ERIC Number: EJ717932
Record Type: Journal
Publication Date: 2005
Pages: 30
Abstractor: Author
ISBN: N/A
ISSN: ISSN-0895-7347
EISSN: N/A
Available Date: N/A
Bayesian or Non-Bayesian: A Comparison Study of Item Parameter Estimation in the Three-Parameter Logistic Model
Gao, Furong; Chen, Lisue
Applied Measurement in Education, v18 n4 p351-380 2005
Through a large-scale simulation study, this article compares item parameter estimates obtained by the marginal maximum likelihood estimation (MMLE) and marginal Bayes modal estimation (MBME) procedures in the 3-parameter logistic model. The impact of different prior specifications on the MBME estimates is also investigated using carefully selected prior distributions. The results indicate that, in general, the MBME provides more accurate item parameter estimates than the MMLE procedure. The impact of different priors on the Bayesian estimates is modest when the examinee sample size is not extremely small.
Descriptors: Simulation, Computation, Bayesian Statistics, Item Analysis, Statistical Analysis, Evaluation Methods, Maximum Likelihood Statistics
Lawrence Erlbaum Associates, Inc., Journal Subscription Department, 10 Industrial Avenue, Mahwah, NJ 07430-2262. Tel: 800-926-6579 (Toll Free); e-mail: journals@erlbaum.com.
Publication Type: Journal Articles; Reports - Evaluative
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A
Grant or Contract Numbers: N/A
Author Affiliations: N/A