Journal of the Learning Sciences · 1997 · 25 citations · 22 references
EducationMathematicsMathematics EducationMathematics InstructionProblem-based LearningSecondary Mathematics EducationGroup Problem SolvingProblem SolvingProblem-solving AssessmentProblem-solving ModelProblem-solving StrategiesSmall-group LearningGrouping For Mathematics InstructionMathematical Problem Solving InstructionMathematics Word Problem InstructionProblem-solving BehaviorProblem-solving GroupsReciprocal TeachingSmall-group Mathematics InstructionTutor Training
The study examined whether a socially assisted group learning method based on modified reciprocal teaching improves fourth graders’ ability to solve complex two‑step word problems. The method involves an adult tutor eliciting students’ solution ideas, facilitating discussion of multiple solutions to reach group agreement, and emphasizing world representations and the transfer of planning skills over time. Students receiving either training scored higher than controls on all problem types, with the modified reciprocal teaching group outperforming the modeling condition on problems requiring external information and mixed‑type quizzes, and they used labeled representations more effectively and inferred implied information earlier.
Abstract A form of socially assisted group learning based on a modified version of reciprocal teaching for mathematics (Campione, Brown, & Connell, 1988) was evaluated in terms of its effectiveness in promoting fourth graders' solution of complex 2-step word problems. The emphasis in this method is for an adult tutor to elicit solution suggestions from students and to explore the multiple solutions in seeking group agreement. The method focuses attention both on world representations underlying numbers and on transfer of problem solution planning skills to children over time. The results were compared with those from a modeling-reinforcement tutoring control condition and a classroom activity control in a 10-week study of fourth graders' solution of 4 types of 2-step math word problems, constructed to avoid keywords. Individuals who received either type of training earned higher scores for all problem types than did controls on quizzes; children in the modified reciprocal teaching condition outscored those in the modeling condition on problems that required information to be added from outside the problem and on a quiz of mixed problem types. Subsequent analysis of solutions and videotaped math behaviors suggested that those in the socially assisted learning condition used labeled representations in problem solving more effectively and deduced implied information earlier and more often than those in the modeling condition.
22