In the past, studies of group differences in intelligence have primarily used summative test scores and group means as a basis for analysis. This type of analysis can lead to the use of comparative, evaluative terms such as superiority and inferiority of groups. The authors propose an alternative way of studying group differences based on ordering analysis which uses item hierarchies as a basis of comparison between two groups. The subjects were 181 sets of black twins and 223 sets of white twins of elementary school age from the Philadelphia area. Subjects were administered the Raven's (1958) Standard Progressive Matrices Test, Subtests A, B, C, and D. Separate item hierarchies resulting from the use of ordering analysis were generated for each subtest for each of the following groups within the sample: Blacks, whites, males, females, monozygotic twins, dizygotic twins, and low and medium SES subjects. The item hierarchies were then compared with respect to the independent variables of race, sex, zygosity, and socioeconomic status by using the Baker-Hubert inference procedure for ordering analysis. Among the findings were that blacks and whites and males and females had similar item hierarchies for complex items but different item hierarchies for simpler items. The implications of this analysis are discussed. © 1986, Sage Publications. All rights reserved.