Abstract
This study examined the effectiveness of a data-based instructional procedure when applied by student teachers in classrooms for children with mild learning handicaps. Student teachers were randomly assigned to training and supervision in either the data-based problem-solving approach to instruction or a conventional training and supervision treatment. Analyses of pupil performances revealed that student teachers in the data-based condition had significantly greater effects upon achievement in reading and greater control of off-task behavior during both seat-work and teacher-led instruction than student teachers in the comparison condition. There was no difference between groups for pupil performance in arithmetic.