Snap Shot
- The question: Can a number or word supplied in a question shape a witness’s estimate of speed, distance, height, weight, duration, or quantity, or a decision maker’s numerical judgment?
- What the research shows: Estimates of uncertain quantities move toward reference values supplied before or within the question, and the effect is among the most replicated in judgment research. Witnesses’ unaided estimates of height, weight, distance, and above all how long a stressful event lasted are often imprecise, and question wording can move them. Sentencing demands influence sentences, including among experienced legal professionals.
- What it does not show: It does not show that a particular witness’s figure was produced by the question, that estimates are worthless, or that every leading question distorts memory; some leading questions in later testing did not reduce accuracy. Anchoring effects are smaller or absent under some conditions, including some uses of plainly random numbers.
- The law: Leading questions are generally barred on direct examination and permitted on cross-examination under Rule 611(c) in federal, military, and Arizona courts. Expert testimony on estimation is governed by Rule 702 and, in courts-martial, Houser; in Arizona, Chapple and Nordstrom allow general testimony but not opinions on a witness’s credibility.
- Issues for each side: Prosecutors should obtain and record estimates with open questions before any figure is suggested and corroborate them with physical evidence; defense counsel should tie anchoring claims to the actual wording and sequence of the questions, recognizing that the research does not make every estimate unreliable.
Witnesses are often asked to put numbers on what they saw: speed, distance, an offender’s height and weight, how long an encounter lasted, or how many people or blows there were. A speed estimate may separate negligence from recklessness, a duration estimate may bear on the witness’s opportunity to view a face, and a height estimate may include or exclude a suspect. Yet people have no precise internal gauge for most of these quantities, and when they must produce a number they draw on whatever reference points are available, including those supplied by the person asking.
Anchoring is the tendency of a numerical judgment to move toward a reference value, the anchor, that is considered before or while the judgment is made. Anchors can come from an investigator’s wording, a figure proposed in an interview or on cross-examination, a witness’s own body size, or a party’s sentencing recommendation, and they affect decision makers as well as witnesses.
This paper summarizes the anchoring research, leading questions with numerical content, witnesses’ accuracy for physical quantities and duration, anchoring in sentencing, and debiasing, followed by the limits of the evidence, the governing law, and the issues for each side.
Anchoring and Adjustment
In a set of demonstrations that defined the field, Tversky and Kahneman (1974) asked people to estimate quantities, such as the percentage of African countries in the United Nations, after first watching a wheel of fortune produce a number. Estimates were higher after a high number than after a low one, even though the number plainly carried no information. The authors described the process as anchoring and adjustment: people start from an available value and adjust away from it, but the adjustment tends to stop short, leaving the final estimate biased toward the starting point. When people generate their own starting value, they tend to stop adjusting once an answer seems plausible (Epley & Gilovich, 2006).
The effect is large and well replicated. A 2026 meta-analysis of the anchoring literature, covering 2,601 effect sizes, found a large average difference between estimates following high and low anchors (Hedges’ g with a 95% confidence interval of 0.77 to 0.88) that remained large after adjustment for publication bias (Schley & Weingarten, 2026). The effect was reduced or absent for incidental anchors, anchors on a different dimension, random numbers, and when incentives or debiasing interventions were present. Anchoring is also not purely automatic. In five studies, participants who intended to use an anchor, or reported having used it, showed larger anchoring effects, and anchors judged to be poor estimates had less influence, although anchoring was not eliminated even among participants who said they did not intend to use the anchor (A. J. L. Harris et al., 2026). In legal settings, an anchor supplied by an investigator or lawyer is rarely random and often appears informative.
Numbers and Words in the Question
The earliest studies of leading questions showed that the form of a question can supply an anchor. When people were asked “How tall was the basketball player?” they estimated about 79 inches on average, compared with about 69 inches when asked “How short was the basketball player?”; asking how long rather than how short a movie was produced a similar difference (R. J. Harris, 1973, as summarized in Loftus, 1975). In an unpublished study described in the same paper, people asked whether they got headaches “frequently” reported more headaches per week than people asked whether they got them “occasionally” (Loftus, 1975).
