Since young children have little experience in answering test questions, misunderstandings easily arise when they are subjected to tests. The experimenter wants to test the children, whereas they expect the experimenter to help them solve the problem. In an experiment the expectations of children about the interaction with the experimenter were influenced in a pre-task. In condition Exp:A (Expectation: Assistance) it was suggested that the experimenter would help the child. In condition Exp.T (Expectation: 'lest) children were made to expect that the experimenter would give no assistance. A third condition was a standard condition (8) without a pre-task. The experimental task was a conservation of number problem. The subjects in condition Exp/I' performed significantly better than the children in the other conditions. It is concluded that unfamiliarity with the rules of testing contributes to children's poor results in this kind of experiment. Making explicit what is intended leads to better performances.