Peer-Allocated Instant Response (PAIR): Computational Allocation of Peer Tutors in Learning Communities

This paper proposes a computational model for the allocation of fleeting peer tutors in a community of learners: a student's call for support is evaluated by the model in order to allocate the most appropriate peer tutor. Various authors have suggested peer tutoring as a favourable approach for confining the ever-growing workloads of teachers and tutors in online learning environments. The model's starting point is to serve two conflicting requirements: 1) the allocated peers should have sufficient knowledge to guarantee high quality support and 2) tutoring workload of peers should be fairly distributed over the student population. While the first criterion is likely to saddle a small number of very bright students with all the tutoring workload, the unconditional pursuit of a uniform workload distribution over the students is likely to allocate incompetent tutors. In both cases the peer support mechanism is doomed to failure. The paper identifies relevant variables and elaborates an allocation procedure that combines various filter types. The functioning of the allocation procedure is tested through a computer simulation program that has been developed to represent the student population, the students curriculum and the dynamics of tutor allocation. The current study demonstrates the feasibility of the self-allocating peer tutoring mechanism. The proposed model is sufficiently stable within a wide range of conditions. By introducing an overload tolerance parameter which stretches the fair workload distribution criteria, substantial improvements of the allocation success rate are effected. It is demonstrated that the allocation algorithm works best at large population sizes. The results show that the type of curriculum (collective route or individualised routes) has only little influence on the allocation mechanism.