ERIC Number: ED264270
Record Type: Non-Journal
Publication Date: 1982-Jul
Reference Count: N/A
Robustness of Adaptive Testing to Multidimensionality.
Weiss, David J.; Suhadolnik, Debra
The present monte carlo simulation study was designed to examine the effects of multidimensionality during the administration of computerized adaptive testing (CAT). It was assumed that multidimensionality existed in the individuals to whom test items were being administered, i.e., that the correct or incorrect responses given by an individual were generated from a specified multidimensional structure, rather than the unidimensional item response theory (IRT) model normally assumed to have generated the observable dichotomous test item responses. The dichotomous response was then treated for CAT item selection and ability estimation purposes as if it had been generated by the unidimensional model. To the extent that the observed item response was affected by dimensions other than the first (which corresponded to the single dimension assumed to underlie the item selection and ability estimation process) errors should be introduced into the adaptive testing process. These errors should affect the ability estimates and the efficiency of CAT. The study focused on the nature and degree of these errors under a variety of multidimensional structures, to determine how robust CAT is to the effects of multidimensionality in examinees' responses to test items. (PN)
Publication Type: Reports - Research; Speeches/Meeting Papers; Numerical/Quantitative Data
Education Level: N/A
Sponsor: Air Force Human Resources Lab., Brooks AFB, TX.; Office of Naval Research, Washington, DC.
Authoring Institution: N/A
Identifiers - Assessments and Surveys: Armed Services Vocational Aptitude Battery