A Comparison of Three Content Balancing Methods for Fixed and Variable Length Computerized Adaptive Tests

Autor/inn/en	Shin, Chingwei David; Chien, Yuehmei; Way, Walter Denny
Institution	Pearson
Titel	A Comparison of Three Content Balancing Methods for Fixed and Variable Length Computerized Adaptive Tests
Quelle	(2012), (41 Seiten) PDF als Volltext Verfügbarkeit
Sprache	englisch
Dokumenttyp	gedruckt; online; Monographie
Schlagwörter	Computer Assisted Testing; Elementary Secondary Education; Test Content; Models; Test Items; Weighted Scores; Comparative Analysis; Simulation; Mathematics Tests; Reading Tests; Certification; Measurement; Correlation; Test Length; Accuracy + Suchen Sie Ihr Suchwort? Testaufgabe; Analogiemodell; Test content; Simulation program; Simulationsprogramm; Lesetest; Abschlusszeugnis; Zertifizierung; Messverfahren; Korrelation
Abstract	Content balancing is one of the most important components in the computerized adaptive testing (CAT) especially in the K to 12 large scale tests that complex constraint structure is required to cover a broad spectrum of content. The purpose of this study is to compare the weighted penalty model (WPM) and the weighted deviation method (WDM) under various conditions including the simple and complicated content constraint structure, different item pools, item exposure control specification, and theta estimation options for both fixed- and variable-length CAT tests. For the tests with simple content constraint structure in which all items are exclusively associated with one and only one constraint, the constraint CAT (CCAT) method was also included in the comparison. For the MATH test, the WPM performed slightly better in the content balancing and measurement precision. However, the disadvantage is that it caused slightly larger maximum item exposure and item-never-used rates. For the RD test, the WPM performed better than the WDM in all the evaluation indices. For the LT test, the WPM and CCAT methods had 100% on-target rate while the WDM has some lower bound violations. The measurement precision and item exposure control results were similar across all three methods. The CRWOB option efficiently reduced the test length from 73 to 68 to reach the termination rule (CSEM = 0.25). The CCAT method with the CRWOB option was the most efficient combination for the LT test because it required the least number of items to reach the same measurement precision while the maximum item exposure rate was controlled around the target rate (0.33). Contains appendices. (As Provided).
Anmerkungen	Pearson. One Lake Street, Upper Saddle River, New Jersey 07458. Tel: 800-848-9500; Web site: http://www.pearsoned.com/
Erfasst von	ERIC (Education Resources Information Center), Washington, DC
Update	2018/2/04