ABSTRACT
This chapter discusses the problem posed by confounds and critically evaluate organizational experiments as a way of overcoming them. Then it explains the implication of this approach for organizations wishing to make sound inferences about which management techniques lead to high performance. A confound occurs whenever a cause of an effect is correlated with another cause of that same effect. Thus, the effect on organizational performance of a cause can be confounded by any other cause of organizational performance. Organizational experiments that feature control groups are a way to avoid confounds. Such experiments are more feasible in large organizations than small, because large organizations are more likely to possess the standardization and comparable subunits. However, organizational experiments can lead to serious downward bias in estimates of the effectiveness of management techniques. Those large organizations that seek to avoid confounding error through organizational experiments may incur bias, leading to conservatism and failure to adopt effective techniques.
