Behavioral anchors as a source of bias in rating

Article Abstract:

Behavioral anchors may affect the way that raters process information about ratees, and may in some cases be a source of bias in rating. This study tested the hypothesis that the presence of behavioral anchors that closely matched behaviors actually observed by raters would bias performance ratings. Subjects (N = 180) viewed videotaped lectures and rated them, using scales that contained examples of either good or bad performance that had actually occurred on the tapes, but that were not representative of the ratee's overall performance. One half of the subjects read the scales before viewing the lectures; the remaining subjects read the scales only after viewing the lectures. There was a significant scale effect, but no Scale x Order interaction; ratings were biased in the direction of unrepresentative anchors. These results suggest that behavioral anchors can be a source of bias in ratings and they may lead to biased recall, but they probably do not bias the observation and encoding of ratee behavior. Our results suggest that behaviorally anchored scales are not necessarily more objective or less prone to bias than are scales without behavioral anchors. (Reprinted by permission of the publisher.)

Analysis, Employee performance appraisals, Performance appraisals

---

Detecting infrequent deception

Article Abstract:

Recent proposals for using the polygraph and similar devices in routine screening have been aimed at detecting deception in situations sometimes characterized by low base rates. Equations are developed that show that extraordinarily high levels of accuracy would be needed to detect infrequent deception. In this context, the debate over the accuracy of these methods is irrelevant; the accuracy needed to detect infrequent deception far exceeds the levels claimed by the most enthusiastic proponents of these detection techniques. The limits on the use of any particular test of deception can be determined by considering the base rate for deception and the proportion of the nondeceptive population that fails the test. When the base rate is less than .10, these limits are extremely restrictive. (Reprinted by permission of the publisher.)

Human resource management, Deception, Lie detectors and detection, Lie detector tests

---

Under what conditions are observed intercorrelations greater or smaller than true intercorrelations?

Article Abstract:

Two possible explanations were examined for finding that observed correlations among dimensions are sometimes greater than and sometimes less than true intercorrelations: It can be explained by (a) variation in true halo or (b) in terms of individual differences in perceived similarities among dimensions. Subjects (N=145) viewed videotapes that exhibited either high or low true halo, rated them, and rated similarities among dimensions. True halo level accounted for significant variance in rating intercorrelation, but perceived similarity did not. Results suggest that neither explanation is adequate to account for the finding that observed intercorrelations can either overestimate or underestimate true intercorrelations. (Reprinted by permission of the publisher.)

Psychological aspects, Performance, Measurement, Statistical hypothesis testing, Evaluation

Similar abstracts:

• Abstracts: A laboratory experiment testing the antecedents of leader cognitions. Cognitive-affective stress propensity: a field study
• Abstracts: The role of the World Bank in international health: renewed commitment and partnership. Role conflict? The World Bank and the world's health
• Abstracts: Expedited removal. The assault on habeas corpus in immigration law. Prisoners of the INS
• Abstracts: Who's afraid of life extension? Is there an 'anti-aging' medicine?
• Abstracts: Professional behavior during conditions of extreme community turmoil: the case of the removal of settlements from Sinai

This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.