Firm Limits on Novel Expert Evidence: Biomechanical Simulations and Probabilistic Credibility Assessments Are Inadmissible Absent a Verifiable Factual Basis and Necessity (Christie v HMA [2024] HCJAC 52)

Court: Appeal Court, High Court of Justiciary (Scotland) — Lady Dorrian (Lord Justice Clerk), Lord Matthews, Lord Beckett

Neutral citation: [2024] HCJAC 52 | Date: 15 November 2024

Introduction

This appeal under section 74(1) of the Criminal Procedure (Scotland) Act 1995 concerns the admissibility of novel expert evidence in a murder trial. The appellant, Dionne Christie, faces trial for the alleged murder of her partner. Central to the appeal was whether the Crown could rely on a report by a consultant ergonomist, Dr David Usher, which deployed a computer simulation rooted in biomechanics and expressed a statistical conclusion that the appellant’s account of the fatal struggle was supported by the biomechanical evidence with a probability of “much less than 0.25%.”

The High Court was asked to determine whether such evidence (i) had an adequate factual foundation; (ii) was necessary to assist a jury on matters beyond ordinary experience; and (iii) respected the jury’s exclusive role in assessing credibility and reliability.

Summary of the Judgment

The Appeal Court allowed the appeal and ruled the expert evidence inadmissible. The core reasons were:

No satisfactory factual basis: The simulation depended on unverified, speculative assumptions (including which hand held the knife, the path and speed of the knife, the “sudden release” of force, and the relative positions of the parties), and even contained a significant arm-length measurement error used in calculations.
Not necessary to assist the jury: The dynamics of a struggle and assessment of the plausibility of accidental self-stabbing lie within the competence of ordinary jurors, particularly given the post‑mortem evidence available.
Usurpation of the jury’s function: Framing the conclusion as a percentage probability that the appellant’s account could be true impermissibly intruded into the jury’s exclusive province to assess credibility and reliability.
Risk of confusion through statistics: The probabilistic expression (0.25%) lacked transparent linkage to the simulations and posed a substantial risk of confusing rather than assisting the jury. The witness was not a statistician.
Misplaced reliance on civil authorities and analogies: The trial judge’s analogy with road traffic reconstructions was inapt because those reconstructions typically rest on objectively verifiable measurements; and the reliance on Kennedy v Cordia Services LLP overlooked the stricter “necessity” requirement in criminal cases.

Factual Background

The deceased died from a stab wound through the sternum into the heart. Only the deceased and the appellant were present. In two police statements, the appellant recounted that the deceased, during a struggle, restrained her from behind, held a knife and sheath, and threatened to cut the baby from her (she was pregnant). She freed herself and afterwards observed blood on the deceased; she did not know how the knife penetrated his chest. Post‑mortem evidence suggested at least a moderate degree of force would be required to produce the wound, which pierced the sternum, rib cartilage, pericardium and right ventricle.

Dr Usher’s report purported to model the biomechanics of “what would have happened” when the appellant escaped the hold, hypothesising that the force the deceased used to restrain her would convert into a rapid, uncontrolled arm movement causing self‑stabbing. He ran thousands of simulations across a wide range of variables and concluded that the appellant’s account was supported by the biomechanical evidence with a probability of less than 0.25%.

Analysis

Precedents Cited and Their Influence

Gage v HM Advocate 2012 SCCR 161 (per LJG Gill at para 22): Reaffirmed that expert evidence in criminal cases is admissible only where necessary to assist the jury on matters outside ordinary knowledge. The Court applied this “necessity” threshold to reject the simulation: the dynamics of a physical struggle and the plausibility of accidental injury are within jurors’ common experience, especially with available post‑mortem context.
Wilson v HM Advocate 2009 JC 336 at para 58: Emphasises limiting expert testimony to areas requiring expertise, not common sense. This case reinforced the gatekeeping role in excluding unnecessary expert pronouncements on ordinary human behaviour.
Snowden v HM Advocate 2014 SCCR 663 at para 67: Stresses that lay experience suffices for interpreting human interactions in a fight. This underpinned the Court’s view that the jury needed no biomechanics to evaluate the appellant’s account.
R v Turner [1975] QB 834 at 841: A classic authority that expert evidence is not admissible on matters within the layman’s knowledge; used to reinforce the proposition that experts should not opine on ordinary human conduct or credibility.
Mitchell v HM Advocate 2018 JC 67: Cited by the Crown to argue that experts can assist juries in drawing inferences from primary evidence they could not otherwise draw. The Court distinguished this, holding that the case did not justify importing expert biomechanics where the fundamental factual substrate was uncertain and the subject matter ordinary.
Kennedy v Cordia Services LLP [2016] UKSC 6: A civil case widely cited for expert evidence principles. The Court cautioned that Kennedy acknowledged a stricter necessity test in criminal cases, which was not within the Supreme Court’s jurisdiction there (para 37). The trial judge erred by focusing on biomechanics as a recognised discipline without ensuring a proper factual foundation or assessing criminal-law necessity.
Chalmers v HM Advocate 1954 SLT 177: Referenced in the context of cautions and admissibility of statements. Although the second statement might have raised caution issues, the defence did not object and the point was not determinative. Its relevance lay in highlighting that the expert selected one statement’s narrative without appreciating discrepancies that mattered to the modelling.
Dickson, Treatise on the Law of Evidence in Scotland (3rd Hamilton Grierson ed, 1887), vol 1, para 397: A classic source reinforcing the limited, necessity-based role of expert evidence in criminal jury trials.

