Affect and managerial performance: a test of the sadder-but-wiser vs. happier-and-smarter hypotheses - includes appendix

Administrative Science Quarterly, June, 1993 by Barry M. Staw, Sigal G. Barsade

In-basket exercises have been shown to be relatively valid and reliable indicators of managerial decision making and administrative skills (Frederiksen, 1966; Thornton and Byham, 1982: 181-184). Such simulations may be especially useful when one of the variables of a theoretical test is likely to contaminate the measurement of performance. Several experimental studies have shown that rated positive affect and likability may lead to inflated performance ratings (Cardy and Dobbins, 1986; Krzystofiak, Cardy, and Newman, 1988; Smither, Collins, and Buda, 1989). Thus, to assess managerial decision making it may be advantageous to have objectively coded tasks such as in-basket exercises rather than in-situ performance ratings.

Assessing Interpersonal Performance

To assess interpersonal performance, we used a leaderless group discussion (LGD) exercise. LGD exercises are frequently used in managerial assessment centers and have been found to be both a valid and reliable measure of interpersonal skills and activity level (Bray and Grant, 1966; Thornton and Byham, 1982: 170-176). The LGD used in the M.B.A. assessment was a competitive exercise involving the allocation of limited resources. Each participant was asked to represent a subordinate's interest in getting a pay raise, as well as the best interests of the company. There were neither sufficient resources to meet all requested raises nor enough time to evaluate fully all the possibilities for allocating funds. Each participant not only had to persuade others of his or her claim on resources but also help move the group toward an overall allocation scheme that was acceptable to all group members. Thus, like many managerial tasks in which both distributive and integrative solutions are possible, participants were asked to work for their own as well as joint interests.

M.B.A. students participated in the LGD in groups of six. Participants were given ten minutes to study the case and then thirty minutes to reach a consensus decision on the allocation of pay raises for each of six candidates. The participants were ranked by their peers (the other five members of the group) on their overall performance and contribution to group effectiveness. They were also ranked by 12 personality assessment staff members who observed the group discussions. Staff ranked the following dimensions: extent of participation, quality of participation, politeness, leadership, criticality, social compliance, mastery of information, and task engagement. The mean staff reliability was .95; the mean peer reliability was .81 (see also John and Robins, 1991).

Other Measures

Managerial potential. M.B.A. students were also assessed on more global dimensions of management. As noted earlier, for each group of 12 participants there were six members of a managerial assessment staff (separate from the 12-member personality staff used in the assessment of positive affect or the LGD exercises). The managerial assessors observed the LGD exercise (but did not formally rate participants on it), read the participants' responses to the in-basket exercise (without formally coding the responses), and conducted in-depth interviews. At the end of all assessment activities, the managerial staff met in three-person groups to discuss each participant's performance and, in turn, rated the participants on the following 15 dimensions: initiative, stress tolerance, energy, leadership, sensitivity to others, analytic thinking, decision making, creativity, factfinding, oral communication, written communication, control, planning and organizing, and delegation. The three-person assessment teams reached consensus ratings on each managerial dimension. If individual staff ratings differed by no more than 1 point (on a 5-point scale), the modal score was chosen as the consensus rating. If any of the staff ratings differed by more than 1 point, the assessors then discussed the participant and again rated that managerial dimension until consensus was reached. These 14 consensus ratings were averaged to create a managerial dimensions scale (|Alpha~ = .90). In addition, a single-item rating scale of the overall managerial potential of the participant was obtained from each assessment team.