Explores problems involved in the quantitative verification of independent variables in investigations of the effects on student outcomes of planned variations in instructional behavior. Specifically addresses (1) gathering of data through direct systematic observation, and (2) analysis of those data through the use of inferential statistics. (GC)

Argues that the complex human behavior studied by educational researchers is by its nature unpredictable if not indeterminate. Concludes that research should focus on descriptions and interpretations of educational phenomena and not on the prediction and control of educational outcomes. (FMW)

Explores reasons for the tenuous relationship between educational research and educational practice. Suggests changes in research and policy to bring educational research directly to bear on educational problems. The major problem is that researchers are searching for variables that relate to improved outcomes, whereas educational innovators are…