Report
Relative Precision of Ability Estimation in Polytomous CAT: A Comparison under the Generalized Partial Credit Model and Graded Response Model
Distributed by ERIC Clearinghouse
2002
Abstract
The purpose of this Monte Carlo study was to evaluate the relative accuracy of T. Warm's weighted likelihood estimate (WLE) compared to maximum likelihood estimate (MLE), expected a posteriori estimate (EAP), and maximum a posteriori estimate (MAP), using the generalized partial credit model (GPCM) and graded response model (GRM) under a variety of computerized adaptive testing conditions. In general, for all four theta estimation methods, conditional and overall bias, standard error (SE), and root mean square error (RMSE) decreased as test length, test reliability, and item bank size increased. The magnitudes of the differences among the dependent variables decreased as the values of the independent variables increased. For both models, WLE outperformed MLE in terms of all the dependent variables studied, and WLE performed better than the Bayesian methods in terms of bias. MLE had less bias than both Bayesian methods. In general, for the fixed length test, both the GPCM and the GRM models, estimation method, and test length has some impact on bias, SE, and RMSE. But, the model factor had the greatest impact on RMSE, accounting for 31.2% of the total variance of RMSE under the GRM. For the fixed test reliability, the model factor had almost no influence on bias, SE, and RMSE under the GRM.(Contains 4 tables and 24 references.) (Author/SLD)
Details
- Title: Subtitle
- Relative Precision of Ability Estimation in Polytomous CAT: A Comparison under the Generalized Partial Credit Model and Graded Response Model
- Creators
- Shudong WangTianyou Wang
- Resource Type
- Report
- Publisher
- Distributed by ERIC Clearinghouse
- Language
- English
- Date published
- 2002
- Academic Unit
- Center for Advanced Studies in Measurement and Assessment
- Record Identifier
- 9984632142902771
Metrics
1 Record Views