Blackwell Reference Online

Extract

Regression toward the mean is a statistical phenomenon which occurs when two quantities are related but imperfectly so ( Campbell and Kenny, 1999 ). The quantities can be repeated observations of the same variable – for example, performance at time t and performance at time t + 1 – or two distinct variables such as innovation and performance. Regression toward the mean is the tendency for extreme values on the first quantity to be associated with values of the second quantity which are closer to the mean rather than with values that are farther away. Regression toward the mean often arises because extreme quantities are the result of chance events that are unlikely to repeat themselves. In such circumstances those who mistakenly attribute the tendency for extreme quantities to be followed by less extreme quantities to complicated causal theories commit the so‐called regression fallacy ( Tversky and Kahneman, 1974 ). Regression toward the mean, however, is not solely due to chance or measurement error. An implication of the fact that regression toward the mean occurs when two quantities have a correlation of less than one is that any factor that weakens the correlation between two variables may be interpreted as facilitating regression toward the mean ( Campbell and Kenny, 1999 ). In line with this logic, organizational researchers have proposed and shown that executives may ... log in or subscribe to read full text