Development of an efficient computerized adaptive test measuring children's self-care performance

The Computerized Adaptive Test for Measuring Children's Self-care Performance (CAT-SC) was developed with an interview administration approach for a limited population.

We developed a new CAT-SC (the CAT-SC-T) based on the original candidate item bank of the CAT-SC, using the caregiver-report approach and targeting children with and without disability.

In this cross-sectional study in the general community, 1289 caregivers of children with and without disability were assessed with the original candidate item bank of the CAT-SC.

The item bank of the CAT-SC-T contained 69 items with unidimensionality. The three stopping rules were (1) administering a minimum of 10 items and a maximum of 16 items, (2) observing the first 10 items receiving the lowest scores, and (3) ensuring a reliability of self-care performance estimate of 0.95. From the simulation analyses, only 10-16 items were needed for each assessment to achieve good reliability and validity. The person reliability for most children (83.9 %) exceeded 0.95. Of the 16.1 % not meeting the 0.95 reliability criterion, 19 % (n = 39) were extreme responders. The good concurrent validity of the CAT-SC-T was proven by its very high correlation coefficient (r = 0.99) between the scores using the entire item bank and the estimated scores of the CAT-SC-T. Significant score differences between children with and without disability indicated good discriminant validity.

The CAT-SC-T appears to be an efficient, reliable, and valid assessment of children's self-care performance. The CAT-SC-T can be applied to children with and without disability.