The Cross-Examiner's Playbook

Almost everything a forensic examiner reads was written for examiners. This wasn't. How to Cross-Examine Forensic Scientists is a real published guide, put together in 2014 by a team of scientists and lawyers led by Gary Edmond and Kristy Martire, and its job is to teach defence lawyers how to take forensic evidence apart on the stand. It even hands them the questions, word for word.

Which is the exact reason you should read it before they do. Every line of attack in it is one you should expect to face. The examiner who has already worked through each question, and answered it straight, in private, is the one it can't catch off guard.

There's no trick to this. The people who wrote the guide spend their careers studying what makes examiners good, and this is the other half of that work: their attack manual, turned around, so you walk in ready for it.

A good cross-examiner starts by taking your credentials off the table, politely. They'll grant that you're qualified, experienced, well thought of. Then they'll note that none of it is evidence your method works.

The whole guide turns on one hard distinction: training, study and experience on one side, demonstrated ability on the other. Your years on the job, the wall of certificates, the fact that courts have waved this kind of evidence through for decades. All of it tells a jury you're credible. None of it shows you can do this exact task accurately. Those are two different claims, and only the second one is worth anything in court.

The classic failure is the expert who's skilled at one thing and slides, without noticing, into another. The podiatrist who can assess a foot, now naming a stranger from grainy CCTV of their walk. The analyst whose validation study was really about a neighbouring problem, not this one. Come ready to show you can do the precise thing your opinion rests on, not something standing beside it.

So they ask the obvious next question. Where's the study? Not a textbook. Not your training. A validation study that tested the technique in conditions where someone independently knew the right answer. That last part is the whole game. Validation only counts when it's measured against ground truth.

This is where a lot of confident-sounding proof falls apart, because so much looks like validation and isn't. The age of your field isn't validation. Your years of casework aren't. Prior admission by the courts isn't. A conviction isn't, even one upheld on appeal. “I held up fine under cross-examination last time” isn't. And a private study you ran but never published certainly isn't.

Even a properly validated technique, though, only reaches as far as the conditions it was tested in. The moment your casework drifts outside them, the validation can stop applying, and you have to say so out loud. A poorer sample, a harder comparison, a different population: any of them can push you past what the study covered.

Run that test on yourself before they do. For your own discipline, can you name the validation studies, the real ones, measured against known answers, and say where your casework sits inside them or outside them? If you can, almost nothing in this section can lay a glove on you. If you can't, it's far better to learn that here than on the stand.

Every method makes mistakes. The guide says so flatly: all techniques have limits, and anything that runs through human hands is error-prone. Then it presses on the parts examiners least like to say out loud.

Start with your error rate. If a study has measured how often examiners using your technique get it wrong, you should know that number cold and be willing to say it. Then your proficiency tests. Passing them feels like proof. But the guide hands lawyers an awkward fact: many proficiency tests are notoriously easy, they pit practitioners only against each other and never against ordinary people, and they may not measure real-world skill at all. Passing an easy test isn't the same as being demonstrably accurate.

Then the question they save for last: the details that didn't fit. In almost every comparison something fails to line up — a feature in the mark with no counterpart in the reference, a measurement sitting outside the expected range. If you resolved each one in favour of your conclusion, called it distortion or artefact and moved on without writing down why, that is exactly where the questions will go: you found a difference — what entitled you to explain it away, rather than treat it as evidence of a different source? And keep three words apart in your head. Proficiency, validity, and reliability are not the same thing. You can be slick at a method that's neither valid nor reliable.

Sooner or later the questioning turns to bias. The cross-examiner asks what you knew about the suspect before you reached your conclusion, and whether that knowledge could have bent it.

The answer the guide expects is yes. It could have, and you might never have felt it happen. Domain-irrelevant information, a confession, a prior record, an investigator who's already certain, can shift and even flip an analyst's conclusion. It works unconsciously, on sincere and skilled people. And once you've been exposed to it, there's usually no way to scrub it back out of your opinion. Brandon Mayfield and Shirley McKie are the standing proof: skilled examiners, full institutional review, confidently wrong. (Mayfield, an Oregon lawyer, was falsely matched to the 2004 Madrid train bombing; McKie, a Scottish police officer, was wrongly said to have left a print at a murder scene.)

Because you can't undo the exposure, the only defence that holds is structural. Control what you're told. Reach your conclusion before the case story reaches you. Verify in a way where the checker doesn't know your answer. Keep a record of what you knew, and when. “I'm too experienced to be biased” is the exact line the guide is praying you'll reach for, and it isn't an answer.

Eventually they reach your words, because your words are the softest target in the room. You wrote “match.” You told the jury the print was the defendant's. Now counsel asks you to explain exactly how you settled on that word. Did it come from a scale, a calculation, a validation study? Or from habit?

The guide red-flags a whole vocabulary as overstatement: match, positive identification, individualisation, to the exclusion of all others, and the lawyerly favourites practical and reasonable certainty. Every one of them smuggles in a confidence the science can't back. Here it goes further than the NAS report. Logically, it argues, a forensic scientist cannot positively identify a single person or object from a trace, full stop. The strongest thing you can defend is tied to what your validation supports and set against the alternative: how much more likely is this evidence if it came from the defendant than if it came from someone else?

So carry a reason for every word. If you can say why you chose what you chose, and keep it inside the limits of your method, the question built to corner you turns into the moment you look most credible.

The last move is often the most uncomfortable. Counsel picks up the NAS report, or the 2012 NIST/NIJ human-factors review, or the Scottish Fingerprint Inquiry, and reads your own field's criticisms back to you. You're familiar with this report, aren't you? You knew it said this about your method? And yet you didn't mention it anywhere in yours, did you?

The guide notes that some analysts meet these reports with open hostility, and that hostility plays terribly in front of a jury. It also points out something examiners forget: your codes of conduct require you to disclose serious disagreement or controversy about your methods even when nobody asks. Silence isn't neutral. Leaving out the known weaknesses of your own discipline can be turned into a suggestion that you're not the impartial witness you're meant to be.

And don't count on the trial to fix it for you. The guide's authors are blunt that the usual trial safeguards haven't been much good at catching forensic overreach. The protection has to come from you. Know the criticisms of your own field, put them in your report before anyone asks, and meet them on the stand with candour instead of a raised guard.

Do all of this and you're the hardest kind of witness to lay a glove on. The cross-examiner in this guide is hoping for the opposite: someone who overstates, props themselves up on the CV, bristles at the mention of bias, and clams up about the reports. Be the other one. The examiner who already knew every question was coming, and brought the answers in with them.