An Experiment On Two-Person Social Choice

Official URL: https://risc01.sabanciuniv.edu/record=b3200000

This study empirically compares four different mechanisms recommended in the arbitrator assignment process: Alternate Shortlisting (ASL), Compromise Rule of k Names (CRk), Gradual Veto (GV), and Shortlisting Mechanism (SL). These four mechanisms differ according to the number of steps they contain, whether they include a first-mover, and whether they contain vetoes. Data were collected in a computer laboratory environment with four different z-Tree treatments in which 290 undergraduate students from Sabancı University participated. A comparison was made on notions such as Truthfulness, Pareto Optimality, Inequality, First Mover Advantage, and Equal Loss Principle. All performance benchmarks are tested with t-tests and OLS regressions with clustered standard errors by session. We found that the Gradual Veto mechanism slightly outperforms the sequential mechanisms regarding playing truthful behavior, but the same success is not valid in achieving an efficient outcome. SL and ASL are more successful for players staying in their upperhalf alternatives. The Gradual Veto mechanism creates a more equal environment for matched players. Among sequential mechanisms, ASL and CRk have lower First Mover Advantage than SL. Also, we found some significant relationship between participant characteristics and round outcomes.