Attraction effect

These illustrate the pattern the research describes. As the Evidence section explains, how often it actually changes choices outside the laboratory is disputed.

The coffee subscription

A coffee roaster offers two plans: 1 bag a month for $15, or 2 bags a month for $26. Most visitors pick the single bag. The roaster adds a third plan, 2 bags a month for $30, and more visitors start choosing the $26 two-bag plan. One says: “The $26 plan is obviously the good deal.”

The new plan is simply the two-bag plan at a higher price, so it loses to it outright. It isn’t clearly worse than the single bag, though: it costs more but delivers more coffee. Nobody picks it. What it does is give the $26 plan an easy win, and that win is what makes it feel like “the good deal”. The comparison that matters, 1 bag for $15 against 2 for $26, hasn’t changed at all. This is the core case, in the number-based form where the effect is best documented.

The third finalist

A hiring committee has two finalists. One has more experience; the other scores higher on the work sample. The committee is split. Then a third applicant is added to the shortlist: slightly less experienced than the first finalist, with a slightly weaker work sample, but still more experienced than the second. A committee member says: “Well, next to her, the experienced candidate clearly stands out.”

The third applicant is worse than the experienced finalist on both counts, but not clearly worse than the other. Beating her says nothing about how experience compares with work-sample quality, which is the question that divided the committee. This shows the claimed pattern outside shopping, in a decision that feels like careful judgment.

“Same trip, one night shorter”

A travel agent describes two packages: a 7-night beach trip for $1,800 and a 5-night city trip for $1,500. Before the customer answers, the agent adds: “There’s also the beach trip at 6 nights for $1,800, but honestly the 7-night one is better value.” The customer books the 7-night beach trip.

Here the decoy is mentioned in passing rather than listed. The 6-night option exists only to lose to the 7-night one. If the customer was drawn to the beach trip because it “beat” an option they’d never have taken, the comparison did the persuading. If they simply preferred a beach to a city, the decoy didn’t matter, and from outside it’s impossible to tell which happened.

The attraction effect is one of three classic context effects, in which the other options on offer change choices between the same two options:

• Compromise effect: an option looks more attractive when it sits in the middle of the range, between an extreme option on each side. Adding a very expensive, very high-end model makes the mid-priced one sell better.
• Similarity effect: adding an option very similar to one of the originals takes choices mainly from that original. A second, nearly identical sci-fi film on a streaming menu splits the sci-fi fans’ votes.
• Repulsion, a reversal rather than a separate classic effect: a nearby inferior option makes its neighbor less popular instead of more. It has been reported under some conditions.

• When the extra option carries information. A third option can tell you something real: what is normal to pay, which features matter, or what a good product in this category looks like. If it changes your understanding of the options, letting it change your choice is reasonable.
• When you were indifferent anyway. If the two real options are equally good for you, breaking the tie by whatever stands out costs nothing, as long as you know that’s what you’re doing.

The test: if the option I’d never pick disappeared, would I still choose the same one?

The plan that showed what mattered

A buyer is choosing between a phone plan with 64 GB of storage and a pricier one with 256 GB, and leans toward the cheaper one. A third plan appears: 128 GB at the higher price, with a note that “people with more than 5,000 photos usually need at least 128 GB.” The buyer checks their photo library, finds 9,000 photos taking up nearly 60 GB, and picks the 256 GB plan.

The third plan is worse than the 256 GB plan, so this looks like a decoy at work. But it changed the choice by prompting the buyer to find out something true about their own needs. The choice now rests on a fact about storage, not on the pricier plan’s easy win. That’s the information condition above.

Choosing by a rule you’d apply anyway

A family comparing three cars rules out one immediately because another model beats it on price, fuel economy and safety. They then compare the remaining two on their own terms and buy the one with more cargo space, which is what they needed.

Setting aside a dominated option is sound reasoning. What would make it the attraction effect is if beating the dominated option became the reason for the final choice. Here, the last comparison is made directly between the two real contenders.

