Development of the Computerized Adaptive Test of Motor Development (MD-CAT) Adopting Multidimensional Rasch Analysis.

OBJECTIVE:This study aimed to develop the Computerized Adaptive Test of Motor Development (MD-CAT) in preschool children based on multidimensional Rasch analysis. DESIGN:A retrospective study with cross-sectional design. SETTING:A medical center. PARTICIPANTS:A total of 1738 children (N=1738). INTERVENTIONS:Not applicable. MAIN OUTCOME MEASURES:MD-CAT. RESULTS:Multidimensional Rasch analysis was used to develop the item bank of the MD-CAT. The item bank of the MD-CAT contained 74 items, with 44 and 30 items, respectively, for the subscales of gross and fine motor skills. High correlation existed between the 2 subscales (r=0.96). Three stopping rules were set for the MD-CAT: (1) the person reliability achieved 0.95 or the limited reliability increase by <0.01; (2) at least 3 items were assessed in each dimension; and (3) the number of items used for assessment reached 16. Based on the 3 stopping rules, the MD-CAT had high correlations with its total test length (r=0.87-0.98 for the 2 dimensions), indicating sufficient construct validity. The MD-CAT also had adequate diagnostic validity (area under the curve=0.72-0.93) and efficiency (an average of 3-6 items used for the assessment). CONCLUSIONS:The MD-CAT has high precision and efficiency, good construct validity, and high diagnostic validity. The results of our study indicate that the MD-CAT can be useful in clinical practice and in research as a diagnostic measure.