Comparison Dimensions and Similarity: Addressing Individual Heterogeneity

Abstract

How many comparison dimensions individuals consider when they are asked to judge how similar two different objects are? I address individual heterogeneity in the number of comparison dimensions with data from a laboratory experiment. I estimate the smallest number of dimensions such that objects may be represented in space where distance corresponds to similarity. I find that the mean smallest number of dimensions in real data is one standard deviation smaller than in randomly simulated data. Furthermore, I find that individuals who find the objects relatively similar to each other are also the ones who implicitly consider fewer dimensions.

Keywords: multidimensional scaling, similarity, latent dimensions, dimensionality

JEL Classification: D12