Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 101 of the 101 references it cites were found in a registry or an index; 31 of the 129 sources its claims cite were checked for support.

Biased and Biasing: The Hidden Bias Cascade and Bias Snowball Effects

Open access
Dror, Itiel E. · 2025 · Behavioral Sciences

Abstract

Cognitive bias is widespread, hidden, and difficult to deal with. It impacts each and every aspect of the justice and legal systems, from the initial engagement of police officers attending the crime scene, through the forensic examination, and all the way to the final outcome of the jurors’ verdict and the judges’ sentencing. It impacts not only the subjective elements in the justice and legal systems but also the more objective scientific elements, such as forensic fingerprinting and DNA. The impact of bias on each of these elements has mainly been researched and considered in silo, neglecting the biasing interactions and how bias cascades and snowballs throughout the justice and legal systems. These should happen rarely, as the Swiss cheese model shows that such errors in the final outcome rarely occur because they require that the shortcomings in each element be coordinated and aligned with the other elements. However, in the justice and legal systems, the different elements are not independent; they are coordinated and mutually support and bias each other, creating and enabling hidden bias cascade and bias snowball effects. Hence, minimizing bias requires not only taking measures to reduce bias in each of the elements but also a wider perspective that addresses bias cascade and bias snowball effects.

Overview

CoChat Summary

TL;DR

The paper argues that cognitive bias in justice and legal systems is widespread, hidden, and usually studied in silos, and that biases interact across elements to create bias cascade and bias snowball effects. It proposes context management, blinding, compartmentalization, linear sequential unmasking, multiple hypotheses, and transparency to minimize these effects.

Key findings

  • Bias affects not only subjective actors but also supposedly objective forensic domains such as fingerprinting, DNA, firearms, and pathology; for example, fingerprint experts are exposed to suspect criminal history in over 40% of formal requests for forensic examination (Gardner et al., 2019).
  • The paper presents eight sources of bias in three categories: case-specific sources (Data, Reference Materials, Contextual Information), person-specific sources (Base Rate, Organizational Factors, Training and Education, Personal Factors), and human nature (Level 8).
  • Bias cascade occurs when one bias propagates to other elements, while bias snowball occurs when additional biases are added and interact in circular loops, amplifying bias and causing double counting of evidence; the Swiss cheese model’s checks and balances fail when elements are not independent.
  • Proposed remedies include blinding to irrelevant information, compartmentalization, linear sequential unmasking (LSU/LSU-E), generating one’s own hypothesis before exposure to others, considering multiple hypotheses, training, and full transparency.

Methods

  • Conceptual/theoretical paper with no empirical study, sample, or dataset; it develops and expands the bias cascade and bias snowball framework from forensic science to the entire justice and legal systems.
  • Uses illustrative examples and cited literature across policing, forensic examination, prosecution, jurors, and judges, including a 2024 Oregon Circuit Court case excluding evidence due to non-blind verification.
  • Main analysis is a narrative synthesis and conceptual taxonomy: eight sources of bias in three categories, the Swiss cheese model, and proposed policy measures.

Limitations

  • The paper states no formal limitations section; it discusses obstacles to addressing bias (bias blind spot, adversarial system, lack of ground truth) but not limitations of its own argument.
  • Evident from the text: no systematic review or empirical test of cascade/snowball effects; examples are illustrative and drawn from prior literature, with no quantitative estimates of prevalence or effect sizes beyond cited figures.
  • Evident from the text: the proposed framework and remedies are conceptual and would require empirical validation and implementation research.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3390/bs15040490

