The “incidental anchor” claim --- that anchors not embedded in any comparative question (e.g., randomly encountered numbers in unrelated tasks) still produce subsequent anchoring on a different judgment --- is more fragile than the comparative-anchoring paradigm. Some incidental-anchor demonstrations have failed to replicate at the originally-claimed magnitude. The robust version of anchoring is the version where the anchor is embedded in a comparison question relevant to the subsequent absolute judgment. The “any random number nearby influences any judgment” claim, sometimes seen in popular-press treatments, is the over-extended version and should be discounted.
Some boundary-condition claims about expertise protection have softened. The original Northcraft finding that real-estate experts are anchored remains, but the question of whether deeper domain expertise systematically attenuates anchoring versus leaves it largely intact is empirically unsettled. The conservative reading is that domain expertise provides partial but not full protection.
Specific quantitative claims about anchoring in pricing strategy --- particularly the consultant-deck versions that promise large purchase-probability lifts from specific anchor configurations --- have a much weaker evidence base than the underlying judgment research and should be evaluated separately. The pricing-anchoring article in this hub treats that applied literature more skeptically; the underlying judgment research treated in this article is the robust core, and the pricing-strategy extensions are downstream applications whose evidence base is patchier.
The honest summary: the core anchoring effect and its Mussweiler-Strack mechanism are one of the most reliable findings in behavioral science. The strongest applied demonstrations (legal sentencing, real-estate appraisal, first-offer negotiation) are well-evidenced. The popular-press version, which often implies that any random number nearby affects any judgment by a large amount, is the over-extended version and should be discounted.
What This Means For Strategists
For practitioners --- negotiators, executives making consequential judgments, marketers, lawyers, hiring managers, anyone whose decisions involve numerical estimation under uncertainty --- the strategic takeaways break in three directions.
Make the first offer when you have the information to do so credibly. The Galinsky and Mussweiler 2001 first-offer finding is one of the better-evidenced applied implications of the anchoring literature. The party who makes the first offer in a negotiation captures the anchor and reliably obtains a more favorable settlement on average, provided the offer is in the plausible-range zone where it gets treated as a credible position rather than dismissed as posturing. The common negotiator instinct to wait for the other side to anchor first is, on the available evidence, usually a mistake. Make the first offer; make it ambitious but defensible; let the other side adjust from your anchor rather than the reverse. The caveat is that the offer has to be in the credibility window --- a wildly extreme first offer can backfire by being dismissed as unserious, breaking the comparison frame the anchoring mechanism requires.
Defend against incoming anchors by reframing the comparison. Galinsky and Mussweiler’s mechanism work showed that the anchoring effect operates through selective activation of anchor-consistent content during the comparison. The corresponding defense is to disrupt that selective activation by deliberately retrieving anchor-inconsistent information before responding. In a negotiation, this means coming to the table with your own pre-prepared alternative reference points and consulting them before responding to an incoming first offer. In a hiring decision, this means consulting candidate qualifications independent of any salary band that was disclosed early. In a sentencing or quasi-judicial context, this means structuring the decision process to consult base-rate information before any specific recommended number is introduced. The structural intervention --- pre-committing to a process that retrieves opposing information --- is what works. Willpower-based resistance (“I will simply not be influenced by the anchor”) does not work, because the mechanism operates pre-consciously during the comparison process.
Recognize that professional expertise does not give you immunity. This is the single most important takeaway from the Mussweiler-Strack applied research program. Judges with decades of trial experience were anchored by sentencing demands generated by dice in front of them. Real-estate appraisers with professional credentials were anchored by listing prices they later denied being influenced by. Negotiators with formal training were anchored by first offers. The base rate of “experts are immune to this bias” is zero, and the assumption that your own professional judgment is the exception is, statistically, the same assumption every other expert population has held immediately before being shown the data. Build the structural defenses into your decision process whether or not you personally believe you need them.
Use anchoring deliberately and ethically in commercial contexts. The same mechanism that makes first offers in negotiation effective makes anchored pricing displays effective. The line between “I am presenting a credible reference point that helps the customer assess value” and “I am using a deceptive reference point to manipulate the customer’s perception of value” is the line between ethical commercial use and consumer deception, and it has real legal consequences in addition to the reputational ones. The version of anchored pricing that survives FTC scrutiny is the version where the anchor is a genuine reference (a real prior price, a real comparable product, a real competitive offering); the version that does not survive is the version where the anchor is fabricated or transparently designed to deceive. The mechanism is the same in both cases; the ethics and legality are not.
