ETHNIC GROUP DIFFERENCES IN SELECTED ASPECTS OF TEST-TAKING BEHAVIOR ON A STANDARDIZED ADMISSIONS TEST
The present study investigated the hypothesis that observed test score differences between ethnic groups might be due, in part, to the different proportion of examinees from each group who were proficient in the use of selected aspects of test-taking behavior. A standardized multiple-choice aptitude test used for admission to a graduate/professional school was employed. The possibility that some examinees guessed on the test while others used either partial information, misinformation, or refused to answer the difficult items could have created ethnic group score differences when the true ability was, in fact, not different. The problem of differentiating those examinees who guesssed throughout the test from those who did not guess was also addressed. Furthermore, the question of whether low scoring examinees tended to guess more than non-low scoring examinees and the extent of such ethnic group differences was investigated. The methodology employed in the present research combined the recent Federal Guidelines for the assessment of equal opportunities in education and employment with tests of statistical significance. Subjects were classified according to ethnicity; examinees were not excluded from the study on the basis of their socioeconomic status, age, sex, or educational level. The personal biserial index which is the correlation of each examinee's correct and incorrect responses with the respective group-determined item difficulties, was used to identify those subjects who probably guessed throughout the test. An examinee identified as a guesser was suspected not to have employed the knowledge required by the examination in selecting responses to difficult multiple-choice test items. Applicants from the white, black, American Indian, asian, and hispanic groups, 95% of whom were college seniors, were compared with respect to the proportion of examinees who (a) obtained extremely low scores, (b) guessed answers throughout the examination, and (c) omitted 10% or more of the verbal multiple-choice items which were not corrected for guessing. Significant differences were obtained between ethnic groups in the proportion of examinees who were identified as low scorers as well as in the proportion of examinees who were identified as guessers. Consistent with past research, a greater proportion of minority examinees were found to be low scorers when compared with the white majority group. However, the only significant post hoc difference detected between ethnic groups in the proportion of guessers was the white-hispanic contrast, perhaps indicating a language difficulty for the bilingual hispanic group. However, no ethnic group differences were discovered in the proportion of guessers when the association between ability level and guessing was known. It was concluded that subjects of low ability would be identified as guessers more often than examinees of higher ability, although some non-low scoring examinees were observed to guess throughout the test. Additionally, the black group was found to exhibit a significantly greater proportion of examinees who omitted 10% or more of the test items when compared with the white majority group. Since, the test was not corrected for guessing, atmosphere bias and poor test-wiseness skills were suggested as possible sources of the observed black-white difference in the omission of large portions of the test. In general, these data were in agreement with the results obtained by other researchers, clarifying some of the incongruities in the literature and reaffirming the usefulness of the personal biserial index as a technique for studying test-taking behavior. Suggestions were included for future research.
BREYER, FRANCIS JAY, "ETHNIC GROUP DIFFERENCES IN SELECTED ASPECTS OF TEST-TAKING BEHAVIOR ON A STANDARDIZED ADMISSIONS TEST" (1981). ETD Collection for Fordham University. AAI8124273.