Cites

101 of 101 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 The eight sources that may bias observations, judgements, and decisions throughout the justice and legal systems. They are organized in a taxonomy within three categories: starting at the top with biases emerging from the specific case (Category A); moving down to biases emerging from the specific person handling the case (Category B); and at the very bottom (Category C), biases emerging from human nature that impact all of us, regardless of the specific case or the specific person who is handling it (I. E. Dror, 2020).
Figure 1 The eight sources that may bias observations, judgements, and decisions throughout the justice and legal systems. They are organized in a taxonomy within three categories: starting at the top with biases emerging from the specific case (Category A); moving down to biases emerging from the specific person handling the case (Category B); and at the very bottom (Category C), biases emerging from human nature that impact all of us, regardless of the specific case or the specific person who is handling it (I. E. Dror, 2020).CC BY · PubMed Central
Figure 2 An adaptation of the Swiss cheese model (Reason, 1990). Top panel: although bias exists in the various elements within the justice and legal systems (the black holes in the Figure), it rarely impacts the final outcome because a bias has to be aligned through all the different layers. E.g., ‘A-1’ in the top panel has been able to go through two layers but becomes stuck and is stopped at the third layer; similarly, ‘B-1’ gets stuck and ends at the fourth layer. As the justice and legal systems have many elements before someone is convicted (many checks and balances even before it goes to court), it would rarely happen that a bias could penetrate and align through all the various elements. However, this all hinges on and assumes that the different elements are relatively independent from each other. As illustrated in the bottom panel, the reality is that supposedly independent elements (that should constrain one another) actually work in concert, influencing and biasing each other, creating bias cascade and bias snowball effects (which are not mutually exclusive). ‘A-2’ is able to go through all the layers (in contrast to ‘A-1’) because the bias cascade has moved the hole in layer three to align with the holes in layers 1 and 2. ‘B-2’ illustrates bias snowball in which multiple biases support and feed each other, creating a bigger and bigger bias so powerful that it can penetrate and go through the last layer.
Figure 2 An adaptation of the Swiss cheese model (Reason, 1990). Top panel: although bias exists in the various elements within the justice and legal systems (the black holes in the Figure), it rarely impacts the final outcome because a bias has to be aligned through all the different layers. E.g., ‘A-1’ in the top panel has been able to go through two layers but becomes stuck and is stopped at the third layer; similarly, ‘B-1’ gets stuck and ends at the fourth layer. As the justice and legal systems have many elements before someone is convicted (many checks and balances even before it goes to court), it would rarely happen that a bias could penetrate and align through all the various elements. However, this all hinges on and assumes that the different elements are relatively independent from each other. As illustrated in the bottom panel, the reality is that supposedly independent elements (that should constrain one another) actually work in concert, influencing and biasing each other, creating bias cascade and bias snowball effects (which are not mutually exclusive). ‘A-2’ is able to go through all the layers (in contrast to ‘A-1’) because the bias cascade has moved the hole in layer three to align with the holes in layers 1 and 2. ‘B-2’ illustrates bias snowball in which multiple biases support and feed each other, creating a bigger and bigger bias so powerful that it can penetrate and go through the last layer.CC BY · PubMed Central
Figure 3 Bias Cascade (left panel) occurs when a bias is introduced and it not only contaminates and impacts that specific element (‘X’ biasing ‘A’ in the left panel of the Figure) but then that same bias (‘X’) cascades to also impact other elements (‘B’, ‘C’, and ‘D’, left panel). Bias Snowball (right panel) occurs when biases accumulate and feed one another. The impact increases as the bias gains greater and greater momentum with more and more biases added. In bias snowball, one bias creates additional new biases, which then support the creation of even more additional new biases, and so forth (including interactive loops adding more bias—the dashed arrows). As different elements of the justice and legal systems are being biased, they then bias others (on the right panel, once ‘A’ is contaminated by ‘X’, then ‘B’ is then contaminated by ‘X’ and by ‘A’; by the time it gets to ‘D’, biases ‘A, ‘B’, and ‘C’ have been created in addition to ‘X’). As this interactive contamination continues, the biases grow and grow, creating the Bias Snowball effect.
Figure 3 Bias Cascade (left panel) occurs when a bias is introduced and it not only contaminates and impacts that specific element (‘X’ biasing ‘A’ in the left panel of the Figure) but then that same bias (‘X’) cascades to also impact other elements (‘B’, ‘C’, and ‘D’, left panel). Bias Snowball (right panel) occurs when biases accumulate and feed one another. The impact increases as the bias gains greater and greater momentum with more and more biases added. In bias snowball, one bias creates additional new biases, which then support the creation of even more additional new biases, and so forth (including interactive loops adding more bias—the dashed arrows). As different elements of the justice and legal systems are being biased, they then bias others (on the right panel, once ‘A’ is contaminated by ‘X’, then ‘B’ is then contaminated by ‘X’ and by ‘A’; by the time it gets to ‘D’, biases ‘A, ‘B’, and ‘C’ have been created in addition to ‘X’). As this interactive contamination continues, the biases grow and grow, creating the Bias Snowball effect.CC BY · PubMed Central
Figure 4 An illustration of why sequencing can be important. Reading from left to right, the first/left-most stimulus can affect the interpretation of the middle stimulus, such that it reads as A-B-14; but reading the same stimuli, from right to left, starting with 14 as the first stimulus, often makes people see the stimuli as A-13-14, i.e., the middle stimulus as a ‘13’ (or a ‘B’) depends on what you start with first (I. E. Dror & Kukucka, 2021).
Figure 4 An illustration of why sequencing can be important. Reading from left to right, the first/left-most stimulus can affect the interpretation of the middle stimulus, such that it reads as A-B-14; but reading the same stimuli, from right to left, starting with 14 as the first stimulus, often makes people see the stimuli as A-13-14, i.e., the middle stimulus as a ‘13’ (or a ‘B’) depends on what you start with first (I. E. Dror & Kukucka, 2021).CC BY · PubMed Central

