Abstract
Field and Hineline have shown how pervasive and insidious is the tendency to make dispositional attributions, even among those who criticize the practice, and they identify a bias for models of contiguous causation as one reason for this tendency. They argue that order can be found at multiple scales of analysis and that in some cases a translation to a model of contiguous causation is impossible. I suggest that pragmatic considerations are sufficient to justify a particular scale of analysis and observe that behavioral principles are fundamentally extended in time. However, I argue that accounting for variance is the goal of science; when events at one level are indeed mediated by those at another, more of the variance can be accounted for by considering both, and there is no principled reason for considering only one