Resumo: In a traditional sorting task, consumers receive a set of
representational items and sort them into piles such that the items in
each pile “go together” according to some criteria specified either by
the researcher or the participants themselves. Although this task
offers good flexibility in accommodating different instructions (and
research questions), no data analytic procedures exist to help
summarize the piles that consumers make, without assuming a particular
cognitive process or without requiring arbitrary conversions of piles
into data that influence the results. This paper introduces a problem
called “Category Covering” to summarize sorting task data, as well as
a proposed optimization model to identify summary piles. The
specification accommodates sorting tasks that feature instructions
other than to sort based on similarity, does not require arbitrary
aggregate conversions of piles into pairwise distances, and does not
require researchers to make parametric assumptions that may not hold.
Rather, a proposed procedure quickly identifies globally optimal
parameter estimates as demonstrated with an empirical application in a
retail setting.