An Investigation Of Bias In Tests Of Writing Ability For Bilingual Hispanic College Students
Date of Award
Doctor of Philosophy (Ph.D.)
Purpose. The purpose of this study was to investigate whether bias exists in the prediction of Hispanics' performance in first year English courses based on direct measures (writing samples) and indirect measures (objective tests) of writing ability.Premise. Existing research has mainly investigated test bias for Blacks, Mexican-Americans and Chicanos, but virtually no studies have investigated test bias for Cuban-Americans, the third largest Hispanic subgroup in the United States. This study is an attempt to fill this void in the literature by targeting the Cuban-American subgroup of the Hispanic population.Design. The performance in a first year college English course of over 2,800 students was analyzed as predicted by their grade in an objective writing test and a holistically scored writing sample. Cleary's Regression Model was used to test for bias between the two ethnic groups: white, non-Hispanics (the majority group) and Hispanics, mainly composed of bilingual Cuban-Americans whose native language is Spanish (the minority group).Findings and Recommendations. The direct assessment of writing ability, the writing sample, appears to have exhibited intercept bias. However, the prediction equation of the majority group overpredicted the performance of the minority group in the English course by such a small amount, that the statistical significance could be attributed to the large size of the sample. Thus, the writing sample test could be treated as unbiased.The indirect assessment of writing ability, the standardized objective test, proved to predict equally well the performance of Cuban-American Hispanics and that of white, non-Hispanics in the first year college English course.In conclusion, amidst the numerous claims that Cuban-American Hispanics are at a disadvantage in taking tests because of their bilingual status, this study has demonstrated that tests of writing ability do not show evidence of bias for a Hispanic subgroup composed of predominantly Cuban-Americans in a first year college English course. This fact should be publicized by both higher education institutions and by test publishers to clarify existing misconceptions.
Education, Tests and Measurements
Isern, Margarita, "An Investigation Of Bias In Tests Of Writing Ability For Bilingual Hispanic College Students" (1986). Dissertations from ProQuest. 1546.