Most American courts determine a child's competence to testify partly by asking some variation of this question: does the child understand the moral obligation to tell the truth when testifying? Children found competent to testify are then required, like all witnesses, to promise to tell the truth. A group of psychologists conducted experiments to see whether either of these steps increased the likelihood that children would tell the truth. Their results are published in Talwar, Lee, Bala, and Lindsay, "Children's Conceptual Knowledge of Lying and its Relation to Their Actual Behaviors: Implications for Court Competence Examinations," Law and Human Behavior, Vol. 26, No. 4, pp. 395-415 (August 2002).

Participants in the experiments ranged from three to seven years old. Each child underwent the experimental procedures individually.

In the first experiment, a researcher would tell a child about Katy, whose teacher told her not to eat candy. The teacher then left Katy alone with the candy, and she ate it. When the teacher returned, she said, "Katy, did you eat the candy?" Each child in the experiment was then asked how Katy should answer. The child was then told that Katy had denied eating the candy, and the child was asked, "Is what Katy said a lie or the truth?"; "Is what Katy said good or bad?"; "Why is it (good/bad)?" Each child was also asked similar questions about a hypothetical situation in which he or she had broken a glass and then was asked by a parent whether he or she had done so. (Asking about such scenarios has been recommended by other psychologists as an improvement over the more abstract questions traditionally used in determining a child's competence to testify. Lyon and Saywitz, "Young Maltreated Children's Competence to Take the Oath," Applied Developmental Science, 3, 16-27 (1999).)

Each child also played a game with the experimenter. In the game the child was to turn his or her back and then guess the identity of an unseen toy. In the midst of the game, the experimenter feigned a need to leave the room and asked the child to promise not to peek at the toy. A hidden camera recorded whether the child did peek. (Most did.) The experimenter then returned and asked whether the child had peeked.

Whether children falsely denied peeking correlated with age. Only 37% of three-year-old peekers lied about it, but 86% of all older peekers did. There was no indication in the results, though, that children who understand lying and its immorality were less likely to do it. In fact, a reverse correlation appeared, because the three-year-olds, who had the weakest grasp of the concept of lying, were the least likely to lie. (See the May 2003 Child Abuse Law story Children’s Willingness to Deceive Fluctuates with Age, Studies Show for its explanation that children so young lack a "theory of mind.")

The second experiment was similar except that each child was asked to promise to tell the truth before being asked whether he or she had peeked. The proportion of peekers who falsely denied peeking was 57%, significantly less than the 80% who falsely denied peeking in the first experiment.

The third experiment presented the same temptation to peek to two groups of children. With each child in one group, the experimenter discussed Katy and lying but did not ask the child to promise to tell the truth. With each child in the other group, the experimenter asked the child to promise to tell the truth but did not discuss the concept of lying or its immorality. The promise was significantly more effective than the discussion of lying in reducing lying. Of peekers who promised to tell the truth, 59% lied. Of peekers who were not asked to promise, 75% lied.

While promising to tell the truth reduced lying, more than half the peekers lied even then. This study provides no information on whether a similar proportion of adults would lie under similar circumstances.