Validation of Score Interpretations for the BDI-2 Using Rasch Methodology
thesisposted on 20.06.2014, 00:00 by Erica M. LaForte
Research has shown the positive impacts of early intervention for children who experience developmental delay. Several challenges exist for professionals tasked with identifying children with developmental delay, designing intervention programs, and tracking the progress of the children who receive intervention services. The Battelle Developmental Inventory, Second Edition (BDI-2) is a measure of early childhood development that can provide a psychometrically sound solution to several of the challenges facing early childhood educators. In this study, I use the validity framework proposed by Wolfe and Smith (2007) and Rasch measurement analyses to gather evidence relevant to the structural, substantive, and generalizability aspects of validity for the BDI-2 Gross Motor subdomain scores. The results of my analyses provide evidence to support the structural and generalizability aspects of validity for the BDI-2 Gross Motor subdomain scores. The Rasch model assumptions of undimensionality and local independence are met. The item and examinee separation indices and separation reliabilities are high. The evidence I gathered relevant to the substantive aspect of validity suggests that examiners may not have used the three-category BDI-2 scoring system as the test developer intended; however, an optimized two-category scoring system produced an examinee ability rank order that was nearly identical to the examinee ability rank order from the three-category scoring system. Additionally, I identified some anomalous examinee score strings in the dataset. Removal of these unexpected scores did not impact the rank-order of the item difficulty measures.