120 likes | 465 Views
Components of Source Credibility. Michael H. Birnbaum Fullerton, California, USA. Outline. I will review studies in which people aggregate evidence from sources A table of analogies can be used to link different content domains
E N D
Components of Source Credibility Michael H. Birnbaum Fullerton, California, USA
Outline • I will review studies in which people aggregate evidence from sources • A table of analogies can be used to link different content domains • Useful to distinguish several factors describing source: expertise, bias, liking • Also useful to consider “bias” of the judge, which I call point of view.
Averaging Model A, B, and C are three sources of information. When a source’s message is not presented, its weight is assumed to be zero. The initial impression is represented by w0 and s0.
Separating Weight and Value: the Zen of Weights • To estimate the weight of factor A we do NOT examine the effect of A (that confounds weight and value). • To estimate the weight of A, we examine the effects of B and C. • To estimate the weight of A, we leave it out.
Source Expertise and Bias • Expertise refers to the validity of a source of information. An “expert” source is one who has skill and knowledge to know the right answer. • Bias refers to tendency for a source to over or underestimate the “true” value. • Point of View refers to the judge’s bias, produced by factors that provide asymmetric losses to the judge.
Some Results • We can reject additive model for all domains in favor of averaging model. • Scholars are usually more amazed by the predictions of the additive regression model than they are by the data. • Expertise magnifies the effect of bias of a source; consistent with scale-adjustment model.
Point of View • Instructions to identify with the buyer, seller, or “fair” judge produce changes in rank order. • Changing preference orders consistent with configural weight model and assumption of scale convergence. • 1979 data refute the “loss aversion” “endowment” model of Tversky & Kahneman (1991). According to that model, ratio of selling to buying price should be constant.
Base Rate and Source Neglect • Data obtained within-subjects do not support the claim of “neglect” of base rates or source credibility. • Between-subjects, it has been shown that 9 is a significantly “bigger” number than 221. • I remain skeptical about findings from between-Ss designs.
Cumulative Prospect Theory • Results with judged prices of gambles refute cumulative prospect theory combined with the theory of “loss aversion.” • Choice data show 12 “paradoxes” of CPT which indicate that this model is not descriptive of choices between gambles.
Summary • Data accumulated over the last 25 years have favored Averaging over Additive models, consistent with claims by N. Anderson. • Inconsistent with Anderson, however, data also show violations of parallel-averaging or differentially weighted models in favor of configural weighting. • Effect of bias of a source is amplified by expertise, consistent with scale-adjustment model. • Data violate CPT account of buying and selling prices; instead, rank orders change, consistent with configural weighting model.