Riskless Choice: Presence of Multiple Attributes. How are decisions made? How do you make such choices ? . Riskless Choice: simplifying heuristics. Presence of multiple attributes, decision alternatives represented as vectors Simplifying heuristics :
Riskless Choice: Presence of Multiple Attributes
How aredecisions made?
Note 1: This is a non-compensatory model. What does it mean?
Note 2: Recall lexicographic semi-order. How does it differ?
Makepairwisecomparisons of alternatives
Estimate the weights for the attributes
Value difference in twoalternatives > epsilon
Not a badmodel in predictingchoices …
Based on the independence principle:
p(x; y) ≥ ½ iff p(x; A) ≥ p(y; A) (provided p(y; A) ≠ 0)
The ordering of x and y, by choice probability, is independent of the considered set of alternatives
Illustrate with an example!
Views riskless choice behavior as a probabilistic process (because of observed inconsistencies, uncertainty in choosing)
Each alternative is viewed as a set of aspects (characteristics)
This paper develops a probabilistic theory of choice, based on a covert elimination process – critique of the Luce model
Consider buying a record for your collection. There are three choices: a Debussy (D) and two different recordings of the same Beethoven symphony (B1, B2). Assume the two Beethoven recordings are of equal quality, and that you are undecided between adding a Debussy or a Beethoven to your collection.
Hence p(B1; B2) = u(B1)/(u(B1)+u(B2))=1/2 and u(B1)=u(B2)=1/2; similarly: p(D; B1) = p(D; B2) = ½ and the corresponding utilities equal each other
According to the Luce model, p(D;B1,B2)= u(D)/ (u(D)+u(B1)+u(B2))=1/3. Intuitively not right! One would expect it to be ½! Lesson: set A matters!