Collaborative learning requires a structured and open environment where individuals can actively exchange and elaborate their ideas to achieve a high-quality problem-solving solution. The use of concept map has been extensively implemented to facilitate idea generation and maintain shared focus during discussion. This study employs the Reciprocal Kit-Build (RKB) approach as a designed activity to support collaborative concept mapping. Though previous studies show the RKB is promising to promote productive discussion and achieve high-quality group products, they have not investigated how individual knowledge differences may potentially influence the effectiveness of collaboration. This study aims to identify the effect of group composition on learning effectiveness at the level of interaction between individual and group and at the level of the group as a whole. At the interaction level, we investigate the amount of knowledge transfer from shared and unshared individual knowledge to group solutions, while at the group level, the quality of collaborative maps is evaluated. Moreover, we explore the affective responses of learners during designated activities. A single group design is applied to illustrate the learning activities in a practical classroom settings where all students receive the same treatment. Our findings show that the transfer of individual shared and unshared knowledge is considerably high in all group conditions. Group composition does not significantly affect knowledge transfer and final group-outcome products; however, it may induce learners to experience different affective states. The results are essential for practitioners who intend to apply the RKB in their classroom to determine the appropriate group settings.