"Can You Be Fair?" Is Not a Question: Designing Voir Dire

Lawnova Editorial • • 4 min read

“Can you be fair?” is not a question

Voir dire is the only part of a trial where the lawyer’s job is to listen rather than persuade, and it is routinely spent doing the opposite. The most common failure is not a bad strike. It is a question set that could not have produced useful information no matter who answered it.

This is about question design — the part that is entirely within your control, costs nothing, and is done badly far more often than it is done well.

Why the standard questions fail

They invite the socially correct answer. “Can you set aside what you have heard and decide this case on the evidence?” has one acceptable response, and everyone in the room knows it. The answer tells you nothing about the person and something about their willingness to give expected answers.

They ask for self-assessment of bias. People are poor witnesses to their own prejudices. A juror who genuinely believes they can be impartial may not be, and one who worries aloud that they cannot may be your best juror.

They are closed. A question answerable with yes or no produces a syllable. Panel members who say more reveal more, and the objective is to get people talking.

They are asked of the group. “Does anyone here have experience with…” produces silence and a few raised hands from the most confident people present.

What works instead

Ask about experience, not attitude. Not “do you trust police officers” but “tell me about any dealings you or your family have had with police.” Experience is concrete, memorable, reportable — and far more predictive than a stated attitude.

Ask about other people. “How do you think most people your age feel about…” lets someone express a view without owning it. What comes back is usually their own.

Ask about the thing you are worried about, directly and without apology. If the weakness of your case is that your client did not call the police for two days, ask the panel what would make someone delay. You will learn who has an answer, and you will have raised it before the other side does.

Ask follow-ups. The first answer is the prepared one. The second is usually more honest. “What makes you say that?” is the most productive four words in voir dire.

Ask individuals, by name. Group questions produce group silence.

A structure that fits the time you get

Voir dire is often short, so decide in advance what you must learn:

  1. Two or three attitudes that could decide this case. Not general fairness — the specific dispositions that map onto your specific facts.
  2. One question per attitude that asks about experience rather than belief.
  3. A follow-up ready for each.
  4. The hardship and disqualification questions, dispatched briskly to preserve time for the rest.
  5. A note-taking scheme you can actually use under pressure — a grid, a code, something recorded consistently for every panel member rather than dense notes on the three who spoke most.

That last point is where technology earns its place, and it is unglamorous. A structured, searchable record of what each panel member actually said beats anybody’s memory an hour later, and it is the thing you will want when you are deciding strikes with limited time.

Where technology helps, and the line it must not cross

Helps: capturing responses consistently; organising public-record information you are entitled to consider; surfacing patterns across a panel; keeping a record of the reason for each strike, contemporaneously.

That last one is worth doing regardless of whether anyone challenges you, because a reason written at the time is worth far more than a reason reconstructed later.

The line: if a tool’s output could shape a peremptory challenge, you are responsible for understanding its methodology well enough to know whether following it would produce an unlawful strike. That is not a policy preference — it is the standard ABA Formal Opinion 517 sets, and the analysis is the same whether the recommendation comes from software, a consultant, or the client. We set that out in what Opinion 517 requires of AI jury tools.

The practical implication is narrow and important: a ranking with no explanation cannot be inquired into, and the duty of inquiry does not disappear because the recommendation arrived as a number.

The part that does not change

The most valuable thing in voir dire remains a lawyer paying attention — to the pause before an answer, to the person who agrees a beat too quickly, to the juror who looks at another when a topic comes up. None of that is in a transcript, and none of it is in a model.

Good questions create the conditions for that observation. That is why question design, not analysis, is where the preparation time should go.

Where VerdictPilot fits

VerdictPilot is built for the preparation half: testing which attitudes actually matter to your case before you write the questions, and rehearsing how the case reads to someone encountering it cold. What happens in the room is yours.

For the wider view on what consultants contribute, see what a jury consultant actually does.

← All Articles
Lawnova Academy — AI certificates for legal professionals · Levels I–IV · 100% online Explore the Academy →