You could then see whether, for any two players, there is a tendency to have more errors in high-leverage situations.(More generally, I think academic sports researchers could design studies that take more advantage of natural controls, and rely less on the regression model to try to do that.)