Treat the pricing-strategy version of anchoring as a separate, weaker question. The robust core of the anchoring literature concerns judgment under uncertainty. The downstream extension to consumer pricing strategy --- specific anchor configurations in pricing-page design, specific decoy effects, specific list-price-to-discount displays --- has a substantially weaker empirical base than the underlying judgment work, partly because of the file-drawer dynamics in commercial A/B-testing literature and partly because pricing-strategy effects are highly contingent on category, price level, customer expertise, and competitive context. Treat the pricing-anchoring article in this hub as the place where the applied pricing-strategy claims get evaluated on their own evidence, separate from the robust core treated here. The core anchoring finding is one of the strongest in behavioral science; the specific pricing-strategy applications deserve case-by-case evaluation against the actual measured lift in your own context.
What This Anti-Example Tells Us About Behavioral Science Overall
The replication crisis is real. The catalog of canonical-then-collapsed findings in this hub is long and important. A reasonable executive could read all of it and conclude that behavioral economics is mostly an academic vanity project --- interesting theory, weak evidence, low practical reliability.
That conclusion would be wrong, and the Mussweiler-Strack anchoring research program, alongside the availability heuristic and the default effect, is among the cleanest counter-examples to it.
What behavioral science can produce, when it does the work properly, is a small number of robust, mechanism-grounded, replicable findings that have meaningfully advanced our understanding of how decisions get made. Anchoring is one of them. Availability is another. Defaults are another. Loss aversion, in its non-inflated form, is a fourth. There are perhaps a dozen findings in the broader behavioral-economics literature that meet this standard --- well-replicated, well-mechanism-specified, well-measured at scale outside the convenience-sample lab paradigm, and confirmed in cross-cultural replication where that has been done.
The job of a sophisticated reader of this literature is to distinguish the dozen-or-so robust findings from the much larger catalog of fragile findings. This hub is a guided tour of that distinction. The Mussweiler-Strack anchoring program is one of the answers you take away as a robust finding to actually use in operational decisions. The collapsed findings catalog --- power posing, ego depletion, money priming, the marshmallow test in its inflated form, facial feedback, social priming, and the rest --- is the part you stop deploying. The synthesis is that behavioral economics is more useful than the replication-crisis critique alone would suggest, and less useful than the original consulting-deck version implied. Anchoring is one of the findings that demonstrates this calibrated middle position is the right one.
Sources
- Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124—1131. DOI: 10.1126/science.185.4157.1124
- Northcraft, G. B., & Neale, M. A. (1987). Experts, amateurs, and real estate: An anchoring-and-adjustment perspective on property pricing decisions. Organizational Behavior and Human Decision Processes, 39(1), 84—97. DOI: 10.1016/0749-5978(87)90046-X
- Strack, F., & Mussweiler, T. (1997). Explaining the enigmatic anchoring effect: Mechanisms of selective accessibility. Journal of Personality and Social Psychology, 73(3), 437—446. DOI: 10.1037/0022-3514.73.3.437
- Mussweiler, T., & Strack, F. (2000). The use of category and exemplar knowledge in the solution of anchoring tasks. Journal of Personality and Social Psychology, 78(6), 1038—1052. DOI: 10.1037/0022-3514.78.6.1038
- Englich, B., & Mussweiler, T. (2001). Sentencing under uncertainty: Anchoring effects in the courtroom. Journal of Applied Social Psychology, 31(7), 1535—1551. DOI: 10.1111/j.1559-1816.2001.tb02687.x
- Galinsky, A. D., & Mussweiler, T. (2001). First offers as anchors: The role of perspective-taking and negotiator focus. Journal of Personality and Social Psychology, 81(4), 657—669. DOI: 10.1037/0022-3514.81.4.657
- Mussweiler, T. (2003). Comparison processes in social judgment: Mechanisms and consequences. Psychological Review, 110(3), 472—489. DOI: 10.1037/0033-295X.110.3.472
- Englich, B., Mussweiler, T., & 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. DOI: 10.1177/0146167205282152
- Klein, R. A., et al. (2018). Many Labs 2: Investigating variation in replicability across samples and settings. Advances in Methods and Practices in Psychological Science, 1(4), 443—490. DOI: 10.1177/2515245918810225
- Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux. (Chapters 11-12 give the popular-press synthesis of the anchoring literature.)
