ERIC Number: ED421534
Record Type: Non-Journal
Publication Date: 1996-Jun
Reference Count: N/A
Comparison of Alternative Models for Item Parameter Estimation with Small Samples.
Parshall, Cynthia G.; Kromrey, Jeffrey D.; Chason, Walter M.
The benefits of item response theory (IRT) will only accrue to a testing program to the extent that model assumptions are met. Obtaining accurate item parameter estimates is a critical first step. However, the sample sizes required for stable parameter estimation are often difficult to obtain in practice, particularly for the more complex models. One approach is to use modified item response models, which may be constructed so additional parameters (e.g. more than one) are included in the model, while limiting estimation. This study investigated several modified IRT models across differing sample sizes and test length in terms of their relative efficiency, accuracy, and precision. Simulated data were generated from the American College Testing program mathematics test. For some of the analyses, performance of the models tended to converge at the larger sample sizes, but at the smaller samples, the modified models displayed some important performance differences relative to the unconstrained models. The strongest pattern of results was for models that displayed the best fit within samples to display the poorest stability across samples. Conversely, models that demonstrated good stability across replications tended to be associated with relatively poorer fit within replications. (Contains 3 tables, 5 figures, and 22 references.) (Author/SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Identifiers - Assessments and Surveys: ACT Assessment