Noncompensatory Information
A noncompensatory set of weights in a linear model means each weight is larger than the sum of all smaller weights. Take the Best has equivalent accuracy when the order of cue validities matches the order of those weights.
THE IDEA
Environments may be compensatory or noncompensatory. Among the 20 environments studied in Section 2, we found 4 in which the weights for the linear models were noncompensatory (i.e., each weight is larger than the sum of all smaller weights in a finite set, such as 1/2, 1/4, 1/8, . . .). The following theorem states an important property of noncompensatory models and is easily proved (Martignon et al., 1997).
THEOREM Take the Best is equivalent in performance to a weighted linear model whose weights form a noncompensatory set.
If multiple regression happens to have a noncompensatory set of weights (where the order of this set corresponds to the order of cue validities), then its accuracy is equivalent to Take the Best.
When to use it
Designing decision rules for environments with limited time or information, where one dominant cue should override all others.
Modeling expert judgment when a single highly valid cue trumps all other evidence (e.g., medical triage based on one critical symptom).
Simplifying linear models for interpretability by ensuring noncompensatory weights that mimic lexicographic heuristics.
FROM THEORY TO EVERYDAY LIFE
In practice
Noncompensatory Weight Set
In a finite model, a set such as {1/2, 1/4, 1/8} is noncompensatory: each weight is larger than the sum of all subsequent weights. No combination of lesser cues can outweigh a more important cue.
A MOMENT TO REFLECT
Put it to work.
In which real-world decision contexts might noncompensatory weights naturally arise?
How would the equivalence of Take the Best and a noncompensatory linear model change if the weights were compensatory?
Often confused with
References
- Martignon et al., 1997 — Martignon, L. and others (1997)
CONTINUE THE CONVERSATION
Source texts
- Heuristics and Biases The Psychology of Intuitive Judgment (Thomas Gilovich, Dale Griffin, Daniel Kahneman)