Because some of the therapists participated in the study with more than one child, the data of the children were not completely independent. For this reason, we initially assessed all variables for the effects of the therapist using a repeated-measures analysis of variance.