Things like production rates, sales statistics, athletic performance, etc., in many situations, vary irregularly around some average level. In such cases if one were trying to predict what the next, as yet uncrystallized, outcome in the performance series would be, settling on a result in the region of the average would be a good bet (this assumes that systematic variations such as seasonal factors are taken into account). A particularly good performance is likely to be followed by a lower one, that is, nearer to the average, while a very poor performance will be improved upon.

Imagine the supervisor of a manager in charge of a sales team, on seeing weak sales figures for the last quarter, gives his junior manager “a kick in the pants” for this “lack of success”. Let’s say that when the next quarterly results are tabulated sales are up and now just narrowly exceed the yearly average. Unfortunately, the senior manager may now feel some self-satisfaction in that “taking a firm hand” with the subordinate has resulted in the ‘turnaround’: punishing his or her lower-ranking colleague’s apparently substandard performance led to success!

Regrettably people frequently exhibit tendencies to make dubious attributions (see: Fundamental attribution error).