Related
- Pricing Anchoring --- the downstream applied-pricing question, with a substantially weaker evidence base than the core judgment work treated here
- Prospect Theory --- the other major Tversky-Kahneman framework, and its post-replication-crisis standing
- Framing Effect --- the related demonstration that the verbal framing of identical outcomes shifts choice systematically
- Representativeness Heuristic --- the third leg of the original Tversky-Kahneman heuristics-and-biases triad
- Availability Heuristic --- the other major heuristics-and-biases finding that survived the replication crisis intact, with the same general anti-example structure as this article
FAQ
Is the anchoring effect actually robust, or is it another finding the replication crisis will eventually take down?
It is actually robust. The Many Labs 2 mega-replication tested it across 36 countries and 15,000-plus participants and found it among the strongest replicators of the 28 classical social-cognition findings examined. The effect size is smaller than the inflated estimates from the early literature, but it is substantial and present in essentially every population tested. The core finding is one of the highest-confidence results in behavioral science.
Does professional expertise protect against anchoring?
Mostly no. The most-cited applied demonstrations (real-estate appraisers in Northcraft 1987, German trial judges in Englich and Mussweiler 2001, judges shown explicitly random anchors in Englich, Mussweiler, and Strack 2006) all involved expert populations operating in their own professional judgment domain, and all showed substantial anchoring. The conservative reading of the boundary-condition literature is that expertise provides partial but not full protection. Do not assume your own professional judgment is the exception.
Why did Mussweiler and Strack’s selective accessibility account replace Tversky and Kahneman’s original insufficient-adjustment story?
Because the predictions of the two accounts diverge in specific experimental setups, and the data consistently favored selective accessibility. The Strack-Mussweiler 1997 paper demonstrated that anchoring effects depend on the semantic relationship between anchor and target in ways that an arithmetic insufficient-adjustment mechanism could not predict but a semantic-priming hypothesis-testing mechanism could. The selective accessibility framework also connected the anchoring effect to a broader literature on comparison-driven assimilation in social judgment, giving it independent theoretical support.
What is the strongest applied implication for someone in a negotiation?
Make the first offer when you have the information to make a credible one. Galinsky and Mussweiler 2001 is the canonical demonstration that the side that makes the first offer captures the anchor and obtains a more favorable settlement on average. The common negotiator instinct to wait for the other side to anchor first is, on the available evidence, usually wrong.
How is the underlying judgment research different from the pricing-strategy applications?
The underlying judgment research --- Tversky-Kahneman 1974, Strack-Mussweiler 1997, Mussweiler-Strack 2000, Mussweiler 2003, the Englich and Galinsky applied papers, Many Labs 2 --- has a large, diverse, mechanism-grounded empirical base that has survived the replication crisis intact. The pricing-strategy applications (specific anchor configurations in pricing-page design, decoy effects, list-price-to-discount displays) have a substantially weaker empirical base, partly because commercial A/B-testing literature has heavy file-drawer dynamics and partly because pricing-strategy effects are highly contingent on category, customer expertise, and competitive context. Treat the pricing-anchoring article in this hub as the place where those applied claims get evaluated on their own evidence.
What is the most striking single demonstration of the effect?
Englich, Mussweiler, and Strack 2006. Experienced trial judges were given a realistic case file, then watched dice being rolled in the courtroom to generate a sentencing recommendation, then asked to recommend a sentence themselves. The dice-generated number, which the judges knew was random and meaningless, still moved their sentencing recommendations by several months of imprisonment. This is the demonstration that most directly refutes the intuition that awareness of randomness inoculates against the anchoring effect. It does not.
If I am building defenses against anchoring in my own decision process, what actually works?
Structural interventions, not willpower. The mechanism operates pre-consciously during the comparison process, so deciding to ignore the anchor in the moment does not work. What works is pre-committing to a decision process that retrieves opposing or base-rate information before the anchor is introduced --- structured interviewing with pre-specified criteria before any salary is disclosed; sentencing guidelines consulted before any specific recommendation; negotiation preparation that includes your own pre-prepared reference points consulted before the other side speaks. The general pattern is to disrupt the selective-accessibility process by making sure non-anchor-consistent content is already activated when the comparison happens.