Loftus (1975) then tested whether numbers embedded in a question about a witnessed event affected later memory of it. Students watched a short videotape in which eight demonstrators entered a classroom. Those asked about the leader of the “twelve” demonstrators later recalled more demonstrators, on average, than those asked about the leader of the “four.” Questions presupposing a detail not in the film, such as a barn, increased later reports of seeing it.
The best-known study in this line used a verb rather than a number. Participants who watched films of traffic accidents gave higher speed estimates when asked how fast the cars were going when they “smashed into” each other than when the verb was “hit,” “bumped,” “collided,” or “contacted.” In a second experiment a week later, 32% of those asked the “smashed” question reported seeing broken glass, which was not in the film, compared with 14% of those asked the “hit” question and 12% of controls who were not asked about speed (Loftus & Palmer, 1974). The wording changed both the estimate and what some participants later reported remembering.
Not every leading question harms accuracy. In two experiments in which adults watched a mock crime, answered simple questions, and later answered cross-examination style questions, negative and double-negative questions sometimes reduced accuracy, but leading questions did not reduce accuracy and sometimes improved it, particularly for participants who had discriminated well between their correct and incorrect answers on the initial test (Wade & Spearing, 2023). Those questions concerned details participants had already reported once, which differs from a number suggested before a witness has offered any estimate; what matters is the timing and content of a suggested figure, not leading form alone.
Estimates of Physical Quantities
Witnesses’ unaided estimates of physical characteristics are often imprecise even without any suggestion. In a large field experiment, 1,588 participants viewed live targets from distances of 5 to 110 meters and estimated their gender, age, height, and weight. Gender was judged correctly in most cases, but estimates within two units of the true height or weight were the minority even at 5 meters, errors of ten units or more were frequent for some targets, and accuracy declined with distance and was lower for young children than for adults (Nyman et al., 2021). Height and build are among the descriptors witnesses report most often, and children’s estimates of height, weight, and age are less accurate than adults’ (Lane & Houston, 2021, pp. 28–37).
Because witnesses lack a precise standard, they may rely on familiar reference points. One familiar reference point is the witness’s own body: estimates of another person’s height and weight may be pulled toward the witness’s own height or toward a general notion of a typical person. This is a reasonable hypothesis drawn from the anchoring literature, but it has received limited published study, its size in real cases is not established, and a witness who is much shorter or taller than the person described is not thereby shown to be wrong.
Estimates of Duration
Duration is especially hard to judge. In three experiments with 469 participants, people who watched a short videotape of a bank robbery overestimated its duration, women overestimated more than men in two of the experiments, and a more stressful version of the event produced larger overestimates than a less stressful one in the third; the accuracy of duration estimates was unrelated to how much or how accurately participants recalled the event (Loftus et al., 1987). Longer exposure generally helps identification, but no study sets a threshold for how much is enough, and people misjudge how long they had to view a face (Lane & Houston, 2021, pp. 30–31). An estimate that an encounter lasted “several minutes” may overstate the time available, though overestimation is a group tendency and some witnesses estimate accurately.
The duration research also bears on interviewing. In three studies of a recorded mock crime, the sum of witnesses’ estimates for each component of the crime was less biased than their estimate of the whole event, while supplying a reference duration or merely breaking the event into parts was less helpful (Gasper et al., 2019). The related effects of breaking an event into parts are discussed in the paper on the subadditivity effect in this hub.
Anchoring in Sentencing and Legal Judgment
Anchoring affects decision makers as well as witnesses. In a series of experiments with experienced legal professionals evaluating a case, sentencing decisions moved toward a sentencing demand even when the demand came from an irrelevant source, when participants were told it had been determined at random, and when participants determined it themselves by throwing dice; expertise and experience did not reduce the effect, and a high demand made arguments for a harsher sentence more accessible (Englich et al., 2006). A meta-analysis of 29 studies (N = 8,549) of numerical legal decisions found significant anchoring effects, moderate in size in studies with a control group, with some evidence of publication bias and preliminary evidence that factors such as legal expertise and the relevance of the anchor moderate the effect (Bystranowski et al., 2021). Actual sentencing recommendations are not random; they are informed by law and the facts, and some influence of a well-founded recommendation is intended. The research shows that a figure can exert influence beyond the reasons offered for it, which is a reason to ask what supports a recommended number.
