ERIC Number: ED359223
Record Type: Non-Journal
Publication Date: 1993-Apr
Reference Count: N/A
A Didactic Example of the Influence of Conditioning on the Complete Latent Ability Space When Performing DIF Analyses.
Ackerman, Terry A.; Evans, John A.
A didactic example is provided, using a Monte Carlo method, of how differential item functioning (DIF) can be eliminated (and thus better understood) when the complete latent space is used. The main source of DIF is that the matching single criterion used in some DIF procedures, Mantel Haenszel or Simultaneous Item Bias (SIBTEST), does not account for all of the abilities used by examinees in both groups of interest. To resolve this problem, several researchers have tried to match on secondary variables, but with no success. In this paper, response data are generated from a 2-dimensional item response theory model for a 30-item test in which items are measuring uniformly spaced composites of theta(sub 1) and theta(sub 2). Two DIF detection methods, the Mantel Haenszel and SIBTEST detection procedures, are used under 3 reference/focal sample size conditions (1,000/250, 1,000/500, and 500/250). When the procedures conditioned on the number correct score, only on theta(sub 1), or only on theta(sub 2), the differential group performance followed predictable patterns. Likewise, when the matching criterion was a function of both thetas (i.e., the complete latent space was identified), the DIF was eliminated for all items as hypothesized. Twelve tables and two figures illustrate the analyses. (SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A