ABSTRACT
As we saw in the previous chapter, confounding, in which a figure is afflicted by a spurious relationship, can be controlled by experiments. However, as we also saw, experiments in organizations are prone to a bias of markedly understating the true effect of a cause on performance. In this chapter, we will propose another way to eliminate confounding: aggregation of data. This reduces confounding without incurring the downward bias from using organizational experiments. It is a novel approach, so it will be argued for at some length.
