This paper states that the real gains in collecting data on the educational system appear to lie in discovering more about the relationships between schools and what they produce. The required data, which could be collected fairly cheaply from existing records, include (1) education outputs as measured by grade level completion, college attendance, and achievement scores; (2) family outputs as measured by parental occupation, parental education, and family structure; and (3) school inputs as measured by school of attendance, special programs, and specific teachers in each year along with the characteristics of the teachers. All these data should be collected annually for each student in the system. Additional avenues of research that seem to be profitable include both broadening the scope of school output measures and ascertaining the contribution of each school output measure to future life styles. (Author)