Noting a general dissatisfaction with the implementation and evaluation of large-scale educational programs, this paper argues that the rational model that is the foundation for large-scale educational programs is faulty. According to the rational model, government mandates call for educational programs to remedy social problems; then rewards and sanctions (applied according to the results of program evaluations) force schools to implement these programs; and finally the resulting changes alleviate the original social problems. One difficulty with the model is that there is often a tenuous link between actual program intent and evaluative criteria. Evaluative criteria chosen are usually those easily measurable, like test performance, rather than those that might actually indicate that the rather ambiguous program goals (such as reduction of poverty) have been realized. Thus, program intent may at times be decoupled from the evaluative criteria used to force program implementation. A second problem concerns the many possible focuses of evaluation: program compliance, program implementation, educational outcomes, or social outcomes. It has not been demonstrated that evaluations of any of these factors are effective in achieving program intent. For this reason, schools often decouple such dubious outcome data from program decisions. (Author/JM)