Why a bad third option can decide which of two you buy

The 1982 experiment that named the effect

In 1982, Joel Huber, John Payne and Christopher Puto published a study in the Journal of Consumer Research testing what they termed an asymmetrically dominated alternative — an option beaten on every attribute by one choice in a set, but not clearly beaten by the other. They ran the comparison across six unrelated product categories: cars, restaurants, beer, lotteries, film and television sets.

Adding the dominated option to a two-item choice set increased the dominating option's share of choices by an average of roughly 9 percentage points, according to a review of the study's design and results. The shift was larger when different groups of people each saw only one version of the choice set (about 9 points) than when the same people saw both versions and could compare their own earlier answer (about 3 points), a gap the researchers attributed to carryover effects.

A result that breaks a basic rule of choice

Classical choice theory includes a principle sometimes called regularity: adding an option to a set should never increase the chance that people pick something they could already have chosen, and it certainly should not change how the existing options are split between each other. An option nobody picks is supposed to be irrelevant to everyone else's decision.

Huber, Payne and Puto's result broke that rule directly. The decoy was irrelevant in the sense that almost nobody chose it, yet its presence changed the outcome for the two options that mattered. Their explanation centered on how easy dominance is to see: an option that clearly beats a decoy looks stronger by that comparison than it does when judged only against a competitor with its own tradeoffs.

The 2014 replications that found its limits

For three decades the finding was treated as one of the sturdier results in decision research. That changed in August 2014, when the Journal of Marketing Research ran two large replication efforts side by side. Shane Frederick, Leonard Lee and Ernest Baskin tested the effect under more realistic conditions — verbal descriptions instead of pure numbers, pictures, and choices carrying real financial stakes rather than hypothetical ones — and found the effect regularly shrank or reversed: in some tests, adding the decoy pulled choice share away from the option it resembled instead of toward it.

A companion study by Yang and Lynn ran 91 attempted demonstrations of the effect across 23 product categories and found only 11 produced a reliable shift, according to a review of the two 2014 papers. Huber, Payne and Puto responded in the same issue, arguing the effect still holds up reliably when the original conditions — numeric attributes and a decoy whose inferiority is obvious — are reproduced closely. They also conceded that the strong, universal version of the effect that had entered popular marketing advice was overstated.

What decides whether the effect fires

Across both sides of the argument, the same conditions keep recurring. The effect is strongest when a choice reduces to two or three clean numbers that are easy to compare — a price and a size, a price and a rating — and weakest once a decision involves qualitative differences, images, brand associations, or money a person will actually spend rather than imagine spending.

It also depends on how obviously the decoy loses. A decoy that is only marginally worse, or worse on some attributes but not others, does less to shift preference than one unambiguously beaten on every visible measure. That combination — numeric, obvious, low-stakes — describes a narrow slice of real purchasing decisions.

The same $125 combined option went from the clear favorite to a minority pick once the option that made it look cheap by comparison was gone.

The Economist subscription that made the effect famous

The clearest applied illustration comes from a classroom experiment described by behavioral economist Dan Ariely in his book Predictably Irrational, built around a real subscription page from The Economist: web-only access for $59, print-only for $125, and print plus web together also for $125. With all three options shown to a group of students, 16 percent picked the web-only plan, 84 percent picked the combined plan, and none picked the print-only option.

When Ariely removed the print-only decoy and asked a separate group to choose only between web-only and the combined plan, the result reversed: 68 percent picked the cheaper web-only plan and 32 percent picked the combined plan. The same $125 combined option went from the clear favorite to a minority pick once the option that made it look cheap by comparison was gone.

Why the tactic clusters in pricing tiers

The structure Ariely tested — a middle option priced close to the top tier but offering meaningfully less — now recurs across subscription pages, software pricing and gym memberships: three lines, a price and a short feature list for each, with the middle line rarely designed to be bought at all. That format shares some of the conditions researchers describe as helping the effect appear — a small number of alternatives, numeric prices, and a dominance relationship a buyer can spot in a glance down a table — but not the low-stakes, feature-free choice sets the lab studies used.

The same research that established the effect also limits its claims. Once a purchase involves comparing options that differ in kind rather than degree, or carries a price tag large enough to prompt real deliberation, the 2014 replications found the pattern far less dependable — in some cases working in the opposite direction from what the 1982 study predicted.