Regression to the mean: a confused concept

It is suggested that psychologists, like other people, are not entirely immune from the tendency to ‘blind themselves with science’. Statistics in particular provide a convenient mystique and thus an encouragement to use catch‐phrases in describing events we do not fully understand. One typical example is the phrase ‘regression to the mean’, often used to explain changes in extreme test score and usually regarded as a wholly statistical problem without psychological implications. This paper, however, by reference to physical and psychological data, attempts to indicate its primarily behavioural bases and to show why groups chosen for study on the basis of extreme test score will tend to regress towards the mean of the general population, while regression of extreme scores in groups not so selected will be towards the mean of their own specific population. There appear to be two main factors at work. (i) The effects of fleeting, inconstant personal alterations in response to the test (the so‐called ‘errors of measurement’ which are more complex than they sound). (2) The essentially non‐linear nature of most individual physical or psychological growth. To say without further qualification that test scores have changed ‘because of regression’ gives no greater information than to state that they have changed because they have changed. 1960 The British Psychological Society