What Is the Decoy Effect? Adding a Bad Option to Change Which Good One Wins

Presented with two options that trade off against each other, so one is better on price and the other on quality, people divide between them. Adding a third option that is worse than one of the existing two in every respect, while not being worse than the other in every respect, shifts choices towards the option it is dominated by. The decoy is not chosen, and its presence changes the comparison by making one option look obviously superior to something, which is a comparison the other option does not win. The effect violates a basic assumption of rational choice, namely that adding an option nobody selects should not change the ranking of the others, and that violation is why the finding attracted so much attention when it was reported in the 1980s.

Commercial application is widespread and the pattern is recognisable once known:

• •Three-tier pricing where the middle option is made to look like obvious value by an unattractive top tier
• •Subscription offers where a print-only option priced identically to a combined print and digital bundle exists solely to make the bundle look free
• •Menu pricing where an expensive item raises the acceptability of the second most expensive
• •Electronics ranges where a deliberately hobbled model makes the next one up look generous
• •Property viewings where a poor option is shown first
• •Recruitment and selection, where the effect has been demonstrated with candidate profiles

The effect has been subject to serious reassessment and the picture is mixed. Large preregistered studies have failed to reproduce it in several forms, particularly with realistic products rather than abstract attribute descriptions, and reviews have argued that the classic demonstrations depended on artificial stimuli where options differ on exactly two numeric dimensions and where the dominance relationship is unmistakable. Other work continues to find it under specific conditions. The current position is that the effect is real in constrained laboratory settings, is smaller and less reliable with realistic choices, and is not the robust general phenomenon that popular accounts describe. That matters for the commercial claims, since the pricing arrangements are widely attributed to it and may be effective for other reasons, including simple anchoring on the highest price.

The reason the finding attracted attention goes beyond marketing. Standard accounts of rational choice assume that preferences exist independently of the options presented, so a person prefers one thing to another and adding a third option nobody wants cannot reverse that, which is the independence assumption. Effects like this one threaten it directly, and if preferences are constructed during the act of choosing rather than consulted, then a great deal of economics, of policy analysis based on stated preferences and of measurement of what people want rests on shakier ground than assumed. That is the substantive stake, and it explains why the replication status of the specific effect matters more than a marketing trick would warrant. The broader evidence for constructed preference is substantial and does not depend on this one demonstration surviving.

The effect belongs to a set of findings about how context shapes choice, and the others have their own standing. The compromise effect finds that an option becomes more attractive when it sits between two extremes rather than at an end, which is why three-tier pricing works even without a dominated decoy. Anchoring shifts valuation towards whatever number was encountered first. Framing changes choices by describing identical outcomes as gains or losses. The common thread is that preferences appear to be constructed at the moment of choosing rather than read off a stable internal ranking, which is a substantial claim about how choice works and is better supported than any individual effect within the family. That general conclusion survives even where specific demonstrations have not replicated.

A third option nobody chooses shifts preference towards the option it is worse than in every respect, which violates the assumption that irrelevant alternatives do not matter. Three-tier pricing and bundled subscription offers are the standard applications. Large preregistered studies have failed to reproduce it with realistic products, and the general finding that preferences are constructed rather than read off survives regardless.