A number of methods have been used to identify potentially biased items within a test. These methods examine one item at a time and do not deal with the complex interrelationships among items or among items and the potentially biasing elements. The use of multivariate procedures to assess whether or not items are biased and to obtain clues about the source of the bias are demonstrated here. A total of 1,294 six and seven year old children from five ethnic groups took the Goodenough-Harris Drawing Test. Principal components analysis and analysis of variance were performed on the results. Other analysis methods are suggested, and are presently being studied. (Author/BW)