A project conducted in Tennessee from 1984 through 1989, Student Teacher Achievement Ratio (Project STAR), serves as a context for a discussion of educational research. The decisions required in major research projects and the problems in carrying out research are seldom discussed in conferences that present research results as completed efforts. Project STAR illustrates the long-term consequences of early decisions and implementation and the additional value research may have. The research question was established through state legislation, but the researchers had to operationalize or limit key variables. The study was conducted to explore the influence of class size on student achievement. How to measure achievement and how to analyze the test results that were chosen as measures of achievement became questions of importance as the project evolved. Relying solely on student outcomes and certain behavioral indicators was not considered adequate for the study, and researchers began to gather other information about schools, students, and teachers. About 100 classes of each type (small, regular, and regular with an aide) were used each year. The size of the database eventually developed, the care researchers had taken, and the in-school design allowed STAR information to be used in subsidiary and ancillary studies. As each inquiry moved further from the initial research question, however, the power of the study and confidence in the results diminished. (Contains 5 tables and 28 references.) (SLD)