An Experimental Test of Sabotage in Tournaments

We use a real-effort task to investigate the responsiveness of both sabotage and performance in a tournament to: (1) changes in the payoff structure of the tournament, and (2) changes in the identity of competitors over a series of tournaments (rematching versus constant pairings). Constant pairings shows significantly lower performance than rematching because of weak performance by low-ability participants. Constant pairings also depresses the rate at which participants choose sabotage, but causes higher sabotage levels given that the sabotage option is selected. Finally, sabotage is used far less effectively in the constant-pairings than it is in the rematching condition.