The Ball Aptitude Battery (BAB) was examined for item bias in a sample of 577 high school students in which males and females, as well as three ethnic groups (Blacks, Whites, and Hispanics) were represented. The objectives of the investigation were: (1) to assess the level of interrater agreement for the judgmental method; (2) to find the level of agreement between the judgmental method and several statistical methods; and (3) if agreement was low, to determine whether correspondence was improved when judgmentally biased items were further classified in terms of content or facial validity bias. Two BAB subtests were analyzed for race and sex bias by the judgmental method and by five statistical methods: TID, partial correlation, Rasch difficulty shift, point biserial, and Scheuneman's modified chi square. In the judgmental analysis, the level of interrater agreement among the three expert judges was not significant. The agreements between the statistical and judgmental methods in identification of biased items were also nonsignificant. A final analysis investigated whether agreement between the statistical and judgmental methods was improved when the judgmentally biased items were further classified into one of two types, content validity or face validity bias. Again the results were nonsignificant. (Author/MAC)