NotesFAQContact Us
Search Tips
ERIC Number: ED463304
Record Type: Non-Journal
Publication Date: 2001-Apr
Pages: 32
Abstractor: N/A
Reference Count: N/A
The Consistency between the Empirical and the Analytical Standard Errors of Multidimensional IRT Item Estimates.
Li, Yuan H.; Yang, Yu N.
An evaluation of the variation of item estimates was conducted for the multidimensional extension of the logistic item response theory (MIRT) model. The empirically determined standard errors (SEs) of marginal maximum likelihood estimation (MMLE)/Bayesian item estimates from 40 items from the ACT Assessment (Form 24b, 1985) were obtained when the same set of items was repeatedly eliminated from test data. These empirically determined SEs were then compared with their corresponding analytical (or formula-based) ones. Both approaches, in general, resulted in similar SE estimates for the same set of items. This empirical comparison implies that the analytical approach has the potential of being used for approximately estimating the magnitudes of SEs of the MMLE/Bayesian item estimates. Tabulation of the analytical SEs for several combinations of item parameters (e.g., low item difficulty, high item discrimination, and low item discrimination was provided as a reference. In addition, the graphical three-dimensional representation of the SEs of item estimates as the bivariate function of item difficulty together with item discrimination was displayed. How to apply the analytical SEs of MIRT item estimates in a MIRT item linking study is illustrated. (Contains 4 tables, 2 figures, and 19 references.) (Author/SLD)
Publication Type: Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A