Despite the conceptual premise of Kerr and Jermier's (1978) substitutes for leadership model, there has been little evidence of empirical support when substitutes for leadership are modeled as interactions with leader behaviors. In this study we attempted to show that one reason that
significant interaction effects have not been found in prior research is the restricted range of the sample. We therefore investigated the interaction between substitutes for leadership and leader behaviors using a broad-based sampling design based on Mintzberg's (1993) model. We found support
for interactions between the substitutes for leadership and leader behaviors. The conclusion we reached is not that the original model of Kerr and Jermier is wrong, but rather that the methods used to test leadership interactions have not been appropriate. In addition, we found evidence in
this research that supports aspects of Mintzberg's (1993) Structure in Fives model.