Premium
An evaluation of the construct validity of the Developmental Test of Visual‐Motor Integration using the Rasch Measurement Model
Author(s) -
Brown Ted,
Unsworth Carolyn,
Lyons Carissa
Publication year - 2009
Publication title -
australian occupational therapy journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.595
H-Index - 44
eISSN - 1440-1630
pISSN - 0045-0766
DOI - 10.1111/j.1440-1630.2009.00811.x
Subject(s) - rasch model , differential item functioning , psychology , construct validity , respondent , test (biology) , logistic regression , logit , item bank , scale (ratio) , psychometrics , clinical psychology , statistics , item response theory , developmental psychology , mathematics , paleontology , physics , quantum mechanics , political science , law , biology
Background: One method of evaluating the construct validity of instruments is the Rasch Measurement Model (RMM), an increasingly popular method used for test construction and validation. Aim: The aim was to examine the construct validity of the Developmental Test of Visual‐Motor Integration 5th Edition (VMI) by applying the RMM to evaluate its scalability, dimensionality, differential item functioning and hierarchical ordering. Method: The participants were 400 children aged 5 to 12 years, recruited from six schools in Melbourne, Victoria, who completed the VMI under the supervision of an occupational therapist. VMI items 1, 2 and 3 were excluded from the Rasch analysis since all of the children achieved a perfect score on these items. Results: None of the items exhibited RMM misfit due to goodness‐of‐fit mean square (MnSq) infit statistics and standardised z (ZStd) scores being outside the specified acceptable range. VMI item 9 (copied circle) exhibited differential item functioning based on gender. In relation to hierarchical ordering of items, several were found to have similar logit difficulty values. For example, VMI items 26, 27 and 29; items 18, 22 and 24; and items 4, 5 and 11 were found to have the same level of challenge. As well, the VMI scale item logit measure order did not match that presented in the VMI test manual. Conclusion: Theoretically, the VMI items are developmentally ordered; however, this ordering was not mirrored by the item logit difficulty scores obtained. This has scoring implications, where scoring a respondent's VMI test booklet is terminated after three consecutive items are not passed. Clinicians should also be aware that item 9 may exhibit bias related to gender.