THREE PARTICULAR PROBLEMS WERE THE CONCERN OF THIS STUDY--(1) THE VALIDITY OF ADMISSIONS DATA USED IN ELEMENTARY TEACHER EDUCATION PROGRAMS, (2) THE EFFECTS OF DIFFERENT STUDENT TEACHING PLACEMENTS, AND (3) THE RELATIONSHIP BETWEEN JUDGMENT RATINGS OF STUDENT TEACHERS MADE BY STUDENT TEACHING SUPERVISORS USING THE REGULARLY EMPLOYED OBSERVATIONAL PROCEDURES OF THEIR OFFICE AND COMPARABLE RATINGS MADE BY RESEARCH OBSERVERS USING PROCEDURES ESPECIALLY SET UP FOR THE PURPOSE OF THIS STUDY. THE DESIGN TREATED FOUR SOURCES OF VARIATION AS EXPERIMENTAL VARIABLES--(1) STUDENT TEACHERS' GRADE LEVEL PREFERENCE, (2) STUDENT TEACHERS' QUARTER OF EXPERIENCE, (3) CLASSROOM GRADE LEVEL, AND (4) ORDER OF PREFERENCE IN TWO DIFFERENT GRADE LEVEL ASSIGNMENTS. THE CONTROLLED VARIABLES (THOSE TREATED IN SUCH A WAY THAT THEY WOULD NOT BIAS THE OBSERVATIONS) WERE--(1) THE PROFESSIONAL EDUCATION OF THE STUDENT TEACHER, (2) THE SEX OF THE STUDENT TEACHER, (3) TRAVEL BY THE STUDENT TEACHER, AND (4) SOCIOECONOMIC LEVEL OF THE CLASSROOM. ON THE BASES OF STUDENT TEACHER OBSERVATIONS AND INTERVIEWS, AND SPECIFIC TEST ADMINISTRATIONS, THE FOLLOWING CONCLUSIONS WERE REACHED--(1) ADMISSION DATA WERE NOT RELIABLE FOR CLASSROOM BEHAVIOR PREDICTIONS, (2) CLASSROOM PLACEMENTS APPEARED TO BE MORE CLOSELY RELATED TO SPECIFIC STUDENT TEACHER RESPONSES TO INDIVIDUAL PUPILS AND TO A CLASS THAN THEY WERE TO GENERALIZED TENDENCIES TO USE CERTAIN CATEGORIES OF VERBAL BEHAVIOR, AND (3) RATINGS BETWEEN THE SUPERVISORS AND THE RESEARCH OBSERVERS ON EITHER PERSONALITY OR CLASSROOM PERFORMANCE JUDGMENTS GENERALLY EXHIBITED A HIGH DEGREE OF AGREEMENT. (JH)