The Raven Progressive Matrices (RPM) were administered to groups of approximately 15 to 25 (for a total of 174) special class, educable mentally retarded Negro or white students, 5 1/2 to 10 1/2 years of age from low socioeconomic status (SES) backgrounds to determine whether individual or group training with learning potential procedure was more effective in improving RPM performance. A training booklet containing test items such as pattern completion and analogy problems was used by each S in two training sessions. Training criteria included strategies to assure child comprehension of task requirements, initial demonstration of task requirements, task requirement concretization in motoric performances (child drew item completing pattern before selecting alternative solutions), and verbal response of child on method of arriving at a solution. Results indicated that both the individually trained and group trained groups demonstrated higher mean increase than the control group; that pretest score, age, SES, race, and training condition were significantly related to posttest scores; that older students, males, or white children received higher posttest scores than younger students, females, or Negroes; that SES was not a significant predictor of performance following training; and that middle class students in special classes had physical involvements not clinically evident in low SES students. (MC)