Data Analysis
The hypotheses with respect to the outcome variables were analyzed using a one-tailed repeated-measures analysis of variance (group [2] 3 time [4]). The probability of Type I error (alpha) was determined as .05. 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.