東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

A multilevel IRT model for group-lev...

The University of Wisconsin - Madison.

FindBook

Google Book

Amazon

博客來

A multilevel IRT model for group-level diagnostic assessment with application to TIMSS.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	A multilevel IRT model for group-level diagnostic assessment with application to TIMSS./
作者:	Park, Chanho.
面頁冊數:	143 p.
附註:	Adviser: Daniel M. Bolt.
Contained By:	Dissertation Abstracts International69-09A.
標題:	Education, Educational Psychology. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoeng/servlet/advanced?query=3327824
ISBN:	9780549803324

A multilevel IRT model for group-level diagnostic assessment with application to TIMSS.
Park, Chanho.

A multilevel IRT model for group-level diagnostic assessment with application to TIMSS. - 143 p.

Adviser: Daniel M. Bolt.

Thesis (Ph.D.)--The University of Wisconsin - Madison, 2008.

Educational tests such as TIMSS are developed to compare units above the student level (i.e., countries). This dissertation proposes and investigates a statistical modeling approach based on application of multilevel item response theory (ML-IRT) for group-level diagnosis, which can be used for cross-national comparisons. The model---referred to as the item feature model (IFM)---is first fitted to the TIMSS 2003 Grade 8 mathematics assessment dataset using a Markov chain Monte Carlo (MCMC) procedure implemented in WinBUGS. The application of the IFM studies the content and cognitive features of items as potential contributors to differential item functioning (DIF) across countries. Application of the IFM to the TIMSS data provides results comparable to the methods already used with TIMSS. However, the current methodology likely more effectively separates the relative effects of the cognitive and content features than the current TIMSS reports based on the distribution of separate proficiencies at the group level. Simulation analyses using conditions similar to those of the TIMSS application reveals that item feature confounding is an important factor to consider when designing and developing a test for diagnostic assessment. Reducing residual variability in item difficulty across countries and maximizing item feature variability also appear to help the model perform better. Another set of simulation analyses suggest that the results should be interpreted with caution as a basis for concluding the significance of feature effects, as Type I error inflation does occur under some forms of model misspecification. Strengths and limitations of the current modeling approach with future directions of research are also discussed.

ISBN: 9780549803324Subjects--Topical Terms:

1017560
Education, Educational Psychology.

A multilevel IRT model for group-level diagnostic assessment with application to TIMSS.
LDR:02675nam 2200289 a 45 001 856069
005 20100708
008 100708s2008 ||||||||||||||||| ||eng d
020 $a 9780549803324
035 $a (UMI)AAI3327824
035 $a AAI3327824
040 $a UMI $c UMI
100 1 $a Park, Chanho. $3 1022822
245 1 2 $a A multilevel IRT model for group-level diagnostic assessment with application to TIMSS.
300 $a 143 p.
500 $a Adviser: Daniel M. Bolt.
500 $a Source: Dissertation Abstracts International, Volume: 69-09, Section: A, page: 3456.
502 $a Thesis (Ph.D.)--The University of Wisconsin - Madison, 2008.
520 $a Educational tests such as TIMSS are developed to compare units above the student level (i.e., countries). This dissertation proposes and investigates a statistical modeling approach based on application of multilevel item response theory (ML-IRT) for group-level diagnosis, which can be used for cross-national comparisons. The model---referred to as the item feature model (IFM)---is first fitted to the TIMSS 2003 Grade 8 mathematics assessment dataset using a Markov chain Monte Carlo (MCMC) procedure implemented in WinBUGS. The application of the IFM studies the content and cognitive features of items as potential contributors to differential item functioning (DIF) across countries. Application of the IFM to the TIMSS data provides results comparable to the methods already used with TIMSS. However, the current methodology likely more effectively separates the relative effects of the cognitive and content features than the current TIMSS reports based on the distribution of separate proficiencies at the group level. Simulation analyses using conditions similar to those of the TIMSS application reveals that item feature confounding is an important factor to consider when designing and developing a test for diagnostic assessment. Reducing residual variability in item difficulty across countries and maximizing item feature variability also appear to help the model perform better. Another set of simulation analyses suggest that the results should be interpreted with caution as a basis for concluding the significance of feature effects, as Type I error inflation does occur under some forms of model misspecification. Strengths and limitations of the current modeling approach with future directions of research are also discussed.
590 $a School code: 0262.
650 4 $a Education, Educational Psychology. $3 1017560
650 4 $a Education, Tests and Measurements. $3 1017589
650 4 $a Psychology, Psychometrics. $3 1017742
690 $a 0288
690 $a 0525
690 $a 0632
710 2 $a The University of Wisconsin - Madison. $3 626640
773 0 $t Dissertation Abstracts International $g 69-09A.
790 $a 0262
790 1 0 $a Bolt, Daniel M., $e advisor
791 $a Ph.D.
792 $a 2008
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoeng/servlet/advanced?query=3327824