Incorporating Thresholds of Indifference in Probabilistic Choice Models

Abstract

Probabilistic choice models are often based on utility concepts which call for assessments by decision makers of available alternatives. These models assume that distinct utilities can be derived from the decision maker's expressed preferences. Such an assumption is generally valid if the decision maker can recognize arbitrarily small utility differences. If the decision maker is indifferent between the alternatives the corresponding utilities are expected to be similar. In this paper, the concept of minimum perceivable difference is introduced into the conventional binary logit model. It is postulated that two alternatives are perceived as different only if the absolute difference in their utilities exceeds a positive constant. For estimating the parameters of the new model a maximum likelihood technique is employed, and an empirical test of the model is conducted using data on travelers' choice of mode for accessing commuter rail service from Lindenwold (New Jersey) to downtown Philadelphia. The model predicts individuals' choices in a holdout sample significantly better than the conventional logit model.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 76 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.