The evaluation of an interdisciplinary, process curriculum for real problem solving required a tool for differentiating that program's application from the actual treatments taken by students receiving more traditional alternatives to the experimental program. Treatments could not be controlled or assigned at random, and these restrictions underscored the need to distinguish between programs in natural settings at the level of student learning activities. An observation instrument, "System to Understand the Dynamics of Education in Non-Traditional Settings (STUDENT)," was developed which enabled reliable, practical representation of the students' behaviors in both groups. Its ease of application and analysis of results were illustrated with data contrasting student behaviors in the experimental and comparison groups for two years of data collection. The results also reflected changes in application of the experimental program over the two years. (Author)