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 consequences of multidimensional...

University of Alberta (Canada).

Linked to FindBook

Google Book

Amazon

博客來

The consequences of multidimensionality to IRT equating outcomes using a common-items nonequivalent groups design.

Record Type:	Language materials, printed : Monograph/item
Title/Author:	The consequences of multidimensionality to IRT equating outcomes using a common-items nonequivalent groups design./
Author:	Ricker, Kathryn Louise.
Description:	153 p.
Notes:	Source: Dissertation Abstracts International, Volume: 68-08, Section: A, page: 3284.
Contained By:	Dissertation Abstracts International68-08A.
Subject:	Education, Educational Psychology. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoeng/servlet/advanced?query=NR29728
ISBN:	9780494297285

The consequences of multidimensionality to IRT equating outcomes using a common-items nonequivalent groups design.
Ricker, Kathryn Louise.

The consequences of multidimensionality to IRT equating outcomes using a common-items nonequivalent groups design. - 153 p.

Source: Dissertation Abstracts International, Volume: 68-08, Section: A, page: 3284.

Thesis (Ph.D.)--University of Alberta (Canada), 2007.

Two studies were conducted to evaluate the consequences of multidimensionality on equating outcomes when IRT true-score equating is employed under a common-items nonequivalent groups (CI-NEG) design. The Stocking-Lord (1983) scale transformation was employed. The first study was a simulation using realistic item parameters with two 68-item test forms, X and Y. A 2 (form parallelism) x 4 (correlation between dimensions) x 3 (group equivalence) x 3 (location of items measuring the second dimension) factorial design was employed, giving a total of 72 conditions. Each condition was replicated 100 times, and measures of Mean Absolute Difference and proportion of examinees with a Score Difference that Matters (Dorans & Feigenbaum, 1994) were examined. The second study used real data taken from two separate licensure tests. Each test was split to create two forms. The scores from each form were equated to each other using equivalent and nonequivalent groups through the common-items nonequivalent groups design. The equating procedures were conducted in a manner identical to those used in the simulation study.

ISBN: 9780494297285Subjects--Topical Terms:

1017560
Education, Educational Psychology.

The consequences of multidimensionality to IRT equating outcomes using a common-items nonequivalent groups design.
LDR:03200nam 2200265 a 45 001 855979
005 20100708
008 100708s2007 ||||||||||||||||| ||eng d
020 $a 9780494297285
035 $a (UMI)AAINR29728
035 $a AAINR29728
040 $a UMI $c UMI
100 1 $a Ricker, Kathryn Louise. $3 1022712
245 1 4 $a The consequences of multidimensionality to IRT equating outcomes using a common-items nonequivalent groups design.
300 $a 153 p.
500 $a Source: Dissertation Abstracts International, Volume: 68-08, Section: A, page: 3284.
502 $a Thesis (Ph.D.)--University of Alberta (Canada), 2007.
520 $a Two studies were conducted to evaluate the consequences of multidimensionality on equating outcomes when IRT true-score equating is employed under a common-items nonequivalent groups (CI-NEG) design. The Stocking-Lord (1983) scale transformation was employed. The first study was a simulation using realistic item parameters with two 68-item test forms, X and Y. A 2 (form parallelism) x 4 (correlation between dimensions) x 3 (group equivalence) x 3 (location of items measuring the second dimension) factorial design was employed, giving a total of 72 conditions. Each condition was replicated 100 times, and measures of Mean Absolute Difference and proportion of examinees with a Score Difference that Matters (Dorans & Feigenbaum, 1994) were examined. The second study used real data taken from two separate licensure tests. Each test was split to create two forms. The scores from each form were equated to each other using equivalent and nonequivalent groups through the common-items nonequivalent groups design. The equating procedures were conducted in a manner identical to those used in the simulation study.
520 $a In the simulation study, equating with parallel forms tended to be robust under most conditions, but equating with nonparallel forms tended not to be robust even under unidimensional conditions. In both simulated and real data studies, when the second dimension items were among both the unique and common items, equating tended to be more robust than when the items were in the other two locations. Equating was least robust when the items measuring the second dimension were only represented in the common items. The simulation study results support the previously established relationship between the correlation between test dimensions and IRT equating robustness. In both studies, IRT equating using a CI-NEG design tended to be more robust when the groups were equivalent. Equating benefits were limited with nonparallel forms. In many nonparallel conditions, error associated with scores was larger when equating was performed than when unequated. Collectively, these results provide evidence that factors beyond just the degree of multidimensionality present on tests mediate the robustness of IRT equating, including form parallelism, the location of the items measuring the second dimension, and group equivalence.
590 $a School code: 0351.
650 4 $a Education, Educational Psychology. $3 1017560
650 4 $a Psychology, Psychometrics. $3 1017742
690 $a 0525
690 $a 0632
710 2 $a University of Alberta (Canada). $3 626651
773 0 $t Dissertation Abstracts International $g 68-08A.
790 $a 0351
791 $a Ph.D.
792 $a 2007
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoeng/servlet/advanced?query=NR29728