Reducing Anchoring
Knowing that an anchor is uninformative does not reliably remove its influence; anchoring was reduced, but not eliminated, among participants who judged an anchor to be a poor estimate (A. J. L. Harris et al., 2026). One strategy with experimental support is to consider the opposite: car experts asked to generate reasons why an anchor was inappropriate showed reduced anchoring in their price estimates (Mussweiler et al., 2000). The meta-analytic evidence suggests that debiasing interventions and incentives reduce anchoring on average but do not reliably eliminate it (Schley & Weingarten, 2026).
For witness estimates, the practical safeguards concern procedure: asking in open form (“How long did that last?”) without suggesting a number or loaded verb, recording the first estimate verbatim, asking what it is based on, asking about each phase of an event for duration (Gasper et al., 2019), and testing the estimate against physical evidence such as video timestamps or scene measurements.
Limits of the Evidence
Most anchoring research uses general-knowledge questions with little personal significance, and most leading-question studies use students, filmed events, short delays, and no consequences. Actual witnesses may have more at stake, different viewing conditions, and repeated questioning. The classic leading-question studies are about fifty years old and some report single experiments; the phenomenon has been replicated, but their effect sizes should not be treated as current estimates. Anchoring effects vary with the anchor’s source, plausibility, and relevance, and the 2026 meta-analysis found that some kinds of anchors, including random numbers, had reduced or no effect (Schley & Weingarten, 2026), which sits in some tension with the dice study in sentencing (Englich et al., 2006). Group research cannot establish that a particular witness’s estimate was shaped by a particular question or whether it is accurate.
Questions to Ask About This Research
When Loftus and Palmer (1974) or Loftus (1975) is cited to show that a question shaped a witness’s estimate, counsel should ask whether the conclusion follows for this witness: the studies show average differences among students who watched films, so the relevant questions are how the questions in this case were worded, whether a number or loaded verb preceded the witness’s estimate, and whether the first estimate was recorded. When Loftus et al. (1987) is cited to suggest that a witness overstated how long an event lasted, counsel should ask how closely the study’s videotaped event resembles the incident and whether any recording or timestamp can test the estimate. When the sentencing studies are cited (Englich et al., 2006; Bystranowski et al., 2021), counsel should ask who was studied and in what task, since case vignettes differ from a contested hearing. When anchoring research is offered to suggest that an estimate is meaningless, counsel should ask whether the research shows more than that estimates are imprecise and can be moved, and whether the leading questions at issue resemble those that did not impair accuracy in Wade and Spearing (2023). These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.
Anchoring and the Law
Federal and Military Standards
Federal Rule of Evidence 611(c) provides that leading questions should not be used on direct examination except as necessary to develop the witness’s testimony, and that the court should ordinarily allow them on cross-examination and when a party calls a hostile witness, an adverse party, or a witness identified with an adverse party. Military Rule of Evidence 611(c) is substantially the same, with the military judge exercising that control; its text refers to a hostile witness or a witness identified with an adverse party and does not separately list an adverse party. The rule governs questioning at trial, not pretrial interviews, where most estimates are first obtained; the wording of those questions is developed through recordings, reports, and examination of the interviewer. Expert testimony about estimation and anchoring is governed by Federal Rule of Evidence 702, as amended December 1, 2023, under which the proponent must show the court that it is more likely than not that the rule’s requirements are met, including that the opinion reflects a reliable application of reliable methods to the case. Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993) made the trial judge a gatekeeper, and Kumho Tire Co. v. Carmichael (1999) extended that obligation to all expert testimony, including testimony based on technical or other specialized knowledge. In courts-martial, Military Rule of Evidence 702 tracks the federal rule, the 2023 federal amendment took effect in the military rule in 2025 through Military Rule of Evidence 1102 (Joint Service Committee on Military Justice, 2025), and practitioners should confirm current text; military judges apply the six factors of United States v. Houser (1993). Such testimony is general and cannot tell the factfinder whether a particular estimate is accurate.
