Reinforcing Punishment and Extinguishing Reward

Abstract

Mawhinney has recently examined the relationship between OBM and SPC technologies. His analysis indicated that a common factor fundamental to both technologies is the capacity to discriminate between random and assignable causes of variation in system performance. This paper expands on the fundamental importance of this factor, for OBM in particular, by examining the deleterious consequences of failure to distinguish between random and non-random changes in human performance. The analysis draws heavily from a recent field experiment by two of the authors which demonstrated that failure to discriminate between random and assignable causes of variation in subordinate performance would lead supervisors, over time, to substantially decrease rewards and increase punishments in the mistaken belief that the latter were more effective in shaping their subordinates' performance. The implications of these effects for OBM are discussed and linked to several of both Deming's and Mawhinney's prescriptions. In this context, particular attention is given to supervisory training and to some simple, yet practical, methods of avoiding misinterpretation of causality in human performance data.