Abstract
It is first argued that the across-octant conclusions that are usually drawn from
F. E. Fiedler's (1967)
contingency model of leadership may be problematic, because virtually all its tests have used only within-octant research designs. Across-octant comparisons were therefore made by analyzing data from 1,282 groups used in previous research. Using nonparametric and parametric meta-analytic procedures, the authors found that the drawing of across-octant conclusions was not supported. Further analyses also showed unequal mean performance levels in octants where levels were supposed to be equal. The contingency model's predictions about the declining performance of leaders by octant, however, were supported in comparison with the predictions of three rival models. The contingency model may therefore need some modification. Implications and directions for future research are briefly discussed.
