Since 1968 educational productivity studies at the University of Florida have been analyzing data from six States and one city. Linear regression was used to identify high and low productive units by measuring the relationship between statistically selected input factors and a measure of student achievement. Discriminant analysis was employed to determine which of several input variables were significantly associated with the classification. The design represents a refinement of efforts to determine the relative importance of factors subject to administrative control in effecting high student achievement. From their analysis, the authors concluded that, of those conditions subject to control by local school district administrators, teacher-related variables constitute the most important group of factors associated with educational achievement. (Author)