Key points are not available for this paper at this time.
Mathematical models of source credibility were tested in five experiments in which judges estimated the value of hypothetical used cars based on blue book value and/or estimates provided by sources who examined the cars. The sources varied in mechanical expertise and in bias; they were described as friends of the buyer or seller of the car or as neutral. Individuals judged the highest price the buyer should pay, the lowest price the seller should accept, and the true value (fair price) of the car. Data indicated that expertise amplifies the effect of the source's bias. This effect is predicted by a scale-adjustment model, in which the source's bias shifts the scale value of the source's estimate. The weight of an estimate depends chiefly on the source's expertise. The weight of an estimate also depends configurally on the other estimates: Judges instructed to take the buyer's point of view give greater weight to the lower estimate, whereas judges who identify with the seller place greater weight on the high estimate. Simple premises about human judgment give a good account of the data.
Birnbaum et al. (Mon,) studied this question.