Effective metaheuristic assignment to improve equivalence in small experimental pedagogical groups

We propose a new metaheuristic algorithm to find “good” solutions for the assignment of small treatment-control groups, minimising the random resource. Using simulated cases, we achieved 100% groups with equivalence levels equal to or higher than those generated with the simple random assignment, complete random assignment and block random assignment designs. In addition, as a secondary objective to test the new algorithm, we found that short out-of-class essays implied that treatment group marks were 14% higher than in the control group.