ERIC Number: ED525528
Record Type: Non-Journal
Publication Date: 2010
Pages: 199
Abstractor: As Provided
ISBN: ISBN-978-1-1244-9550-7
ISSN: N/A
EISSN: N/A
Comparing Accuracy of Parameter Estimation Using IRT Models in the Presence of Guessing
Fu, Qiong
ProQuest LLC, Ph.D. Dissertation, University of Illinois at Chicago
This research investigated how the accuracy of person ability and item difficulty parameter estimation varied across five IRT models with respect to the presence of guessing, targeting, and varied combinations of sample sizes and test lengths. The data were simulated with 50 replications under each of the 18 combined conditions. Five IRT models were employed for the data analyses, including (a) the Rasch model, (b) the 3PL model, (c) the Rasch model plus a fixed lower asymptote, (d) the Rasch model plus CUTLO = 1 to adjust for mismatched ability and difficulty in the simulated data, and (e) the Rasch model plus CUTLO = 2. The parameter recovery across the models was evaluated using four criteria: (a) means, standard deviations, 2.5 and 97.5 percentiles of the ability and difficulty parameter estimates; (b) correlations (Kendall's tau-b) between the ability parameters and estimates, and correlations (Kendall's tau-b) between the difficulty parameters and estimates; (c) root mean squared deviations of the ability and difficulty estimates from their respective parameters; and (d) mean signed differences between the ability and difficulty parameters and their respective estimates. The accuracy of parameter estimation by the five models varied with all three factors--guessing, targeting, and combinations of sample sizes and test lengths. The accuracy also depended on the evaluation criteria used. Each model outperformed some other models for particular combined conditions. The practical implications of the findings were discussed along with limitations of the present research and suggestions for the future research. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]
Descriptors: Item Response Theory, Guessing (Tests), Accuracy, Computation, Difficulty Level, Test Items, Test Length, Sample Size, Correlation, Evaluation Criteria
ProQuest LLC. 789 East Eisenhower Parkway, P.O. Box 1346, Ann Arbor, MI 48106. Tel: 800-521-0600; Web site: http://www.proquest.com/en-US/products/dissertations/individuals.shtml
Publication Type: Dissertations/Theses - Doctoral Dissertations
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A
Grant or Contract Numbers: N/A