Arizona
Arizona Rule of Evidence 611(c) follows the federal formulation on leading questions. Arizona Rule of Evidence 702 adopted the federal formulation and the Daubert standard effective January 1, 2012, and was amended effective January 1, 2024, to conform to the 2023 federal amendment (Ariz. Sup. Ct. Order R-23-0004). State v. Chapple (1983) is Arizona’s leading case on expert testimony about memory and eyewitness reliability; excluding such testimony was an abuse of discretion on its facts, while admission otherwise remains within the trial court’s discretion (State v. Nordstrom, 2001). An expert may describe general principles but may not opine on whether a particular witness is truthful (State v. Lindsey, 1986, 149 Ariz. at 475).
Special Issues in Legal Practice
The principles described above apply with equal force to every party in a legal proceeding. The science operates the same way regardless of who is asking the questions, and the research summarized here does not by itself favor either side. What differs between the parties is not the science but the vantage point — the particular places where these issues tend to surface, and the particular errors each party is best positioned to guard against. The two summaries below identify those recurring issues. They are offered for orientation, not as litigation strategy; in every case the controlling question is what the evidence in that case will, and will not, support, and the expert’s role is to inform the trier of fact rather than to advocate for either side.
Special Issues for Prosecutors
The research supports several propositions prosecutors can legitimately present. An estimate obtained through an open question, recorded before any figure was suggested, and consistent across early accounts is not shown by the anchoring research to be unreliable. Leading questions on a later occasion do not necessarily reduce accuracy, and imprecision in a number does not mean the witness did not see the event or the person. Prosecutors are well positioned to ensure that investigators obtain estimates without supplying a number or loaded description, record the first estimate verbatim, and seek physical evidence that can test it.
The research also counsels caution. A precise-sounding figure may reflect the form of the question as much as the observation, and in one set of 1987 laboratory studies of a videotaped robbery (Loftus et al.), a more stressful version produced larger duration overestimates, and this paper reviews no measurements from real events. Presenting an estimate as a measurement, or relying on a duration estimate alone to establish a lengthy opportunity to view, goes beyond the research, and a sentencing recommendation is most defensible when its basis is stated explicitly.
Special Issues for Defense
The research supports several propositions defense counsel can legitimately present. Question wording, suggested numbers, and presupposed details can move estimates and later reports; witnesses’ unaided estimates of height, weight, distance, and duration are often imprecise; and, in one set of 1987 laboratory studies of a videotaped event, a more stressful version produced larger duration overestimates. Where the record shows that a number or loaded verb preceded the witness’s estimate, or that the estimate changed after such a question, expert testimony explaining the research is a legitimate way to help the factfinder weigh it.
The research also counsels caution about its limits. It does not show that a particular estimate was caused by a question, that estimates are meaningless, or that every leading question distorts memory, and some anchors have small or no effects. A mismatch between an estimate and a defendant’s actual height or weight may reflect ordinary imprecision rather than error about identity, the same research applies to defense witnesses’ estimates, and an expert cannot tell the factfinder that a witness’s figure is wrong.
This paper is for educational purposes only and is not legal advice. The author is a psychologist, not an attorney; summaries of law are general and must be verified against current authority or with a licensed attorney.
