Effect of Multidimensionality on Separate and Concurrent Estimation in IRT Equating

Anton A. Béguin, University of Twente
Bradley A. Hanson, ACT, Inc.
Cees A. W. Glas, University of Twente

Paper presented at the Annual Meeting of the National Council on Measurement in Education (New Orleans, April, 2000)

Abstract: The relative performance of separate and concurrent unidimensional IRT estimation could be affected by multidimensionality of the data. This paper reports the results of a simulation study comparing the relative performance of unidimensional estimation methods on multidimensional data. Data based on a two-dimensional IRT model are simulated according to equivalent and nonequivalent groups designs. The results of separate and concurrent unidimensional estimation are compared with the results of concurrent estimation under the two-dimensional model. In this study it becomes clear that multidimensionality of the data can effect the relative performance of separate and concurrent unidimensional estimation methods. The relative performance of separate and concurrent estimation was different for the equivalent and nonequivalent groups conditions. In the nonequivalent group conditions, the error for the unidimensional estimation methods was very large compared to the error obtained using two-dimensional IRT estimation.

PDF Download paper in PDF format (340 KB, compressed). Because of the large size of this PDF document the file is compressed (a Zip archive). It will need to be uncompressed before being viewed with Acrobat Reader. Version 3.0 or later of Adobe Acrobat Reader (which is available for free) is needed to view this paper.


Papers by Brad Hanson

Brad Hanson's Home Page

я нашел на этом сайте