Matching with indifferences: A comparison of algorithms in the context of course allocation

We evaluate six one- and eight two-sided matching mechanisms with preferences based on a collection of 28 field data sets. Although important properties of matching mechanisms such as strategy-proofness or Pareto efficiency can be shown by formal proofs, the size, the average rank, and the popularity of matchings ask for an empirical evaluation. We introduce different metrics to compare the results. The study shows trade-offs between various design desiderata, which are relevant in the field. The results provide guidelines for the selection of matching mechanisms.