This paper presents a case study of a sixth grade remedial math class which illustrates the thesis that only the "experimental attitude," not the "experimental method," is appropriate in the classroom. The thesis is based on the fact that too many independent variables exist in a classroom situation to allow precise measurement. The case study involved 23 sixth graders and their teacher, who referred himself and his class for help on behavioral management and instruction because he was concerned with students' resistance to work and hostility toward himself. Baseline data taken over a 5-day period produced certain assumptions about the class. From these assumptions, specific recommendations about the teacher's instructional presentation and the assigned work period were made in an effort to engineer behavior change and increase the students' grades. Subsequently, math grades and behavior did improve. A reversal to conditions prior to the experimental changes again produced disruptive behavior and poor grades. The findings were considered to be a result of the experimental conditions, but which of the conditions produced the change could not be discerned. Data tables comprise approximately one-half of the document. (SDH)