The process of dealing with real world problems often requires solutions that would not be totally acceptable in the world of "pure" scientific research. A systems analytic viewpoint demands that everything possible be done to define the constraints of the system and that the procedures used to construct a model out of imperfect data be specified. Four types of problems are encountered in doing systems research: (1) specification of output measures, (2) data incompatibility where different instruments are used, (3) data incompatibility where the concept is not precisely captured by existing data, and (4) missing data. Procedures for solving the problems in each of these problem areas are suggested. (HW)