Decision making factors in estimating SDy in utility analysis: effects of frame, stimulus salience, and anchoring
The present research is concerned with the cognitive dimensions of procedures for estimating the standard deviation of job performance (SDy) in utility analysis. The overall goal was to integrate SDy estimation with advances in the area of decision making, and with advances in the social psychological study of person perception. Three variables were considered: problem frame, that is, whether the estimation task is posed in terms of the gain or loss of employees, stimulus salience, that is, the level of detail regarding the employee whose worth is being evaluated, and anchor values from previous estimates provided to judges as the starting point in forming their judgments.
In previous research by Shetzer & Bobko (1987), estimates of overall worth obtained under negative frames were significantly greater than estimates obtained under positive frames. Experiment 1 tested whether the effect of frame would be as evident with high salience stimuli as with the traditional low salience scenario. A significant effect due to salience was found and the study concluded that salience is a primary variable. Experiment 2 examined the relation between the effect of framing and anchor values provided to subjects as the starting point in estimation. Experiment 2 found a significant effect for anchoring but no effect due to frame, suggesting that subjects' estimates are anchored on initial values. Anchoring was also found to reduce the variability of estimation. The reduction in the variability of negatively framed estimates appears to be relatively greater when anchors are provided than is the reduction in the variability of positively framed estimates. These findings confirm earlier research concerning the efficacy of the sequential feedback procedure for reducing within-cell variance.
The results of the two experiments suggest that the effect of the problem frame is not as important a variable in SDy estimation as are salience and anchoring. This conclusion should be welcomed by utility analysts, since it suggests that the estimation procedure can be made more precise by providing judges with the maximum amount of relevant information thus mitigating the impact of more peripheral variables, such as how the problem is framed.