US Expert Testimony: Rule 702 and Forensic Evidence

• Separate an expert’s qualifications from the reliability of a particular opinion.
• Summarise the current federal Rule 702 requirements without treating them as universal state law.
• Recognise overstatement, missing assumptions and unsupported probabilities.

Under the current federal rule, the party offering the expert must establish the specified conditions on a more-likely-than-not basis. The testimony must assist the fact-finder, rest on adequate facts or data and reliable methods, and reflect a reliable application of those methods to the case. The rule’s 2023 amendment made the burden and application requirement explicit.

This threshold concerns admission of the expert evidence. It is not the criminal standard of proof for guilt and is not an instruction to give every scientific conclusion a numerical probability. The court assesses the offered opinion; a witness’s confidence is not a substitute for its foundation.

Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), explains federal reliability gatekeeping under Rule 702. Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999), applies that gatekeeping responsibility beyond scientific testimony to technical and other specialised knowledge. The inquiry depends on the proposed testimony; it is not a mechanical checklist that every method passes identically.

Do not call an instrument, laboratory or training course “Daubert certified.” Courts decide admissibility in cases. State adoption and application of expert-evidence standards vary, so a statement about a federal decision needs an explicit federal context.

A report should let another appropriately qualified person follow the work. Distinguish the material received, what was examined, the measurements or observations made and the reasoning connecting them to a conclusion. Identify assumptions supplied by others and explain how the opinion changes if an important assumption is wrong.

For a comparison, state what characteristics were assessed, the quality and limitations of the material, the decision framework used and the significance the method can support. A similarity is an observation; an inference about source or activity needs further reasoning. An analyst should not leap from a shared feature to a claim that only one person or object in the world could have produced it.

Report inconclusive results as inconclusive. Do not convert an instrument’s detection threshold into a universal exclusion, or treat a laboratory’s accreditation as proof that every opinion it produces is reliable.

An error rate, a measurement uncertainty and the probability of a hypothesis are different quantities. Describe which quantity is supported, how it was obtained and whether it applies to the actual conditions. Avoid phrases such as “100% accurate,” “zero error” or “scientific certainty” where the method cannot justify them.

A numerical result is not automatically the probability that the accused committed the offence. For example, a statistic describing the occurrence of a DNA profile under stated assumptions does not independently give the probability of guilt. The report must identify the propositions and assumptions used rather than invite a reversal of the conditional probability.

Authentication, hearsay, disclosure obligations and applicable confrontation rights may raise separate issues. Passing a Rule 702 inquiry does not resolve every question about a report or exhibit. Nor does admission require the fact-finder to accept the opinion as conclusive.

Experts should preserve records, identify contrary findings and correct material errors through the appropriate documented process. The scientific account should remain the same whether the result assists the prosecution, the defence or neither party.

US Forensic Laboratories and Evidence Submission. UK Expert Evidence and Forensic Quality Standards.

• Use the rule applicable to the actual court.
• Relevant expertise, a reliable method and supported case application all matter.
• Show limitations and avoid probabilities the data do not establish.
• An admitted opinion is not a scientific guarantee or a verdict.