Researchers have proposed several explanations, and none is settled:

• Comparisons are local. People compare options one attribute at a time, and the target wins those comparisons against the decoy while the competitor can’t be compared cleanly. The effect is strongest when attributes are numbers or short descriptions laid out side by side, which make those comparisons easy, and weak or absent with pictures or real experience of a product.
• Preferences are partly built on the spot. Many choices don’t come from fixed values. People construct them from the options in front of them, so the set on offer can shape the result. This is the same lesson as the Framing effect, where presentation rather than substance changes a choice. It also resembles the Anchoring effect, in which a reference point that shouldn’t matter shifts a judgment, though the two are studied as separate effects.

• Strike the options you’d never choose, then compare what’s left as if they were the only two.
• Ask what the extra option tells you. If it teaches you something about your needs or about prices, use that. If it only makes one option look good by contrast, set it aside.
• Be wary of lineups built by a seller. A plan or package that nobody would buy may be there for comparison. These are common-sense checks rather than remedies that have been tested.

Status: contested. The effect in its original laboratory form has been replicated many times, but whether it survives realistic presentation, and how much it changes what people actually choose, is seriously disputed. The 2014 debate remains unresolved, and the largest synthesis so far is a 2026 meta-analysis that has not yet been peer reviewed.

The original work. Huber, Payne and Puto (1982) showed that adding an “asymmetrically dominated” option, one worse than one item in the set but not the other, can increase the probability of choosing the item that dominates it. This broke a principle called regularity, built into standard models of rational choice: adding an option to a set shouldn’t make any existing option more likely to be chosen. Their stimuli described products by attributes such as price and quality ratings, for example six-packs of beer.

The challenge. Two papers published together in 2014 argued that the effect is far narrower than its reputation.

• Frederick, Lee and Baskin (2014) proposed that it may be restricted to stylized products in which every attribute is a number. In their studies it typically didn’t occur when people experienced the product, such as tasting a drink, or when even one attribute was shown perceptually, such as a hotel room’s quality shown in a photo. They questioned the effect’s practical significance.
• Yang and Lynn (2014) made 91 attempts to produce the effect across 23 product classes and found reliable effects in only 11, significantly fewer than their studies’ statistical power should have delivered. With verbal descriptions or pictures, effects appeared only at chance levels.

The replies. Huber, Payne and Puto (2014) argued the effect remains robust when the conditions of the original studies are essentially replicated, and treated the new findings as information about what weakens or strengthens it. Simonson (2014) argued that a decoy’s influence depends on stronger drivers of choice, so failing to find it where those dominate is unsurprising; that it is often observed when price is an attribute, including in the real world; and that many of the failed attempts did not test it properly.

The meta-analysis. Cai and Pleskac (2026, posted as a preprint) pooled 83 studies comprising nearly a million individual choices, and adjusted for signs that small studies and selective publication had inflated effects.

• The attraction effect held up better than the other two context effects: there was very strong evidence that a decoy shifts choices toward the target relative to its competitor.
• But the evidence favored no average increase in the share of people choosing the target. That increase is the regularity violation Huber and colleagues highlighted, and the one the popular version (“add a decoy and sell more of the target”) depends on.
• Effects varied greatly between experiments. They were credibly positive with numeric and verbal attributes, and not credibly different from zero with pictorial or experience-based ones. They were also larger when displays encouraged comparing options attribute by attribute.

Outside the laboratory. Devine and colleagues (2025) analyzed 3.6 million wine purchases at UK grocery stores and found that dominated options on the shelf did make the dominating wines more likely to be chosen, but the effect was modest, roughly a 1% change in preference.

What remains uncertain is where the effect’s boundaries lie. That people’s choices can be nudged by dominated options in simple, number-based comparisons is well documented. Whether decoys are a dependable way to change real choices, and why they work in some displays but not others, is not.