Leadership Research Summary:
• Current leadership rating scales emphasize assessing behaviors rather than other characteristics. Yet an examination of their content, as well as the current research literature, suggests that they tap much more.
• In the interest of increasing the utility of leadership ratings, a relatively narrow conception of “behavior” is advocated, and leadership-behavior is contrasted with leadership-perception and leadership-effectiveness.
• Recent discoveries in social-cognition regarding factors that influence encoding and retrieval of behavior episodes are reviewed. They suggest a number of steps that might be taken to increase the accuracy of leadership ratings as measures of leaders’ actual behaviors.
Leadership Research Implications and Findings:
• The work on encoding and Tulving’s (1983) principle of encoding specificity suggestthe importance of providing relevant retrieval cues, and of having subjects generatetheir own cues. If subordinates encode behaviors as evidence of particular traits, orinto particular behavioral categories, those encodings will serve as effective retrieval cues for the behaviors themselves (Uleman, 1987).
• So in addition to a general instructionto mentally recreate the context of the behaviors (the meeting rooms, the memos, the184 LEADERSHIP QUARTERLY Vol. 2 No. 3 1991topics, etc.), subordinates could be asked to generate (and could be provided with)categories that are likely to tap the relevant behavioralencodings as well as organizingframeworks.
• An underlying premise for most of these suggestions is that we want observers torecall specific behaviors and episodes, not to make summary ratings or attempt toestimate the frequency of classes of behaviors. Too much is known about the inaccuracyof such summaries and estimates to accord them high validity (see above and Nisbett& Ross, 1980). So anything that requests and can assist recall of particular episodesshould be used.
• Reducing the delay between observing the episode and recalling it will also increaserecall accuracy. This suggests a time-sampling strategy in which observers are askedat the end of the day to recall as many relevant episodes as possible from that sameday. A larger sample of episodes could be obtained by repeating this procedure on anumber of days. (See Bass, 1990,for discussions of recent applications of this technique.)Psychometric considerations suggest additional steps that should improve accuracyof recall.
• First, use severalsubordinates to recall the behaviors of any particular leader,and combine these reports in some way that gives greater weight to episodes recalledby several subordinates. Second, increase the frequency of time samples. And third, aggregate the “items” over multiple observations of the same kind of behavior.
• Whether or not these suggestions are practical depends on the settings in which weand the constraints they impose. Only future research can determine how useful or important they are for obtaining more accurate ratings of leaders’ behaviors