Anchoring Persistence Error

The first piece of information a person hears has a way of outlasting its usefulness. Even once new facts show it was wrong, decisions keep tracking back to it.

This error names the bias by which initial information keeps disproportionately shaping every judgment and estimate that follows it. Additional evidence rarely dislodges it — the decision consistently under-adjusts from the original anchor.

A dealer quotes $40,000. Negotiating down to a "big discount" of $36,000 feels satisfying — anchored entirely to that opening figure. A market search showing comparable cars selling for $31,000 would have set a far lower starting point, and a far lower final offer.

An acquiring firm's analyst opens a discounted-cash-flow model with a precedent transaction multiple of 12x EBITDA, drawn from a peak-market deal. Comparables analysis later yields a central estimate of 8.5x, yet the final fairness opinion still anchors near 10.8x — a classic under-adjustment from that initial reference point. The endowment-induced anchor from the precedent transaction propagates through the evidence accumulator, suppressing the posterior variance collapse that should occur around the unanchored central estimate and biasing the final number toward the inflated multiple.

A first clue sets the starting point for the thinking that follows. Later information gets ignored, or adjusted for too little, because that starting point already feels right.

The initial anchor imposes a reference constraint on the decision representation, with early evidence assigned more weight than anything that arrives later. That asymmetry — mediated by the initial reference node and the downstream update pathways — limits how much correction the estimate can actually undergo.

Weighing the first idea against other numbers or opinions before deciding helps expose its pull. Asking how the judgment would look without that first suggestion is a useful check.

Eliciting and recording an unanchored estimate before any reference value appears, then applying calibrated debiasing informed by external benchmarks, keeps the anchor from ever taking hold. Procedural checks that reweight incoming evidence further counteract initial reference dominance.

Insufficient adjustment; Overreliance on initial source; Delayed correction

An adversarial actor can deliberately introduce an extreme or fabricated initial figure — a salary offer, a negotiation price, a risk estimate, or a casualty projection — knowing that even after partial correction the final judgment will remain biased toward that anchor. This technique is routinely weaponized in high-stakes negotiations, auction design, and disinformation campaigns where the first number to reach a target audience sets a durable reference-point constraint that subsequent rebuttals struggle to fully displace.

Before any reference value is introduced, practitioners should elicit and commit to an unanchored prior estimate in writing, making the baseline explicit and harder to silently revise. Structured adversarial review — assigning a team member to argue from an alternative anchor — forces explicit reweighting of downstream evidence against the initial reference node. Calibration training with historical base rates and external benchmarks builds the habit of reference-class calibration, reducing default weighting assigned to the first datum encountered.