Contrast Effect: Why the Same Price Depends on the Last One

A judgment needs a frame of reference, and the nearest available frame is whatever was just looked at. The 1,700 dollar apartment in the demo above has not changed between the two questions and neither has the 600 dollar bike. What changes is the number sitting next to it, and the number next to it is what the rating is made against. Harry Helson called the pooled effect of recent stimuli the adaptation level, and treated every judgment as a reading of where the current item falls relative to it.

Allen Parducci's range-frequency model, published in 1965, sharpens this. A category judgment is a compromise between two things: where the item sits in the range of items you have seen, and how many of them it beats. Both parts depend on the set, which is why adding one very expensive apartment to a morning of viewings shifts every later rating, and why the shift is a change in how the scale is used rather than proof that the apartment now looks different. That distinction is worth keeping. A demo like the one above measures ratings, so it cannot tell a changed experience from a changed use of the slider.

Contrast is also not the only thing a comparison can do. Muzafer Sherif, Daniel Taub and Carl Hovland set out the condition that decided the direction in their task: an anchor inside or just beyond the series pulled judgments toward itself, which they called assimilation, while an anchor far outside the series pushed them away, which is contrast. That was a result about categorizing lifted weights, not a general law that remote numbers repel estimates: anchoring studies get attraction from numbers far outside the plausible range.

In perception the same word covers a related but distinct mechanism. A gray square on white looks darker than the identical square on black because of how neighboring regions of the retina and visual cortex interact, not because of how a scale is being used. Pages about the contrast effect often mix the two cases. They behave alike in one respect, that the judgment of a thing depends on its surroundings, and in markets that is enough to make the ordering of options a design decision. Itamar Simonson and Amos Tversky showed in 1992 that the set of options on offer changes which one people choose, and sellers who present a high-priced item first are relying on it.

Muzafer Sherif, Daniel Taub and Carl Hovland published "Assimilation and contrast effects of anchoring stimuli on judgments" in the Journal of Experimental Psychology in 1958. The judgments were about lifted weights, which suited the question because the experimenters could set the physical values exactly and the participants had no price, brand or reputation to reason from.

In the first session, participants lifted weights ranging from 55 to 141 grams one at a time and sorted each into one of six categories, where one was the lightest and six the heaviest. That session established how the series was judged with nothing else to compare it against. In the later sessions the weights came in pairs: participants first lifted an anchor weight, which was described to them as belonging in the topmost category, and then the weight to be judged. The anchors ranged from 141 grams, the top of the original series, to 347 grams, well above anything the participants had lifted in the first session.

The direction of the shift depended on where the anchor sat. An anchor at the end of the series, or only slightly beyond it, displaced the scale of judgment toward itself: the series was absorbed into a frame that now included the anchor, which is assimilation. A remote anchor did the opposite. Once the heaviest thing in the session was far heavier than anything in the series, the series itself was pushed toward the lower categories, so the same weights were judged lighter than they had been with no anchor at all. That is the contrast effect, and the paper's contribution was to show that the two outcomes are the same process read at different distances.

The authors were interested in attitudes as much as in weights, and the result became one of the foundations of social judgment theory: a position far from your own is pushed further away, while a position close to your own is pulled in. The demo above is a consumer version of the same design, with a price standing in for the anchor weight and a rating scale standing in for the six categories.

Wherever options arrive in a sequence, the sequence is part of the judgment. The rows below are illustrations, not findings from the studies cited on this page.

The comparison set is doing the work, so the counter is to choose the set yourself.

• Fix your criteria in writing before you look. A rent limit, a salary range or a scoring sheet gives a frame that the next item on the screen cannot move.
• Widen the set deliberately. Looking at ten comparable options, rather than the two in front of you, makes any single extreme item a smaller part of the frame.
• Score each option as you go. A rating made immediately after one apartment is contaminated by the previous apartment; a rating made at the end of the day is contaminated by all of them.
• Ask what the first number was for. When a price, a salary or a tier is shown first and is far outside the range, treat it as part of the presentation rather than as information.
• Rotate the order when you are the one presenting. In interviews, tastings and demo days the ordering changes the scores, so the only fair sequence is one that varies across judges and is recorded.

Contrast effects are easy to confuse with the other ways a comparison changes a judgment.

• It is not anchoring, where a number seen first pulls an estimate toward it. They are the two directions a comparison can move a judgment, toward a reference point or away from it. Sherif, Taub and Hovland got assimilation from a near anchor and contrast from a remote one in their 1958 weight task, but distance alone does not predict the direction: the anchoring demo on this site moves estimates toward anchors of 9 and 140.
• It is not the decoy effect, where an unattractive third option is added to shift the choice between the other two. The decoy works on the structure of a choice set; contrast works on the scale a single option is rated against.
• It is not the framing effect, where the same outcome described as a gain or as a loss changes the decision. Framing changes the wording; contrast changes the company the option keeps.