Similarity of the Multidimensional Space Defined by Parallel Forms of a Mathematics Test.

Autor/inn/en	Reckase, Mark D.; und weitere
Titel	Similarity of the Multidimensional Space Defined by Parallel Forms of a Mathematics Test.
Quelle	(1989), (26 Seiten) PDF als Volltext kostenfreie Datei Verfügbarkeit
Sprache	englisch
Dokumenttyp	gedruckt; online; Monographie
Schlagwörter	Achievement Tests; Factor Analysis; High School Students; High Schools; Item Analysis; Latent Trait Theory; Mathematics Tests; Multidimensional Scaling; Test Construction; Test Format + Suchen Sie Ihr Suchwort? Achievement test; Achievement; Testing; Test; Tests; Leistungsbeurteilung; Leistungsüberprüfung; Leistung; Testdurchführung; Testen; Faktorenanalyse; High school; High schools; Student; Students; Oberschule; Schüler; Schülerin; Studentin; Itemanalyse; Latent-Trait-Modell; Mehrdimensionale Analysis; Testaufbau; Testentwicklung
Abstract	The purpose of the paper is to determine whether test forms of the Mathematics Usage Test (AAP Math) of the American College Testing Program are parallel in a multidimensional sense. The AAP Math is an achievement test of mathematics concepts acquired by high school students by the end of their third year. To determine the dimensionality of the multidimensional space, each test form was factor analyzed at content area and item levels, and the number of dimensions needed to define the space was determined. A multidimensional item response theory analysis (MIRT) determined where in space each item provided the most information. A sample of 2,500 individuals for each form was taken from the equating administration in 1988 of five forms of the AAP Math: 29B, 29C, 29D, 29E, and 29F. A third step was to compare the information provided by each form at 49 points in the space defined by item-person interactions. Factor analysis showed that the tests had a very dominant first factor and a possible weaker second factor. MIRT analysis showed a fairly clear distinction between story-problem items and those only requiring manipulation, but the constructs measured by these item types were highly correlated. The information comparison indicated that the test forms all had the same basic structure. Considering that these test forms were not constructed with multidimensional considerations in mind, they are very similar. Eleven tables and five graphs present study data. (SLD)
Erfasst von	ERIC (Education Resources Information Center), Washington, DC