References
Bystranowski, P., Janik, B., Próchnicki, M., & Skórska, P. (2021). Anchoring effect in legal decision-making: A meta-analysis. Law and Human Behavior, 45(1), 1–23. https://doi.org/10.1037/lhb0000438
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. https://doi.org/10.1177/0146167205282152
Epley, N., & Gilovich, T. (2006). The anchoring-and-adjustment heuristic: Why the adjustments are insufficient. Psychological Science, 17(4), 311–318. https://doi.org/10.1111/j.1467-9280.2006.01704.x
Gasper, H. L., Roy, M. M., & Flowe, H. D. (2019). Improving time estimation in witness memory. Frontiers in Psychology, 10, Article 1452. https://doi.org/10.3389/fpsyg.2019.01452
Harris, A. J. L., Phillips, T., Bhaskaran, S., Krasniqi, J., Zhang, Y. S., & Shanks, D. R. (2026). Awareness of anchoring: Intentional processes contribute to anchoring effects. Consciousness and Cognition, 145, Article 104137. https://doi.org/10.1016/j.concog.2026.104137
Harris, R. J. (1973). Answering questions containing marked and unmarked adjectives and adverbs. Journal of Experimental Psychology, 97(3), 399–401. https://doi.org/10.1037/h0034165
Joint Service Committee on Military Justice. (2025). Amendments to the Military Rules of Evidence since 1984 (G. E. Maggs, Ed.; 2025 ed.). https://jsc.defense.gov/Portals/99/MRE%20Amendments%202025.pdf
Lane, S. M., & Houston, K. A. (2021). Understanding eyewitness memory: Theory and applications. New York University Press.
Loftus, E. F. (1975). Leading questions and the eyewitness report. Cognitive Psychology, 7(4), 560–572. https://doi.org/10.1016/0010-0285(75)90023-7
Loftus, E. F., & Palmer, J. C. (1974). Reconstruction of automobile destruction: An example of the interaction between language and memory. Journal of Verbal Learning and Verbal Behavior, 13(5), 585–589. https://doi.org/10.1016/S0022-5371(74)80011-3
Loftus, E. F., Schooler, J. W., Boone, S. M., & Kline, D. (1987). Time went by so slowly: Overestimation of event duration by males and females. Applied Cognitive Psychology, 1(1), 3–13. https://doi.org/10.1002/acp.2350010103
Mussweiler, T., Strack, F., & Pfeiffer, T. (2000). Overcoming the inevitable anchoring effect: Considering the opposite compensates for selective accessibility. Personality and Social Psychology Bulletin, 26(9), 1142–1150. https://doi.org/10.1177/01461672002611010
Nyman, T. J., Antfolk, J., Lampinen, J. M., Korkman, J., & Santtila, P. (2021). The effects of distance and age on the accuracy of estimating perpetrator gender, age, height, and weight by eyewitnesses. Psychology, Crime & Law, 27(3), 231–252. https://doi.org/10.1080/1068316X.2020.1798425
Schley, D. R., & Weingarten, E. (2026). Fifty years of anchoring effects: A theoretical reintegration and meta-analysis. Management Science. Advance online publication. https://doi.org/10.1287/mnsc.2023.03238
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124–1131. https://doi.org/10.1126/science.185.4157.1124
Wade, K. A., & Spearing, E. R. (2023). The effect of cross-examination style questions on adult eyewitness accuracy depends on question type and eyewitness confidence. Memory, 31(2), 163–178. https://doi.org/10.1080/09658211.2022.2129066
Legal Authorities
Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).
Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).
State v. Chapple, 135 Ariz. 281, 660 P.2d 1208 (1983).
State v. Lindsey, 149 Ariz. 472, 720 P.2d 73 (1986).
State v. Nordstrom, 200 Ariz. 229, 25 P.3d 717 (2001).
United States v. Houser, 36 M.J. 392 (C.M.A. 1993).
Fed. R. Evid. 611(c).
Fed. R. Evid. 702.
Mil. R. Evid. 611(c).
Mil. R. Evid. 702.
Mil. R. Evid. 1102.
Ariz. R. Evid. 611(c).
Ariz. R. Evid. 702.
Ariz. Sup. Ct. Order R-23-0004 (amending Ariz. R. Evid. 702, effective Jan. 1, 2024).
Related topic papers: The Misinformation Effect · Reconstruction and Recall · Memory for Persons · Improving Eyewitness Memory · The Subadditivity Effect · Confidence and Accuracy
Related topics from Arizona Forensic Services: False Confessions: Suggestibility · Courts-Martial