Verification

Claims the paper makes

29 Valid · 1 Not valid · 62 Not checked · 82 Own results and interpretations

86 more claims · 13 with their cited sources checked

Create a free account to see all 174 claims and every source verdict.

See all claims — free account
  • It may state two separate claims as one. 49
  • Probably not a substantial claim. 48

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Self-check results appear here once an independent check has confirmed them.

Cited works that were retracted

Works this paper cites that were later retracted, from the retraction notices publishers deposit. Major when the notice came out before this paper was published.

1 of 1

Citations

Cites

1 retracted

Show all 100 references

Research works printed but not linked: 7

  1. Davis, A. J. (2007). Arbitrary justice: The power of the American prosecutor, Oxford University Press.
  2. Delattre, E. J. (2011). Character and cops: Ethics in policing, AEI Press. [6th ed.].
  3. Ensminger, 2020, The use and abuse of dogs in the witness box, Suffolk Journal of Trial and Appellate Advocacy, 24, 1
  4. Grometstein, 2007, Prosecutorial misconduct and noble cause corruption, Criminal Law Bulletin, 43, 63
  5. Haber, 2021, Racial recognition, Cardozo Law Review, 43, 71
  6. Noble, S. U. (2018). Algorithms of oppression, New York University Press.
  7. Wagenaar, W. A., van Koppen, P. J., and Crombag, H. F. M. (1993). Anchored narratives: The psychology of criminal evidence, St. Martin’s Press.

Cited by

See all 30

Metadata

Authors
Dror, I. E.
Affiliations
Cognitive Consultants International (CCI-HQ), London, UK
Corresponding author
Dror, Itiel E.
Venue
Behavioral Sciences
Publisher
MDPI AG
Published
2025-04-08
Volume
15
Issue
4
Pages
490
ISSN
2076-328X (ISSN-L)
OpenAlex
W4409296383
Cited by
29
References
100
Citations by year
2026: 21 · 2025: 8
FWCI
179.53
Type
journal article
Language
en
Open access
Locations
  • Behavioral Sciences (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Social Sciences › Law › Legal processes and jurisprudence
  • Social Sciences › Decision Sciences › General Decision Sciences › Decision-Making and Behavioral Economics
  • Social Sciences › Psychology › Social Psychology › Deception detection and forensic psychology
Keywords
cognitive biases, hidden bias, criminal justice system, forensic evidence, fingerprint examination, DNA evidence, jury verdicts, sentencing, Swiss cheese model, bias mitigation

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-23, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-23 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Biased and Biasing: The Hidden Bias Cascade and Bias Snowball Effects. https://app.cochat.ai/papers/10.3390/bs15040490
The paper
Dror, I. E. (2025). Biased and Biasing: The Hidden Bias Cascade and Bias Snowball Effects. Behavioral Sciences. https://doi.org/10.3390/bs15040490

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.