Individual versus collaborative problem solving: divergent outcomes depending on task complexity

• Published in: Instructional science 2013-11, Vol.41 (6), p.1153-1172
• Main Authors: Sears, David A., Reagin, James Michael
• Format: Article
• Language: English
• Subjects: Acceleration (Education); Achievement Gains; Collaboration; Collaborative learning; Conventional Instruction; Cooperative Learning; Cooperative Programs; Coordination; Difficulty Level; Dyadic relations; Education; Educational Psychology; Grade 7; Group Dynamics; Group performance; Intermode Differences; Laughter; Learning and Instruction; Mathematical complements; Mathematical problems; Mathematics Education; Mathematics Instruction; Middle school students; Pedagogic Psychology; Performance Based Assessment; Pretests; Problem Solving; Productivity; Reagins; Task Analysis; Teaching Methods; Transfer of Training
• ISSN: 0020-4277; 1573-1952
• DOI: 10.1007/s11251-013-9271-8

Many studies have tested external supports for promoting productive collaboration, but relatively few have examined what features characterize naturally productive collaborative tasks. Two lines of research have come to distinct conclusions on the primary task feature associated with productive collaboration: demonstrability versus complexity. This study examined the problem-solving performance of 110 seventh grade students on a demonstrable mathematical task, including 69 in three traditional math classrooms (for whom the task was complex) and 41 in two accelerated math classrooms (for whom the task was not complex). Students were further assigned to one of four conditions split by two factors: grouping (individual versus dyad) and number of problems (one or two). For the accelerated math classes, individuals performed significantly better than dyads. For the traditional math classes, dyads performed significantly better than individuals and exceeded the truth-wins criterion (a theoretical maximum indicating how individuals would perform if they shared knowledge perfectly). A complex-demonstrable task framework is proposed for characterizing naturally productive collaborative tasks.