Elimination by aspects
We cut options by ruling out whatever fails one attribute at a time.
What it means
Elimination by aspects is a noncompensatory choice rule in which we whittle a large set down by taking one attribute at a time and discarding everything that lacks it, rather than scoring every option on every dimension. The mechanism is sequential filtering: an aspect is selected with probability tied to its importance, all options missing it are eliminated, and the process repeats until one survivor remains. It describes how people actually work through filter panels, sift resumes, and pick a restaurant, which is why faceted search and comparison tables feel so natural. Because a strength on one attribute cannot offset a failure on another, an excellent option can be knocked out early by a single missed cutoff, and the order in which aspects come up changes the winner. It matters because whoever controls the filters, and their sequence, quietly controls the choice.
A formal model, not just a rule of thumb
Tversky introduced EBA in 1972 as a precise probabilistic choice model, not a loose description of shortcut behavior. Each option is a bundle of aspects; the model gives every aspect a weight, draws aspects in proportion to those weights, and eliminates whatever lacks the drawn aspect. Formally it is a random-utility model that generalizes Luce's and Restle's earlier theories. Its headline achievement was escaping the independence of irrelevant alternatives, the awkward property built into simpler models whereby adding a third option cannot change the relative odds between two others. Because similar options share aspects, EBA lets a new entrant steal share mainly from its near-neighbors, which is why economists such as McFadden saw it as a serious tool for modeling real substitution patterns.
What the evidence shows
People do not use one strategy for everything; they switch, and EBA-style elimination is what they reach for when a choice gets big. Process-tracing studies that track which pieces of information a person actually inspects find that as the number of options or attributes rises, and as time pressure mounts, decision-makers abandon exhaustive option-by-option scoring for sequential, attribute-wise elimination. Payne, Bettman and Johnson framed this as an effort-accuracy trade-off: noncompensatory rules are cheap and, in many environments, nearly as accurate as the full calculation they replace. The evidence is therefore not that EBA is how we always choose, but that it is a strategy the mind deploys selectively, mostly to prune an unmanageable set down to a shortlist.
The tyranny of the cutoff
Because a strength cannot buy back a weakness, EBA punishes the balanced all-rounder. An option that is second-best on every aspect but exceptional on none can be eliminated before its overall quality is ever assessed, while a lopsided option that clears each hurdle survives. Applied to continuous attributes the rule is harsher still: a hard cutoff treats a laptop at $1,001 as failing 'under $1,000' exactly as a $3,000 one does, discarding a near-miss that a compensatory weighing would have ranked highly. The rule also throws away trade-off information; it never learns how much cheaper one option is, only that it passed. And because outcomes depend on which aspect happens to be drawn first, EBA can yield intransitive, order-dependent choices that no single stable ranking explains.
Using it well
In practice EBA works best as a first pass, not the whole decision. A robust real-world pattern is the two-stage strategy: eliminate by a few must-have aspects to get from hundreds of options down to a handful, then switch to careful, compensatory comparison among the survivors, where trade-offs finally matter. That second stage guards against discarding a strong all-rounder in the final round. For anyone designing a choice, the order of search facets, which filters are offered, and where default cutoffs sit are not neutral conveniences; they set which options ever reach comparison. And for the chooser, the discipline is to notice when a single cheap cutoff has quietly deleted options that were worth a second look.
Examples
Shopping for a laptop, you click 'under $1,000,' then '16GB RAM,' then 'in stock,' and each filter wipes out whole swathes of the catalogue. You buy whatever survives.
A recruiter screens 400 applicants by rejecting anyone without a degree, then anyone under five years' experience, then anyone needing a visa, never comparing the survivors' actual strengths.
Picking dinner with friends, you rule out anywhere without vegetarian options, then anywhere over a twenty-minute walk, then anywhere fully booked, and the last place standing wins by default.
Deciding what to cook for dinner, you cross off every recipe that needs an ingredient you lack, then every one that takes more than half an hour, then anything too spicy for the kids, and settle on a dish without ever judging which would actually taste best.
Registering for next term, you drop every course that meets before ten, then every one held on Fridays, then every section with no seats left, and enroll in whatever survives without comparing the instructors.
First described in Tversky (1972); Tversky & Sattath (1979).
Key references
- Payne, J. W., Bettman, J. R., & Johnson, E. J. (1988). Adaptive strategy selection in decision making. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14(3), 534-552. doi.org/10.1037/0278-7393.14.3.534
- Johnson, E. J., & Meyer, R. J. (1984). Compensatory choice models of noncompensatory processes: The effect of varying context. Journal of Consumer Research, 11(1), 528-541. doi.org/10.1086/208989
- Tversky, A., & Sattath, S. (1979). Preference trees. Psychological Review, 86(6), 542-573. doi.org/10.1037/0033-295X.86.6.542
- Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological Review, 79(4), 281-299. doi.org/10.1037/h0032955