Self-generated anchoring
Even the anchor you invent yourself drags your final answer toward it.
What it means
Anchoring that occurs when the starting value is one the person generates internally rather than receiving from outside — typically a known reference point from which they adjust toward the unknown answer. In these cases adjustment really does seem to be the operative process, and it stops too soon because people halt as soon as they reach a value within the plausible range, leaving the estimate biased toward the self-generated anchor. This distinguishes it from externally provided anchors, where priming and selective accessibility play a larger role, and it supports a dual account in which anchoring has more than one mechanism. The insufficiency of adjustment is worsened by cognitive load, time pressure, and low motivation, and reduced when people are prompted to consider reasons the true value lies further away. It matters for everyday estimation, where so many guesses begin from a fact we know and creep too little toward the fact we don't.
Examples
To estimate when George Washington was elected president, Americans start from the known date 1776 and adjust upward — but stop short, landing too early.
Asked how long a new report will take, an analyst starts from the last similar one — three days — and nudges upward, landing at four when the unfamiliar data will really take a fortnight.
Guessing a neighbouring city's population, you begin from your own town's figure and adjust down a little; the tweak is too timid, so the estimate stays anchored near home.
First described in Epley & Gilovich (2001, 2006).