Anchoring revisited: the role of the comparative question

When people estimate a numeric value after judging whether it is larger or smaller than a high or low anchor value (comparative question), estimates are biased in the direction of the anchor. One explanation for this anchoring effect is that people selectively access knowledge consistent with the anchor value as part of a positive test strategy. Two studies (total N = 184) supported the alternative explanation that people access knowledge consistent with their own answer to the comparative question. Specifically, anchoring effects emerged when the answer to the comparative question was unexpected (lower than the low anchor or higher than the high anchor). For expected answers (lower than the high anchor or higher than the low anchor), however, anchoring effects were attenuated or reversed. The anchor value itself was almost never reported as an absolute estimate.