Illustrating the Three Models

These three contrasting positions constitute different “lenses” for viewing leadership. The differences among them can be illustrated by considering a hypothetical experiment. On one hand, we have an array of people who view themselves or are viewed by others as candidates for a leadership position. On the other hand, we have a set of situations, each requiring the exercise of leadership. These situations could vary markedly in the magnitude and nature of the challenges they pose, including the culture in which the challenge is embedded, the values of those whose energies are to be tapped, and so forth. The people could also vary in such things as their aptitudes, abilities, and other personality dispositions.

Now let us imagine the perfect experiment in which we could observe each person in each situation and we could judge their effectiveness in each. Thus, we could represent the results of our experiment as a matrix, with people represented by rows and situations as columns. Aggregating by rows and then by columns would enable us to test the three positions we have described. If leadership is a generalized trait, we would observe most of the variance in effectiveness occurring among the row means. There would be large differences in the overall effectiveness of different people, meaning that one could predict a person's ability to demonstrate leadership in one situation from his or her performance in a totally different situation. On the other hand, the situational point of view would lead us to expect large differences among the column means. Some challenges would enable all people to shine as leaders whereas others would defy the skills of all.

Finally, the contingency model would be consistent with those observations within the matrix that could not be predicted from either the person's mean score (reflecting a general leadership trait) or the situation's mean score. In other words, the person's success or failure would be a surprise, not explicable by either the pure situational or pure dispositional approach but potentially explained by a contingency theory.

Where then should we look to understand effective leadership—to the person, to the situation, or to both in interaction with one another? All three models have some explanatory value but we shall argue that relative potency depends on the nature of the problem one seeks to answer. In our hypothetical experiment, if the situations are highly similar to one another (and the persons are different), performance in one is likely to be predictive of performance in another. Experiments in which managers have been rotated from highly performing units to poorly performing ones doing the same kind of work shows that they tend to carry their level of performance with them. Personal qualities tend to transcend situational differences when situations are highly similar.

On the other hand, if the situations are highly dissimilar and the persons very similar, we make the opposite prediction. Situational effects will dominate the effects of personal qualities. These are precisely the kinds of situations that Pfeffer cites to support his situational view. Studying cities and towns and athletic teams over substantial time periods, he finds little or no evidence that changes in policies, practices, or achievements are attributable to changes in leadership. Pfeffer concludes that, “If one cannot observe differences when leaders change, then what does it matter who occupies the position or how they behave.” But mayors, athletic coaches, and, indeed, most leaders must pass through a strict screening and selection process that greatly limits the variability among those who are finally selected. If the persons are carbon copies of one another, situation effects should be the residual source of variance and leadership effects should disappear.

Leadership effects should also disappear when leaders have minimum discretion in the conduct of their jobs. There are some leadership roles in which leaders' choices are highly constrained by institutional traditions. The leader is, to use Pfeffer's term, an actor, with the script largely written by others or dictated by the organizational context.

I once had a colleague who had a mechanical model of an organization. It was a set of gears assembled in the shape of a pyramid. Turning the gear at the top of the pyramid about 10 degrees produced rapid 360-degree rotations at the pyramid's base. Pfeffer's analysis envisages a drastically different gearing—one in which turning the uppermost gear has little or no effect on the behavior of other gears!

But what about the contingency model? When is it most likely to represent the blueprint for thinking about and acting on leadership? The answer is clear: Wherever there is diversity among leaders and leadership candidates, and whenever there is diversity and fluidity in the kinds of challenges facing leaders, matching of leaders and environments will represent the key to effectiveness. This is precisely the environment that will be experienced by leaders in the 21st century.

In the present rapidly changing and complex world, the heroic model is no longer adequate to guide policies for selecting and training leaders. Similarly, the situational model, while appropriately drawing our attention to the “luck of the draw” in attributions of leadership, gives us no levers by which the quality of leadership may be enhanced. Matching of leadership qualities and behaviors to situational demands has been and will continue to be necessary for optimizing organizational effectiveness. To use an historical example with clear parallels to the present day, Patton and Gandhi may both have had the personal qualities necessary for the immense challenges they faced, but it is very unlikely each could have succeeded in the other's role.