An Examination of Test Bias and Response Characteristics for Six Candidate Groups Taking the ATGSB.

Autor/in	Echternacht, Gary J.
Institution	Educational Testing Service, Princeton, NJ.
Titel	An Examination of Test Bias and Response Characteristics for Six Candidate Groups Taking the ATGSB.
Quelle	(1972), (33 Seiten) PDF als Volltext kostenfreie Datei Verfügbarkeit
Dokumenttyp	gedruckt; online; Monographie
Schlagwörter	Admission Criteria; Black Students; Business Education; Females; Graduate Students; Racial Bias; Racial Factors; Response Style (Tests); Sex Differences; Test Bias; Test Wiseness; White Students; Graduate Management Admission Test + Suchen Sie Ihr Suchwort? Admission; Admission procedures; Zulassungsbedingung; Zulassungsverfahren; Zulassung; Wirtschaftserziehung; Wirtschaftspädagogik; Weibliches Geschlecht; Graduate Study; Student; Students; Aufbaustudium; Graduiertenstudium; Hauptstudium; Studentin; Racial discrimination; Rassismus; Antwortverhalten; Sex difference; Geschlechtsunterschied; Testkritik
Abstract	Differences between black and white, male and female groups who took the Admission Test for Graduate Study in Business (ATGSB) during February 1971 were examined in three studies. The studies were: (1) a study of the biasedness of the test with respect to the groups, (2) a comparison of the mean criterion scores for those candidates who omit items, and (3) a comparison of the response randomness for the sub-groups involved. The ATGSB in this study consisted of seven separate tests: Reading Recall I, Reading Recall II, Antonyms, Analogies, Sentence Completion, Mathematics, and Data Sufficiency. The tests were conducted in two types of settings, a regular test center which charged a fee and at fee-free centers. A total of 2,930 candidates, who were attending institutions in the West, North Central, South, Northeast, and Foreign Institutions, were sampled for each research question. The groups and their sizes were: fee-free males 485, fee-free females 370, regular center black--males 630, females 150, white males 995 and white females 300. The results of Study One: Statistically Defined Test Bias showed that if item-group interaction is accepted as a definition of test bias, then each section of the ATGSB is biased in some way. It appeared that racial bias is contributing more than sex bias in each of the subtests. In Study Two: Omit Behavior, the white group mean criterion score was the lowest among the three groups in the majority of cases. Study Three: Randomness of Response results showed the randomness in chosing distractors was greatest in the fee-free groups and there was more randomness in female groups than male groups. Conclusions, recommendations and graphs are given. (DB)
Erfasst von	ERIC (Education Resources Information Center), Washington, DC