Experts in court

Core idea

The forensic psychologist in the witness box looks, to a jury, like a scientist delivering verdicts about another human mind. The reality is narrower and more disciplined. An expert witness is permitted to translate a defined body of empirical knowledge into terms that help a court answer a legal question — and nothing more. The expert is not there to decide guilt, predict crime, or vouch for a defendant's character. The job is to tell the court what psychological research, properly applied, can and cannot say about a specific issue in the case.

Two pressures constantly distort this role. The first is legal gatekeeping: judges decide what counts as admissible expert evidence using standards (Frye in some jurisdictions, Daubert in most US federal and state courts) that test whether the underlying method is reliable enough to put in front of a jury. The second is the adversarial system itself: each side hires the expert, pays the expert, and rewards the expert who is most persuasive — not the one who is most circumspect. The legitimate work of the forensic expert sits in the narrow band that survives both pressures.

Author's argument: The forensic expert's authority is borrowed from the discipline, not owned by the individual. The moment testimony exceeds what the underlying research actually supports, it becomes advocacy dressed as science.

Why it matters

Verdicts hinge on expert testimony far more often than the public realizes. Whether a confession was coerced, whether an eyewitness identification can be trusted, whether a defendant could form intent at the time of the offence, whether they are competent to stand trial today — none of these questions can be answered by lay common sense alone, and all of them can be answered badly by an overconfident expert. The cost of bad expert evidence is asymmetric: a jury that believes a charismatic witness will convict on a story the science does not actually support, and the appeal courts only sometimes catch it.

Why the public overestimates what experts can do

Television has trained jurors to expect that a forensic psychologist can read a defendant the way a pathologist reads a wound. Real research findings are far softer: probabilistic, sample-dependent, and laden with effect sizes that translate poorly to "did this person, on this night, do this thing?" When a real expert speaks cautiously, juries can read that caution as weakness. The structural temptation is to harden the answer to match the audience's expectation.

Why adversarial systems amplify the distortion

In an inquisitorial system, the court appoints the expert and pays the expert; the expert reports to the court. In an adversarial system, the prosecution and defence each retain their own expert, brief their own expert, and present their own expert in chief before the other side cross-examines. The selection itself is filtering: lawyers retain experts who are willing to support their theory and quietly drop those who are not. Over a career, an expert who too often says "the evidence does not let me answer that" stops being briefed.

Key takeaways

Mental model

Copy sourceZoom

Practical application

What experts can legitimately offer

What experts should refuse to offer

The same training that licenses an expert to testify in one domain disqualifies them from many others. A psychologist asked to "say whether the defendant did it" is being asked the ultimate question, which is reserved for the jury. A psychologist asked to "predict whether this person will reoffend" must answer in probabilities anchored to a validated risk instrument and explicitly state the confidence interval — anything more specific is fiction. A psychologist asked whether a victim is "telling the truth" is being asked a question psychology cannot answer; credibility assessment is the jury's job.

Frye versus Daubert in one breath

Frye asks the field: is this method generally accepted? Daubert asks the judge: is this method scientifically reliable, judged by testability, peer review, known error rate, controlling standards, and acceptance? Frye is easier for judges (delegate to community consensus) but slow to admit new science. Daubert is harder for judges (they must understand the method) but more responsive to genuine evidence. Both standards exist precisely because courts have learned that confident-sounding expertise is not the same as reliable expertise.

The adversarial pressure to over-claim

A working rule for honest testimony

The clearest discipline is to separate three sentences and never let them merge:

1. What I observed in this individual. Test scores, behaviour during interview, records reviewed.
2. What the literature says about people like this. Cited studies, sample sizes, effect sizes, base rates.
3. The inference I draw from (1) and (2). Where the inference relies on assumption rather than evidence, say so.

Cross-examination will probe each layer separately. An expert whose three sentences are clearly distinguished can be challenged on any layer without the others collapsing.

Example

A 19-year-old with documented mild intellectual disability is convicted of burglary largely on the strength of a written confession produced after eleven hours of police interview without a lawyer present. The defence retains a forensic psychologist to evaluate the reliability of the confession. The temptation — and the lawyer's hope — is for the expert to say: the confession is false.

The legitimate version of the report instead does three things. It documents the defendant's measured intellectual functioning and scores on the Gudjonsson Suggestibility Scales, both of which show high compliance and suggestibility. It summarises the published research on interrogation-induced false confessions, including the elevated risk among suspects with intellectual disability under prolonged questioning, citing effect sizes and known case counts. It then states the inference carefully: the conditions of this interrogation, combined with this defendant's measured vulnerability profile, place this confession in a category for which the false-confession rate is materially elevated. The report stops there.

In cross-examination, the prosecution will push for the further step — "so you are saying the confession is false?" The disciplined answer is no — I am saying the confession was produced under conditions psychology has shown to produce unreliable confessions at meaningfully higher rates, and that fact bears on how much weight the jury should give it. The jury decides whether the confession is true. The expert decides whether the conditions under which it was produced are the kind that psychology can speak to. That distinction is the whole job.

A second expert, retained earlier in the case and quietly dropped, had been willing to testify that the confession was "almost certainly" false. The defence team replaced them not because the conclusion was wrong but because no published study supports a single-case probability statement of that kind, and any half-competent cross-examination would have destroyed the expert's credibility — taking the defence's whole theory down with it. The disciplined report is harder to deliver and harder for the jury to summarise, but it is the one that survives appeal.

Related lessons