Home › What the research says

Anchoring, and Why You Should Set Your Number First

The wheel of fortune

Amos Tversky and Daniel Kahneman described anchoring in Science in 1974 as part of their account of judgment under uncertainty. People make estimates by starting from an initial value and adjusting, and adjustments are typically insufficient. Their demonstration was deliberately blunt. For each quantity, a number between 0 and 100 was determined by spinning a wheel of fortune in the subjects' presence, then subjects estimated a real quantity. The median estimates of the percentage of African countries in the United Nations were 25 and 45 for groups that received 10 and 65 as starting points. A visibly random number moved answers by twenty percentage points.

Anchors that are obviously meaningless still work

Dan Ariely, George Loewenstein, and Drazen Prelec pushed this further in the Quarterly Journal of Economics in 2003. Across six experiments, initial valuations of familiar products and simple hedonic experiences were strongly influenced by arbitrary anchors, in some cases derived from the participant's own Social Security number. The part that gives the paper its title is what happened next: subsequent valuations were still coherent with respect to real differences in quality or quantity. People were internally consistent about relative value while their absolute level had been set by a meaningless number. They called it coherent arbitrariness, stable demand curves without stable preferences.

Experts are anchored too, and say they are not

Gregory Northcraft and Margaret Neale tested this outside the lab in Organizational Behavior and Human Decision Processes in 1987. Students and real estate agents toured actual properties and made pricing decisions. All received identical information packets except for the listed asking price, which was manipulated. Estimates of value came out positively correlated with the asking price they had been given, consistent with anchoring and adjustment in an information-rich real-world setting. The detail people remember is that most agents reported that the asking price was not relevant to their judgment. Domain expertise did not prevent the effect, and introspection did not detect it, which is the combination that makes anchoring hard to defend against by willpower.

How well the finding holds up

This matters because a lot of famous psychology has not survived retesting. Richard Klein and a large group of coauthors ran the Many Labs replication project, retesting 13 classic and contemporary effects across 36 different samples with 6,344 participants, and reported that replicability depended more on the effect itself than on the sample or setting. Anchoring items were among those included, and they were among the effects that reproduced across samples rather than among the ones that failed. That is a reasonable basis for treating anchoring as durable, while remembering that reproducing in a lab task is not the same as measuring how much a store display moves your grocery bill.

The envelope change to make

You cannot decline to be anchored. What you can do is arrive with a number already in place, decided somewhere calm, so that the retailer's number has something to compete with rather than an empty field to fill. That is what a funded envelope is: a figure you set in advance, in a room with no price tags in it, attached to a specific category. Before a category where you know the anchors are aggressive, check the envelope balance first and treat it as the reference. Deciding what a thing is worth to you after you have seen what they are asking is deciding second, and second is the weaker position.

Sources

Every source below was retrieved and checked. Findings are reported as the source states them.

• Judgment under Uncertainty: Heuristics and Biases — Science, 1974
  People estimate by starting from an initial value and adjusting, and adjustments are typically insufficient; with starting points set by a wheel of fortune spun in the subjects' presence, median estimates of the percentage of African countries in the United Nations were 25 and 45 for groups given 10 and 65.
• Coherent Arbitrariness: Stable Demand Curves Without Stable Preferences — The Quarterly Journal of Economics, 2003
  In six experiments, initial valuations of familiar products and simple hedonic experiences were strongly influenced by arbitrary anchors, sometimes derived from a participant's Social Security number, while subsequent valuations remained coherent with respect to perceived quality or quantity differences.
• Experts, Amateurs, and Real Estate: An Anchoring-and-Adjustment Perspective on Property Pricing Decisions — Organizational Behavior and Human Decision Processes, 1987
  Students and real estate agents toured properties and priced them from identical information packets differing only in listed asking price; value estimates were positively correlated with the manipulated asking price, consistent with anchoring and adjustment in an information-rich real-world setting.
• Investigating Variation in Replicability: A Many Labs Replication Project — Social Psychology, 2014
  Thirteen classic and contemporary psychological effects, including anchoring items, were retested across 36 samples with 6,344 participants; results indicated that replicability depended more on the effect itself than on the sample or setting, with the anchoring effects among those that reproduced.

Common questions

Does knowing about anchoring protect you from it?

The evidence is not encouraging. In Northcraft and Neale's study, real estate agents valuing actual properties were anchored by the manipulated asking price while most reported that the asking price was not relevant to their judgment. Expertise in the domain did not prevent the effect and self-report did not detect it. That is why the useful defense is structural rather than mental: decide your number before you see theirs, and write it down somewhere you will actually look at the moment of the decision.

Is a struck-through original price an anchor?

It fits the pattern the research describes, which is that a presented number shifts subsequent valuation. But the specific studies here used wheels of fortune, Social Security digits, and real estate list prices, not retail markdown displays, so calling it an anchor is an inference rather than a measured result. The practical response is the same either way. If your own number for that category exists before you look, the comparison is against your number, not against theirs.

How is an envelope different from just setting a spending limit?

Mostly in timing and specificity. A limit you recall vaguely at the register is being reconstructed after you have already seen prices, which is exactly when anchoring operates. An envelope funded in advance for one category is a number that existed before the encounter and can be checked in a second. Neither of these is proven by the anchoring literature to reduce spending. The claim is narrower: one of them gives you a competing reference point and one does not.

Related