Development of a computerized adaptive test of children's gross motor skills

Objectives: To (1) develop a computerized adaptive test for gross motor skills (GM-CAT) as a diagnostic test and an outcome measure, using the gross motor skills subscale of the Comprehensive Developmental Inventory for Infants and Toddlers (CDIIT-GM) as the candidate item bank; and (2) examine the psychometric properties and the efficiency of the GM-CAT. Design: Retrospective study. Setting: A developmental center of a medical center. Participants: Children with and without developmental delay (N=1738). Interventions: Not applicable. Main Outcome Measures: The CDIIT-GM contains 56 universal items on gross motor skills assessing children's antigravity control, locomotion, and body movement coordination. Results: The item bank of the GM-CAT had 44 items that met the dichotomous Rasch model's assumptions. High Rasch person reliabilities were found for each estimated gross motor skill for the GM-CAT (Rasch person reliabilities =.940-.995, SE=.68-2.43). For children aged 6 to 71 months, the GM-CAT had good concurrent validity (r values =.97-.98), adequate to excellent diagnostic accuracy (area under receiver operating characteristics curve =.80-.98), and moderate to large responsiveness (effect size =.65-5.82). The averages of items administered for the GM-CAT were 7 to 11, depending on the age group. Conclusions: The results of this study support the use of the GM-CAT as a diagnostic and outcome measure to estimate children's gross motor skills in both research and clinical settings.