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 examination of local item depend...

Dresher, Amy R.

Linked to FindBook

Google Book

Amazon

博客來

The examination of local item dependency of NAEP assessments using the testlet model.

Record Type:	Language materials, printed : Monograph/item
Title/Author:	The examination of local item dependency of NAEP assessments using the testlet model./
Author:	Dresher, Amy R.
Description:	271 p.
Notes:	Adviser: Clement A. Stone.
Contained By:	Dissertation Abstracts International63-10B.
Subject:	Education, Tests and Measurements. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3066946
ISBN:	0493863559

The examination of local item dependency of NAEP assessments using the testlet model.
Dresher, Amy R.

The examination of local item dependency of NAEP assessments using the testlet model. - 271 p.

Adviser: Clement A. Stone.

Thesis (Ph.D.)--University of Pittsburgh, 2002.

The purpose of this study was to compare the item and ability parameter estimates when item dependency was present using two methods: (1) a modified item response theory (IRT) model that models a testlet interaction using Markov chain Monte Carlo (MCMC) techniques in SAS; and, (2) marginal maximum a posteriori estimation (MMAP) using PARSCALE, a typical IRT estimation procedure which ignores the dependency. Because the MCMC method has only recently been applied to IRT estimation, a secondary purpose of the study was to compare standard methods for estimating IRT models with MCMC methods.

ISBN: 0493863559Subjects--Topical Terms:

1017589
Education, Tests and Measurements.

The examination of local item dependency of NAEP assessments using the testlet model.
LDR:03002nam 2200301 a 45 001 930432
005 20110429
008 110429s2002 eng d
020 $a 0493863559
035 $a (UnM)AAI3066946
035 $a AAI3066946
040 $a UnM $c UnM
100 1 $a Dresher, Amy R. $3 1253994
245 1 0 $a The examination of local item dependency of NAEP assessments using the testlet model.
300 $a 271 p.
500 $a Adviser: Clement A. Stone.
500 $a Source: Dissertation Abstracts International, Volume: 63-10, Section: B, page: 4960.
502 $a Thesis (Ph.D.)--University of Pittsburgh, 2002.
520 $a The purpose of this study was to compare the item and ability parameter estimates when item dependency was present using two methods: (1) a modified item response theory (IRT) model that models a testlet interaction using Markov chain Monte Carlo (MCMC) techniques in SAS; and, (2) marginal maximum a posteriori estimation (MMAP) using PARSCALE, a typical IRT estimation procedure which ignores the dependency. Because the MCMC method has only recently been applied to IRT estimation, a secondary purpose of the study was to compare standard methods for estimating IRT models with MCMC methods.
520 $a A Monte Carlo study was performed in which mild and extreme dependency were simulated using three different techniques and the composition of the test was varied by the number of items, the number of testlets, and the percent of items in a testlet. In addition, the estimation procedures were applied to National Assessment of Educational Progress (NAEP) data. The parameter estimates were compared using the bias, or the difference between the parameter estimates and the true values, the root mean square error (RMSE), which provided an estimate of the spread or unsigned difference between the estimates and true values, the correlation of the estimated and true values, and the relative precision, or ratio of the test information functions.
520 $a When locally independent data was simulated, no clear advantage of using one estimation procedure over the other was found. When mild dependency was present, the MCMC procedure that estimated a testlet effect yielded smaller bias and RMSE values, indicating that it was more precise in recovering the item and ability parameters. When extreme dependency was present, both estimation methods produced large bias and RMSE values, indicating that neither method yielded good recovery of the parameters. However, the presence of such extreme dependency may be unlikely in real applications. For the NAEP analysis, there was evidence suggesting that the MCMC procedure that estimated a testlet effect yielded more accurate estimates.
590 $a School code: 0178.
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 University of Pittsburgh. $3 958527
773 0 $t Dissertation Abstracts International $g 63-10B.
790 $a 0178
790 1 0 $a Stone, Clement A., $e advisor
791 $a Ph.D.
792 $a 2002
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3066946