Blocking chart · where it stands
Decoy Effect
No single mechanism is settled. Huber, Payne & Puto (1982) proposed two broad families: (a) perceptual/weighting accounts — a decoy that extends the range or frequency on the attribute where the target is strong shifts attribute weights or rescales perceived differences (cf. Parducci's range-frequency theory), making the competitor's advantage seem smaller; and (b) process accounts — attribute-by-attribute pairwise comparisons or 'count-the-wins' tournaments in which the easily-beaten decoy adds wins to the target. Later work frames it as context-dependent valuation via pairwise comparison and divisive normalization, and notes a 'value-shift' versus 'process-shift' debate (Wedell 1991) that has never been cleanly resolved.
First demonstrated by Huber, Payne, and Puto (1982), who showed that inserting an asymmetrically dominated "decoy" — an option worse than the target but not worse than the competitor — raised the target's choice share by about 9 percentage points between-subjects across six product categories. The effect (also called the attraction or asymmetric-dominance effect) violates the "regularity" axiom of standard choice models. It is real under tightly controlled, numeric, hypothetical lab conditions, but a string of larger studies (Yang's 2013 meta-analysis; Frederick, Lee & Baskin 2014; Yang & Lynn 2014) found it is fragile, shrinks or vanishes with verbal/pictorial or experienced stimuli, and may have limited practical relevance.
Huber, Payne & Puto (1982) — original demonstration
9.2%
Yang (2013) — meta-analysis of the literature
14.7%
Yang & Lynn (2014) — robustness challenge
on the mark
Frederick, Lee & Baskin (2014) — 'The Limits of Attraction'
on the mark
Adding a third option that is dominated by (clearly worse than) one of two existing options on every relevant attribute can raise the chance that people pick the dominating "target," reversing the preference they would have had in the two-option set.
Stage left · in the wild
The Economist subscription pricing (Ariely classroom demonstration)
Stage right · the exit
Strip the set down to genuine contenders and evaluate each finalist on its own absolute merits, not by in-set comparison. Ask: 'If this clearly-worse option weren't here, would I still prefer this one?' Ignore any option that is dominated (worse on every attribute) — it carries no information about the others. Because the effect collapses with real stakes and experienced/non-numeric attributes, converting abstract numbers into concrete consequences ('what do I actually get?') also defuses it.
Downstage · the charge
Decoy Effect
Adding a third option that is dominated by (clearly worse than) one of two existing options on every relevant attribute can raise the chance that people pick the dominating "target," reversing the preference they would have had in the two-option set.
Mid-stage · the marks
- 1. Huber, Payne & Puto (1982) — original demonstration · 9.2%
Adding the dominated decoy increased the target's share, violating regularity in a predicted direction; range-increasing decoys were most effective, frequency decoys weakest.
- 2. Yang (2013) — meta-analysis of the literature · 14.7%
The effect exists under conservative methods but the methodological factors creating it remain elusive; effect weakens as the target already captures more share in the control condition; no other choice/methodological characteristic reliably predicted its magnitude.
- 3. Yang & Lynn (2014) — robustness challenge · on the mark
Only 11 of 91 attempts produced reliable effects — far fewer than the studies' power implied; meaningful verbal descriptions and pictures reduced effects to chance levels.
- 4. Frederick, Lee & Baskin (2014) — 'The Limits of Attraction' · on the mark
The attraction effect was largely confined to fully numeric, stylized stimuli; it typically did not occur when consumers experienced the product or when one attribute was perceptual, leading the authors to question its practical validity.
The set · mechanism
No single mechanism is settled. Huber, Payne & Puto (1982) proposed two broad families: (a) perceptual/weighting accounts — a decoy that extends the range or frequency on the attribute where the target is strong shifts attribute weights or rescales perceived differences (cf. Parducci's range-frequency theory), making the competitor's advantage seem smaller; and (b) process accounts — attribute-by-attribute pairwise comparisons or 'count-the-wins' tournaments in which the easily-beaten decoy adds wins to the target. Later work frames it as context-dependent valuation via pairwise comparison and divisive normalization, and notes a 'value-shift' versus 'process-shift' debate (Wedell 1991) that has never been cleanly resolved.
Standing · replication
The effect reliably reproduces in tightly controlled, hypothetical, fully numeric two-attribute choices, but is fragile beyond them. Yang & Lynn (2014) got reliable effects in only 11 of 91 attempts; Frederick, Lee & Baskin (2014) found it largely vanishes when any attribute is perceptual or the product is experienced; Yang's (2013) meta-analysis puts the average share gain near 14.7% but argues earlier figures were inflated by methodological artifacts and that the effect weakens as the target's baseline share rises. Lichters, Sarstedt & Vogt (2015) note that of 52 marketing experiments reviewed, essentially only Doyle et al. (1999) used real incentives, questioning practical relevance. The original authors pushed back in Huber, Payne & Puto (2014, 'Let's Be Honest About the Attraction Effect'), conceding boundary conditions while defending the core phenomenon. Pre-registered replications (e.g., Frederick-tradition and Ariely/Wallsten-Connolly lines) give mixed results.
The marks in detail
Huber, Payne & Puto (1982) — original demonstration. Adding the dominated decoy increased the target's share, violating regularity in a predicted direction; range-increasing decoys were most effective, frequency decoys weakest.
Joel Huber, John W. Payne, Christopher Puto, 1982 · N=153 (between-subjects); 93 subjects retested two weeks later (within-subjects), 558 choices
Yang (2013) — meta-analysis of the literature. The effect exists under conservative methods but the methodological factors creating it remain elusive; effect weakens as the target already captures more share in the control condition; no other choice/methodological characteristic reliably predicted its magnitude.
Sybil S. Yang (advisor Michael Lynn), 2013 · Meta-analysis aggregating many published studies (exact k per author's appendix); plus original experiments
Yang & Lynn (2014) — robustness challenge. Only 11 of 91 attempts produced reliable effects — far fewer than the studies' power implied; meaningful verbal descriptions and pictures reduced effects to chance levels.
Sybil Yang, Michael Lynn, 2014 · 91 attempts; 23 product classes; 73 decoyed sets
Frederick, Lee & Baskin (2014) — 'The Limits of Attraction'. The attraction effect was largely confined to fully numeric, stylized stimuli; it typically did not occur when consumers experienced the product or when one attribute was perceptual, leading the authors to question its practical validity.
Shane Frederick, Leonard Lee, Ernest Baskin, 2014 · Several studies (article reports per-study Ns); published JMR 51(4):487-507
Catch it in the act
You see a third option that nobody would rationally pick — it's worse than one of the other options on every dimension — yet it sits there making one specific option look like the obvious 'smart' choice. Classic tell: a pricing menu where one tier is strictly worse than the tier next to it (e.g., print-only priced the same as print+web). If removing the never-chosen option would change which remaining option feels best, a decoy is at work.
Cast list
- Compromise effectAdding an extreme option makes a middle option more attractive by making it a 'compromise'; like the decoy effect it is a context/menu effect, but the added option is not dominated.
- Similarity effectA new option steals disproportionate share from the most similar existing option — the decoy effect actually reverses this, since a decoy similar to the target helps (not hurts) the target.
- Phantom decoy effectAn attractive but unavailable (out-of-stock) option that dominates the target shifts choice toward the target; a variant of asymmetric dominance with an unavailable decoy.
- Regularity (axiom of choice)The decoy effect is defined precisely as a violation of regularity — the principle that adding an option cannot raise the share of an existing option.
- AnchoringBoth involve a reference point biasing choice, but anchoring works via a numeric reference for a judgment; the decoy works via a dominated third option changing relative comparisons within a set.