People's judgments are often inordinately influenced by the first information that comes to mind. Judgments of others' perceptions are unduly tied to one's own (Gilovich & Savitsky, 1999; Savitsky, Epley, & Gilovich, 2001; Keysar & Barr, 2002), impressions of others' personalities overly influenced by initial dispositional inferences (Gilbert, 2002) and answers to general knowledge questions anchored on irrelevant information considered early in the processing stream (Chapman & Johnson, 2002; Wilson, Brekke, & Centerbar, 2002). These effects, and many like them, have traditionally been explained as insufficient adjustment from an initial anchor or starting point, using what Tversky and Kahneman (1974) called the "anchoring and adjustment" heuristic. In short, people make judgments under uncertainty of all kinds by starting with (or anchoring on) information that is presented to their attention or that comes quickly and automatically to mind, and then serially adjusting their initial impression until an acceptable judgment is reached. Although originally used to explain numerical estimates, anchoring and adjustment has figured prominently in numerous theories of social judgment.
Recent evidence has cast doubt on the "anchoring and adjustment" heuristic. In particular, it has come to light that the experimental paradigm long used to investigate anchoring and adjustment does not, in fact, involve adjustment at all. These results, accompanied by failures to find conclusive evidence for a process of adjustment outside this paradigm, have led some researchers to reject the anchoring and adjustment heuristic altogether (Chapman & Johnson, 2002).
Research conducted with funding from a current NSF grant indicates that this conclusion is premature: The anchoring and adjustment heuristic is alive and well when one looks in the right places (Epley & Gilovich, 2001). This research indicates that people adjust in a serial fashion from anchors they automatically generate themselves as a value known to be close to the right answer but in need of tinkering. It appears, furthermore, that the processing of such "self-generated" anchors is common and pervasive in everyday life, making an accurate understanding of them critical to a broader understanding of social judgment.
The studies proposed here are designed to expand the work begun on the existing grant by: a) developing new methodologies for tracking the operation of true anchoring and adjustment; b) more clearly identifying the exact nature of anchors that stimulate adjustment and those that do not; c) expanding the scope of the anchoring and adjustment heuristic beyond questions of general knowledge and into social judgment more generally; and d) developing techniques for overcoming insufficient adjustment.
