On the Futility of Aggregating Individual Learning Curves

Abstract
It is argued that a mean or aggregate learning curve yields little information about the form of the individual curves from which it arose. Hypothetical examples and empirical evidence in support of the argument are discussed. Also cited is a mathematical argument showing that aggregation of certain forms of individual curves systematically results in an aggregate curve of a different functional form. Finally, it is argued that drawing conclusions about the effects of different treatments on the parameters of the learning curve is risky unless the form of the individual curves is known.