Making trade-offs: a probabilistic and context-sensitive model of choice behavior

Area of Science:

• Decision Science
• Cognitive Psychology
• Mathematical Psychology

Background:

• Decision-making models often simplify the complex process of evaluating attribute differences.
• Understanding how individuals weigh attribute value differences is crucial for predicting choices.
• Existing models may not fully capture context-dependent variations in decision sensitivity.

Purpose of the Study:

• To introduce and test the stochastic difference model, specifically the proportional difference (PD) model.
• To evaluate the PD model's ability to account for individual and group choice data.
• To assess the model's capacity to describe violations of standard choice axioms.

Main Methods:

• The study employs a stochastic difference model framework.
• Choice probabilities are modeled as a function of a normalized difference variable (d) and a decision threshold (delta).
• The proportional difference (PD) normalization was utilized and tested against nine diverse datasets.

Main Results:

• The proportional difference (PD) model demonstrated strong explanatory power for both individual and group data.
• The decision threshold (delta) effectively captured context-dependent sensitivity to attribute value differences.
• The model successfully described observed violations of stochastic dominance, independence, and stochastic transitivity.

Conclusions:

• The stochastic difference model, particularly the PD variant, provides a robust framework for understanding decision-making.
• The decision threshold parameter offers valuable insights into context effects on choice behavior.
• The PD model's ability to account for choice axiom violations enhances its predictive and descriptive capabilities.