Questioning a child, what the question does to the answer

In 1900 Alfred Binet published a book entitled La Suggestibilité, in which he showed that schoolchildren changed their answers depending on the form of the questions asked, notably when these presupposed a detail absent from what they had seen.

The subject long remained secondary. It returned to the fore from the late 1980s, when American court cases resting on repeated interviews with young children raised doubts about the way the testimony had been gathered.

In 1993 Stephen Ceci and Maggie Bruck published a historical and critical review of the available research. It concludes that preschool children are on average more susceptible to suggestion than older children and adults, without the latter being immune.

The review stresses an often forgotten point: young children questioned in a neutral way produce short but generally accurate accounts. The fragility lies less in their memory than in the conditions of the interview.

In 1995 Michelle Leichtman and Stephen Ceci published an experiment conducted in preschools. A visitor briefly comes into the classroom without damaging anything. Depending on the condition, the children then hear talk of his clumsiness and are questioned several times with leading questions.

When the two procedures are combined, a substantial share of the children end up reporting that the visitor damaged objects, sometimes with invented details. Without them, the accounts remain overwhelmingly faithful to what happened.

Several mechanisms combine. A child interprets a repeated question as a sign that their first answer was not the right one. They will more readily answer a closed question than say they do not know the answer. They also distinguish less well between what they experienced and what they heard others recount.

These mechanisms do not concern testimony alone. They apply to any research questionnaire or interview in which an adult questions a young child about their experiences.

Age is not the only factor. Studies also find differences linked to the status of the interviewer, an adult perceived as an authority more easily winning acceptance of their suggestions, and to the time elapsed since the event, which weakens the original memory.

Structured interview protocols, including the one developed by Michael Lamb and his colleagues under the aegis of an American public research institute, rest on these results: open invitations to free recall, with closed questions pushed back to the end of the interview.

Evaluations of these protocols mainly compare the amount of information obtained through open questions. Its accuracy is harder to check in the field, since what actually happened is not known.

Methodological recommendations for questionnaire surveys of children point in the same direction: wordings that do not presuppose an answer, an explicit option to say one does not know, and attention to the number of consecutive closed questions.