Description
Measures similarity judgements in both directions to reveal which interface or concept users treat as the prototype, using Tversky's feature-matching model of common and distinctive features.
Tversky's contrast model treats similarity as a weighted function of common features minus each object's distinctive features, rather than a symmetric distance in a geometric space. Because the weights differ, similarity is often asymmetric: people say a variant resembles the prototype more than the prototype resembles the variant. The direction of the asymmetry reveals which item functions as the reference. Applied to UX, asking 'how similar is our app to X?' and 'how similar is X to our app?' separately reveals whether users treat a competitor, a legacy system or a familiar convention as the standard against which new designs are judged. Feature listing then identifies which shared and distinctive features drive judgements. The method was briefly used in early HCI similarity research and has largely disappeared.
Attributes
- Discipline
- Cognitive Psychology & Perception
- Loss type
- Extinct
- Data
- Quantitative
- Effort
- Days
Steps
- Present pairs of interface elements.
- Ask 'does A resemble B?' then 'does B resemble A?' Measure asymmetry.
- Large asymmetry reveals the prototype.
- Redesign unfamiliar elements toward the prototype direction.
Sources
- Preference, Belief, and Similarity — Amos Tversky
- Tversky, Features of similarity (Psychological Review, 1977)
Also known as
- Feature contrast model
- Asymmetric similarity judgement
Categories
- User Research
- Cognitive Psychology & Perception