Abstract
Although most previous studies assume that the anchoring effect is elicited through numerical processes, a recently proposed semantic model (Mussweiler & Strack, 1999a) assumes that it occurs due to specific pieces of knowledge activated by the anchor. To test that model, this study examines the magnitude of the anchoring effect when participants have little knowledge about the inferential target. They estimated the height of a well-known personality (i.e., Tokugawa Ieyasu) or a little-known personality (i.e., Tokugawa Ietsuna) after an anchor was presented. The semantic model predicts that the anchoring effect should be smaller for the little-known personality. However, the prediction was not confirmed in this study. An analysis in terms of a skew index (Epley & Gilovich, 2006) suggests that the anchoring effect is produced by numerical processes when the participants have little knowledge about the inferential target.
