ITEM BIAS IN THE MCCARTHY SCALES OF CHILDREN'S ABILITIES FOR ANGLO AND MEXICAN-AMERICAN CHILDREN

This study investigated cultural bias in the 46 Verbal items of the McCarthy Scales of Children's Abilities (MSCA). Two separate approaches to the examination of item bias were utilized. The first approach examined item bias empirically by comparing performance differences in terms of correct item responses given by examinees from Anglo and Mexican-American cultures. The second approach addressed the issue of face validity by obtaining the opinion of Anglo-American (AA) and Mexican-American (MA) judges regarding their perception of item bias for AA and MA examinees. The two criteria used to judge item bias were examinee opportunity to learn item content and examinee familiarity with the language used in the items. The Verbal items of the McCarthy Scales were administered to 59 Anglo-American and 59 Mexican-American examinees matched for age and sex. Ability differences were controlled by generating common ability intervals for both groups based on overall Verbal scale score. The obtained data was tested utilizing procedures for the analysis of contingency tables. Two statistical analyses applied to the data were Scheuneman's modified chi-square and the log-linear technique using the likelihood ratio chi-square statistic. Statistically significant performance differences between the two groups, identified by both analyses, were found for only two items of the Word Knowledge II subtest which measures vocabulary comprehension. The differences in perception of the two groups of judges in terms of opportunity to learn item content and familiarity with item language across AA and MA examinees was examined. The obtained ratings of item bias were tested with the chi-square statistic. Significant differences in ratings of two groups of judges on the opportunity to learn dimension were found for 16 items. It was revealed that more AA judges perceived 14 of these items to be fair for both groups of examinees in terms of opportunity to learn, while more MA judges perceived the AA child as having more opportunity to learn the content of these items. However for two of the 16 items the statistically significant difference between the two groups reflected the perception of MA judges that bias favored the MA examinees. There was a high degree of agreement between the two groups of judges in their rating on the familiarity dimension, with both groups indicating more familiarity with item language for AA subjects for most items. Significant differences in the ratings of the two groups were identified for four items. These items were from the Word Knowledge and Verbal Memory subtests. For three of these items the obtained data indicated that the greater number of MA judges perceived these items as favoring AA subjects. For the remaining item the results pointed out that more MA judges perceived that MA subjects seemed to have greater familiarity with the item language as compared with their Anglo counterpart. There was no discernible pattern of judgment in terms of item difficulty, with easier items perceived as more biased than more difficult ones in some cases. Judge responses appear more related to specific item content than level of difficulty or verbal complexity. The findings from both studies led to the conclusion that the majority of the verbal items in the McCarthy Scales seemed to be fair for Anglo and Mexican-American subjects. Implications of these findings were discussed and recommendations were made for future studies intended to examine item bias