ABSTRACT
BACKGROUND: There is a need for appropriate tests to identify developmental delays early in life. OBJECTIVE: This study investigates the Test of Infant Motor Performance (TIMP) unidimensional model and its ability to explain variance in items response, hierarchic order of the items, levels of difficulty, and the extent to which the items were appropriate to distinguish different levels of function for infants in Brazil. METHODS: Six hundred fifty-five infants were assessed. Families provided sociodemographic information. RESULTS: Appropriate items reliability (0.99), infit (mean = 0.99; SD = 0.46), outfit (mean = 1.34; SD = 1.58), and point-biserial correlations (-0.36 to 0.92) were obtained. The model explains 74.3% of the variance, and 76.2% of the items had adequate factor relationship. CONCLUSION: The TIMP measures a single construct, has items with different levels of difficulty, and is sensitive to detect differences in the performance of distinct groups for infants in Brazil.