The purpose of this research was to develop alternative models of a dissertation production system in educational administration. Interviews were conducted with 40 research administrators in public and private research settings; e.g., Bell Laboratories and the National Cancer Institute. Based on information derived from interviews and appropriate readings, three preliminary models were developed, and evaluative criteria were formulated. In addition, a 14-member panel was consulted. Each preliminary model was evaluated by the panel. On the basis of these evaluations, modifications of the models were made. Should these models prove viable, a more effective means of carrying out the dissertation activity will be available. (Author)