There is ample evidence that incentive pay structures such as tournaments result in increased performance, but whether this is due to selection or increased individual effort is less clear. We show that empirical specification is the key. Misspecification masks individual effort and interprets it as selection. Looking at data on horse racing, we compare a pure selection model to the Lazear-Rosen tournament model. While both models organize the data, the tournament model does a better job, and it says that nearly two-thirds of the increased performance associated with higher prizes is due to increased individual effort. This estimate is very similar to estimates found in industrial field studies where performance pay is not structured as a tournament. We corroborate the horse-racing results by looking at dog racing. Dogs are not expected to respond as elastically as horses, and empirically they do not.