Anchors as Midpoints: It’s Not the Size of the Adjustment That Counts, It’s The Direction

People’s estimates of unknown quantities are tethered to values they have previously considered, a phenomenon known as anchoring (Tversky & Kahneman, 1974). Previous research has attributed anchoring effects to a process of insufficient adjustment from anchors (e.g. Epley & Gilovich, 2006; Tversky & Kahneman). In this report, we propose an alternative explanation for anchoring effects, “anchors as midpoints”, which instead focuses on the direction of adjustment. Specifically, people treat anchors as if they were a mental midpoint of the number line. Consequently, compared to what their unanchored estimates would suggest, people are more likely make estimates that are higher than high anchors, and lower than low anchors. In our paper, we test this prediction, as well as three other predictions: First, the original prediction will be most evident when people’s intuitions about the magnitude of the stimulus match the magnitude of the anchor. Second, anchoring effects will be stronger when people can adjust from an anchor in either direction, compared to when an arbitrary direction of adjustment is imposed. Finally, people will be less likely to adjust from arbitrary anchors in the direction of the correct answer than unanchored estimates would suggest.