The Court’s Legal Reasoning

1) No objectively verifiable factual foundation

The Court found a striking absence of stable, verified facts necessary to support any meaningful biomechanical reconstruction:

Unknown critical variables: Which hand held the knife; grip type; arm trajectory; angle, speed and direction of approach; the nature and constancy of force used; the relative positions of the parties at release; whether force was “suddenly” released. The expert openly conceded “We don’t know how he was holding her … there is uncertainty about how the movement starts.”
Assumptions over facts: The model proceeded on a “generalist assumption” of instantaneous escape producing an uncontrolled arm movement; assumed the knife was in the right hand; and assumed a curved trajectory into the chest. None of these were established.
Hearsay foundation for pivotal detail: The key scenario (escape under the right arm) came from a verbal report of a police demonstration the expert never witnessed and lacked particulars to translate into model parameters.
Measurement error: A plainly improbable arm-length value (713mm elbow-to-grip) was used in formulae, undermining confidence in calculations.

Result: without an “underlying accepted or established factual basis,” any simulation was “entirely speculative and theoretical,” lacking evidential value. The Court rejected the trial judge’s analogy with road traffic reconstructions, which typically rest on objective measures (skid marks, projection distances, vehicle damage, final positions) and well‑researched norms.

2) Necessity: the jury did not need expert biomechanics to decide this case

Scots criminal law requires expert evidence to be “necessary” — i.e., on subjects beyond jurors’ ordinary knowledge. The Court held that assessing the plausibility of accidental self‑stabbing in the course of a struggle, in light of the post‑mortem’s description of the wound path and degree of force likely required, is a task within ordinary experience. The case was “not particularly complicated or sophisticated” in a way that demanded specialist assistance. The jury can evaluate the account alongside the medical findings and other evidence.

3) Usurpation of the jury’s function — probabilistic credibility

By expressing that the appellant’s account is supported by the biomechanical evidence with probability “much less than 0.25%,” the witness effectively quantified the truthfulness of the account. That invades the jury’s exclusive role to determine credibility and reliability. The Court was clear that expert testimony should not translate disputed narratives into numerical probabilities of truth, a move that implicitly converts juror judgment into statistical thresholds.

4) The dangers of statistics in the criminal trial

The expert was neither a statistician nor a probability expert. The relationship between the numerous simulations and the precise 0.25% figure was not explained, and the Court highlighted the notorious difficulty of communicating probability in legal proceedings, referencing the Royal Society of Edinburgh/Royal Society publication, “The Use of Statistics in Legal Proceedings” (p 25). Rather than aiding, such quantification risked confusing the jury and potentially necessitating dueling statistical experts — distracting from the central question.

5) Misapplication of civil authority (Kennedy v Cordia) and an inapt reconstruction analogy

Kennedy v Cordia: While recognising biomechanics as a legitimate discipline could be relevant, the criminal test requires necessity and a secure factual underpinning. The trial judge focused on the discipline’s status but did not ensure the essential evidential groundwork or necessity. That was a misdirection.
Road traffic analogy rejected: Without objective, measurable inputs, the comparison to collision reconstruction fails. Those reconstructions combine stable datapoints with validated physical models; here, core inputs were speculative.

Impact and Future Significance

Key principles established or reinforced

Novel simulations must be built on verifiable facts: Biomechanical or digital modelling cannot stand on speculative, assumed, or hearsay premises. Where critical inputs are unknown, the product is inadmissible.
Necessity remains a hard gate in criminal cases: If the subject is within ordinary juror experience, expert testimony is unnecessary and inadmissible. The dynamics of a struggle and assessment of accidental injury plausibility generally fall within lay competence, particularly where medical/pathological evidence is available.
Experts cannot quantify credibility: Expressing a statistical probability that an accused’s account is true (or “supported”) impermissibly usurps the jury’s function and risks framing “reasonable doubt” as a number.
Beware statistical expressions without statistical expertise: Probabilistic conclusions must be methodologically transparent, statistically sound, and truly assistive. Otherwise, they confuse and mislead.
Civil-law frameworks do not dilute criminal-law safeguards: Kennedy’s general principles about expert disciplines do not displace the stringent necessity and factual-basis requirements in criminal trials.

