東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

The effects of model specification e...

Jiao, Hong.

Linked to FindBook

Google Book

Amazon

博客來

The effects of model specification error in Item Response Theory-based computerized classification test using sequential probability ratio test.

Record Type:	Electronic resources : Monograph/item
Title/Author:	The effects of model specification error in Item Response Theory-based computerized classification test using sequential probability ratio test./
Author:	Jiao, Hong.
Description:	152 p.
Notes:	Source: Dissertation Abstracts International, Volume: 64-02, Section: A, page: 0478.
Contained By:	Dissertation Abstracts International64-02A.
Subject:	Education, Tests and Measurements. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3081443

The effects of model specification error in Item Response Theory-based computerized classification test using sequential probability ratio test.
Jiao, Hong.

The effects of model specification error in Item Response Theory-based computerized classification test using sequential probability ratio test. - 152 p.

Source: Dissertation Abstracts International, Volume: 64-02, Section: A, page: 0478.

Thesis (Ph.D.)--The Florida State University, 2003.

This study investigated the effects of model specification error on classification accuracy, error rates, and average test length in Item Response Theory (IRT) based computerized classification test (CCT) using sequential probability ratio test (SPRT) in making binary decisions from examinees' dichotomous responses.Subjects--Topical Terms:

1017589
Education, Tests and Measurements.

The effects of model specification error in Item Response Theory-based computerized classification test using sequential probability ratio test.
LDR:03080nmm 2200289 4500 001 1855541
005 20040610112948.5
008 130614s2003 eng d
035 $a (UnM)AAI3081443
035 $a AAI3081443
040 $a UnM $c UnM
100 1 $a Jiao, Hong. $3 1943349
245 1 0 $a The effects of model specification error in Item Response Theory-based computerized classification test using sequential probability ratio test.
300 $a 152 p.
500 $a Source: Dissertation Abstracts International, Volume: 64-02, Section: A, page: 0478.
500 $a Major Professor: Akihito Kamata.
502 $a Thesis (Ph.D.)--The Florida State University, 2003.
520 $a This study investigated the effects of model specification error on classification accuracy, error rates, and average test length in Item Response Theory (IRT) based computerized classification test (CCT) using sequential probability ratio test (SPRT) in making binary decisions from examinees' dichotomous responses.
520 $a This study consisted of three sub-studies. In each sub-study, one of the three unidimensional dichotomous IRT models, the 1-parameter logistic (IPL), the 2-parameter logistic (2PL), and the 3-parameter logistic (3PL) model was set as the true model and the other two models were treated as the misfit models. Item pool composition, test length, and stratum depth were manipulated to simulate different test conditions. To ensure the validity of the study results, the true model based CCTs using the true and the recalibrated item parameters were compared first to study the effect of estimation error in item parameters in CCTs. Then, the true model and the misfit model based CCTs were compared to accomplish the research goal,
520 $a The results indicated that estimation error in item parameters did not affect classification results based on CCTs using SPRT. The effect of model specification error depended on the true model, the misfit model, and the item pool composition. When the IPL or the 2PL IRT model was the true model, the use of another IRT model had little impact on the CCT results. When the 3PL IRT model was the true model, the use of the 1PL model raised the false positive error rates. The influence of using the 2PL instead of the 3PL model depended on the item pool composition. When the item discrimination parameters varied greatly from uniformity of one, the use of the 2PL IRT model raised the false negative error rates to above the nominal level. In the simulated test conditions with test length and item exposure constraints, using a misfit model in CCTs most often affected the average test length. Its effects on error rates and classification accuracy were negligible. It was concluded that in CCTs using SPRT, IRT model selection and evaluation is indispensable
590 $a School code: 0071.
650 4 $a Education, Tests and Measurements. $3 1017589
650 4 $a Psychology, Psychometrics. $3 1017742
690 $a 0288
690 $a 0632
710 2 0 $a The Florida State University. $3 1017727
773 0 $t Dissertation Abstracts International $g 64-02A.
790 1 0 $a Kamata, Akihito, $e advisor
790 $a 0071
791 $a Ph.D.
792 $a 2003
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3081443