GIVE CREDIT FOR GOOD FORM. If the only dimension of performance we needed to be concerned with were the number of widgets produced each hour – without reference to how this was achieved (regarding waste of material impact upon fellow workers, being on time, etc.) – then no formal, performance-appraisal system would be needed.

However, rarely is this the case. Usually, physical conditions, external events, and the behavior of others (e.g. time required for the best team-building efforts to have effect) interact to determine outcomes. Therefore, adequate attention should be given to productive, work-relevant behaviors – not just immediate, physical output – on the part of each individual.

IDENTIFY WORK-RELEVANT BEHAVIORS VIA THE CRITICAL INCIDENT TECHNIQUE. Court decisions stress the need for appraisal criteria that are based upon explicit, job-analysis data, rather than general characteristics; such as, attitude toward people, resourcefulness, leadership, capacity for growth, and loyalty to the organization.

The Critical Incident Technique (CIT) produces quite explicit data. Those who are most knowledgeable about a job (supervisors, incumbents, peers, subordinates, clients, etc.) are asked to describe, independently, specific incidents of efffective and ineffective job behavior they have observed over the past 6-12 months. Next, they meet to classify those incidents that they agree are positively or negatively critical into different perforrmance dimensions and assign relative weights to these. These incident data can then be used to inform new job inductees, guide remedial action, and indicate the the types of behavioral documentation that will be required for supporting “star” and “inadequate performer”
nominations.

EXCLUDE THE MIDDLE – SOLICIT RATINGS FOR ONLY EXTREME PERFORMERS. In virtually any organization, you will obtain much better agreement as to who are the “stars,” and who are the “inadequate performers,” than you will as to who is “above average,” “average,” and “below average.” Yet, in most organizations, assignment of these middle-three ratings consumes a disproportionate amount of supervisory time and causes the most dissension and resentment among the troops.

The label “average” has a negative conotation to good employees, for we know that average performance in a superior organization is not the same thing as average performance in a mediocre one. Why not simply tell these middle people that they are valued menbers of the team and give them CIT data on how to be “stars” and avoid being “inadequate performers.”

USE MULTIPLE SOURCES AND, WHERE FEASIBLE, MULTIPLE RATERS FOR A SOURCE. Studies show that ratings from those closest to the performance dimension being rated – in terms of knowledge about the dimension and opportunity to observe behavior relative to it – are more valid than ratings from other sources. In addition, the pooling of ratings from several knowledgeable raters for the same source, when available, are preferable to one rater for that source, in terms of measurement reliability and inclusiveness. Of course, all “star” and “inadequate performer” ratings should be documented with CIT determined behaviors.

USE A COORDINATION PANEL TO PROCESS EVALUATIONS WHEN THERE IS MORE THAN ONE ORGANIZATIONAL UNIT. A Coordination Panel (comprising a representative from each organizational unit and chaired by a top-level officer) should be established to check on the adequacy of each unit’s CIT preparation, review inputs from the various units to assure interunit fairness, handle appeals from aggrieved personnel, and otherwise monitor the operation of the system. The panel also determines how to make trade-offs between goal achievement and goal difficulty.

SOLICIT FEEDBACK FROM RATEES. No supervisor or other rater is all-seeing and all-knowing. In a rating-review session, the supervisor should regard the overall rating of a ratee to be tentative until the ratee (previously given all CIT data about his or her job) has had the opportunity to input any additional critical-incident data not communicated by the supervisor or challenge any communicated data. Then, if necessary, the supervisor may need to confirm the validity of such data before continuing the performance review.

A more complete presentation will be found in my article: “Improving Performance Appraisal Systems,” NATIONAL PRODUCTIVITY REVIEW. Winter 1987-1988, Vol. 7, No. 1, pages 20-27.

William M. Fox
gryfox@bellsouth.net
Professor Emeritus, Management and Organizational Behavior
University of Florida
6605 SW 37th Way
Gainesville, FL 32608
(352) 376-9786