inferring choice
criteria with mixture irt models: a demonstration using ad hoc and goal-derived
categories
;Steven
Verheyen;Wouter
Voorspoels;Gert
Storms
nederlands tijdschrift voor geneeskunde2015Vol. 10pp. 97-114
162
steven
verheyen2015judgmentinferring
Abstract
Whether it pertains
to the foods to buy when one is on a diet, the items to take along to the beach
on one's day off or (perish the thought) the belongings to save from one's
burning house, choice is ubiquitous. We aim to determine from choices the
criteria individuals use when they select objects from among a set of
candidates. In order to do so we employ a mixture IRT (item-response theory)
model that capitalizes on the insights that objects are chosen more often the
better they meet the choice criteria and that the use of different criteria is
reflected in inter-individual selection differences. The model is found to
account for the inter-individual selection differences for 10 ad hoc and
goal-derived categories. Its parameters can be related to selection criteria
that are frequently thought of in the context of these categories. These
results suggest that mixture IRT models allow one to infer from mere choice
behavior the criteria individuals used to select/discard objects. Potential
applications of mixture IRT models in other judgment and decision making
contexts are discussed.