Practical consequences for stakeholders

Prosecutors: Before instructing biomechanical or similar simulations, ensure a solid bedrock of objectively verified facts (positions, trajectories, timings, measurements) and confirm the expert’s remit does not drift into credibility or probability of truth. Anticipate challenges under Gage and Wilson if the subject is within juror competence.
Defence: Challenge speculative modelling at the gatekeeping stage, focusing on factual uncertainties, necessity, and any probabilistic framing of credibility. Section 74 appeals remain a viable route to exclude such evidence before trial.
Experts: Stay within your core discipline. Avoid probabilistic expressions of a party’s truthfulness. Be explicit about assumptions, demonstrate how each is grounded in evidence, and refrain from opining where foundational facts are not established.
Judges: Apply a rigorous gatekeeping function. Ask: (i) Is there an objective factual substrate? (ii) Is expert input necessary for the jury? (iii) Does the testimony risk usurping the jury’s role, especially through probabilistic framing?

Complex Concepts Simplified

Ergonomics vs Biomechanics: Ergonomics focuses on human interaction with systems and environments, often in occupational safety. Biomechanics applies mechanical principles to human movement and forces. While overlapping, expertise in one does not automatically confer authority in the other or in statistics.
Necessity (criminal expert evidence): Evidence from experts is admitted only when it addresses matters beyond ordinary juror understanding. If lay jurors can decide the issue using common sense and ordinary experience (aided by, say, medical evidence), expert testimony is unnecessary and inadmissible.
Usurping the jury’s function: Experts cannot opine on the ultimate question of credibility. Expressing a percentage likelihood that an account is true effectively tells the jury whom to believe, which is impermissible.
Objective factual foundation: Expert models must be grounded in verified facts (measurements, observations, data). Models built on untested assumptions, hearsay, or unknown parameters lack admissible worth.
Statistical probability in trials: Numerical probabilities (e.g., 0.25%) risk misunderstanding (e.g., the “prosecutor’s fallacy”) and can misframe reasonable doubt. Unless the statistic is robust, transparent, and necessary — and the witness is properly qualified — it can mislead more than assist.
Section 74(1) appeals: A procedural route to appeal certain preliminary rulings (such as admissibility of evidence) before trial, avoiding prejudice from inadmissible material reaching the jury.

How the Court Dealt with the Appellant’s Statements

The Court noted variations between the appellant’s first and second statements and observed that the expert’s modelling relied on only the second, potentially raising caution issues. Although the defence did not seek to exclude the second statement, the Court highlighted that selective reliance obscured discrepancies relevant to any modelling (e.g., relative positions, mechanics of release). The point underlined the broader theme: the model had no stable factual platform.

Why the Road Traffic Reconstruction Analogy Failed

The trial judge’s comparison to collision reconstruction was decisively rejected. Traffic reconstructions typically involve:

Measured skid marks, distances, angles and final rest positions;
Documented vehicle damage characteristics and pedestrian projection distances;
Validation against established physical laws with known inputs.

By contrast, in Christie the essential inputs were unknown or assumed (grip, trajectory, hand dominance during the act, force dynamics, release mechanism). Without objective inputs, a “simulation” becomes conjecture, not assistance.

Key Takeaways

Expert simulations in criminal trials must be anchored in verifiable facts; speculative modelling is inadmissible.
Expert evidence is unnecessary — and therefore inadmissible — where jurors can decide using common experience, particularly about human interactions in a struggle.
Experts must not present percentage probabilities that an accused’s account is true. That invades the jury’s core role.
Statistics in the courtroom require extreme caution: transparent methods, proper expertise, and a genuine prospect of assisting the jury are essential.
Judges must apply criminal-law gatekeeping rigor, not import civil-law reasoning to dilute necessity and factual-basis requirements.

Conclusion

Christie v HMA sets a clear boundary in Scots criminal law for novel biomechanical and computer-simulation evidence. The Appeal Court emphatically held that without a secure factual foundation and demonstrable necessity, such evidence is inadmissible — doubly so where it purports to express probabilistic judgments about the truthfulness of an accused’s account. The judgment is a strong reaffirmation of the jury’s central role in assessing credibility and of the court’s gatekeeping responsibility to exclude expert evidence that is speculative, unnecessary, or confusing.

Going forward, parties seeking to deploy biomechanical or similar digital reconstructions must ensure: (i) the model is rooted in objective, measured facts; (ii) the expertise squarely matches the tasks undertaken; (iii) the testimony helps, rather than supplants, the jury; and (iv) the presentation avoids statistical quantifications of credibility. Christie will likely be a touchstone for excluding overreaching simulations and probabilistic “truth” assessments in criminal trials.