ERIC Number: ED373547
Record Type: RIE
Publication Date: 1993-May
Reference Count: N/A
A Study of Item Bias in the Maine Educational Assessment Test.
Smith, James Brian
A study used four statistical item bias analysis strategies to determine the French cross-cultural validity of the Maine Educational Assessment, a standardized test administered in six content areas to students in grades 4, 8, and 11. Analysis was performed on eighth grade pupil performance in test year 1988-89, in the areas of the 100 common reading and mathematics items that all pupils take. The four statistical procedures used were: Scheuneman's modified chi-square; Rudner and Convey's TID-45 degree item difficulty p-value; the Rasch one-parameter latent trait model; and the Mantel-Haenszel procedure. Item response comparisons were made with two of Maine's pupil populations: 336 French bilingual/English-fluent speakers and 336 monolingual English-speakers. Findings show that 8 of 50 mathematics items and 9 of 50 reading items indicated differential validity. However, they also indicated differential functioning in not favoring significantly either language group. It is concluded that lower mean scores on the test among bilingual/English-fluent students are not due to item bias, as suggested in an earlier study. It is suggested that misidentification of bilingual/English-fluent students is a primary cause of the discrepancy in scores. (MSE)
Publication Type: Dissertations/Theses - Doctoral Dissertations
Education Level: N/A
Authoring Institution: N/A
Identifiers: Maine Educational Assessment
Note: Doctoral dissertation, Boston College.