A problem is that evaluators differ in the average numerical ratings they give to others. Because of what is known as the restriction of range problem, some people provide uniformly high, lenient ratings; others are inclined to give stingy, low ratings; and still others gravitate toward the center of the numeri- cal scale. In all cases, people who use a restricted range fail to make adequate distinctions. Sometimes the differences among raters are considerable, as was seen in a study of managers employed in numerous organizations—in part because of differences in their personality. Individuals who have agreeable personalities tend to be lenient in their ratings of others, while those who are highly conscientious tend to be harsher. In a meta-analysis of 25 studies, John Georgesen and Monica Harris (1998) also found that people who are in power, compared to those who are not in power, consis- tently give lower performance ratings to others who are in subordinate positions.
In addition to problems with rating error and bias, supervisors may intention- ally distort their evaluations, depending on their objectives. Evaluating students in a human management resources course, for example, rat- ers gave higher ratings overall when their goal was to encourage group harmony, to ensure fairness and accuracy, or to motivate those they were rating than when the purpose was merely to help members to identify strengths and weaknesses. In the workplace, performance evaluation is not just a mea- surement process but one that serves social and communication purposes as well.