10.2. DIFFICULTIES WITH EMPLOYEE APPRAISAL

When a supervisor appraises a subordinate, the process of appraisal can be analyzed as follows. First, the supervisor must have observed some performances. However, such observations in the case of R&D personnel are unlikely to be sufficiently coherent to be valid. If the supervisor were to observe a simple operation, he might be able to judge it. But R&D work is complex, and doing any one thing well is unlikely to provide a clue to the total performance. Thus, rather than observe an individual's specific performance, the supervisor is much more likely to observe large chunks of performance, such as the presentation of a research plan or the completion of a project. Usually these are products of groups rather than individuals. It then becomes difficult to know how much the particular scientist has contributed to the group product.

Second, the observations must be integrated into some sort of "schema." Unfortunately, there are several biases in the formation of such schemata. For example, research has shown that first impressions are extremely important. If the scientist has a good reputation, many acts that are ambiguous will be evaluated positively. Also, recent events tend to be given more weight in the formation of such schemata than events that occurred during the middle of the period of observation.

The fact that negative events are given more weight in such judgments than positive events creates a further bias. If the supervisor has observed ten events, and eight are positive and two are negative, the negative ones will be given more weight because they "stand out" as "figures" against the "background" of the eight positive events. This is because in our own lives we generally encounter few negative events, but when we do they are major negatives (e.g., loss of loved ones). On the other hand, although we encounter mostly positive events, they are seldom major positive events (e.g., getting married, winning a million dollars), and so we become especially vigilant about the negatives. Furthermore, a manager who overlooks the negative may be far sorrier than a manager who overlooks the positive. After all, if the manager does nothing about a major mistake, top management will blame the manager; but the manager is not likely to be blamed for failing to praise good performance.

Once the "schema" has been formed, regardless of the observations on which it is based, and it has been distorted by various biases, there is the problem of remembering it long enough to register it on paper. Here we have additional biases influencing what happens. Humans tend to remember positive events better than negative ones. Also, we are helped in our memories by previous schemata, such as stereotypes. That means we tend to distort what we remember in order to make it consistent with our stereotypes.

When making such judgments we often make major mistakes. For example, we generally do not give proper weight to the "base rates" of events. Suppose a department has a 15 percent failure rate for projects that are carried out. When judging a particular failure we usually do not take this into account.

Employee appraisal used in salary administration is affected by numerous other problems such as the "halo effect" (because an employee is good at one thing, he or she is seen as good at a lot of other things) and the "leniency effect" (everybody is first rate). Appraisal ratings tend to correlate with job difficulty, age, pay, and seniority, and they tend to increase each year. Such correlations suggest that they are contaminated by other factors.

In order to overcome these difficulties, psychologists have developed a number of strategies. These include forcing comparisons among subordinates (in a paired comparison format of N employees the supervisor makes N(N − 1)/2 comparisons). This may not really solve the problem because if all of them are excellent they should all get a high rating. Another approach is to ask supervisors to check specific statements using behaviorally anchored scales, or ask them to make forced choices between two statements describing the employee that are of equal social desirability but only one of which correlates highly with performance. In this way, only the selection of the "valid" alternative will ensure that the employee gets a high rating. These matters are beyond the scope of this book and are mentioned here only to sensitize managers to some of the methods that the personnel department can use that are slight improvements over the simpler rating scales.

There is a good deal of research that shows that observers attribute the other's behavior to internal factors (ability, attitude, personality, effort), while actors attribute their behavior to external factors (the situation, pressure from others, perceived consequences in the particular environment). Such biases are due, in part, to what observers and actors are looking at. The observer looks at the actor and naturally attributes the actor's behavior to internal causes. The actor looks at the environment and also knows that on other occasions he has acted differently. So, it is the particular environment on this occasion that is seen as determining the action.

The attributions we make give meaning to the behavior. If a failure is due to the difficulty of the task, it can be excused; if it is due to laziness, it cannot. People also tend to use their own behavior as a measure of how "typical" some action of the actor is. They underuse base rates (see above) and they see the actor as more responsible for acts leading to rewards than for acts that prevent failures. Furthermore, if they like the actor, they will attribute good actions to the person and bad actions to the environment; but if they dislike the actor, good actions are attributed to the environment and bad actions to the person.

If you keep these biases in attribution in mind, you might be able to improve your judgments. But in any case, a discussion with your subordinates of the way you look at their actions can increase the convergence between your attributions and the attributions made by your subordinates. If you two agree on the attributions, the subordinates will feel that the evaluation has been fair.

The performance appraisal, as usually carried out by an organization's management, is very often incompatible with the needs of research personnel. For most of the research and development personnel, these performance appraisals do more harm than good. Wilson (1994) proposes a two-tier approach. A management performance appraisal is normally conducted in most organizations, augmented by a professional development appraisal conducted by a highly respected mentor not directly involved in the management structure. The performance appraisal conducted by the mentor would focus on professional growth and contribution to the individual's field.