Studying the best can mislead: selection bias in expertise research

What we think we know about how people reach exceptional performance could be distorted by who gets studied. Studying elite performers has been highly informative for understanding the cognitive mechanisms of expertise but might be less suitable for explaining how expertise develops. Elite samples are filtered on the very success researchers want to explain. Predictors that appear beneficial early in development often weaken or even reverse among world-class adults. These findings are commonly interpreted as evidence that the route to excellence changes at higher performance levels. Here we demonstrate that they can also arise from selection alone. Outstanding adult performance is rare, depends on many factors, and can be reached in more than one way. Focusing exclusively on those who succeed can create misleading trade-offs, make statistical relationships look smaller, and make weak positive and weak negative effects look the same. Simulations and longitudinal data from chess and athletics illustrate these patterns and show that they depend on who remains in the sample and how narrowly elite performance is defined. We offer practical principles for studying elite performers without mistaking associations within selected samples for evidence about how excellence develops.