An efficient method for obtaining similarity data

Abstract

Although this is sound advice, their core assumption (that the pairwise method is de facto superior) is not clearly supported by evidence. The pairwise method has a rich empirical history and is clearly an excellent approach. It also has obvious drawbacks, and there are doubtless many examples of pairwise experiments that did not "work" as researchers may have hoped. Given that few MDS solutions can clearly be deemed as "correct," we suggest that both methods come with their own caveats.

The Novel Object and Unusual Name (NOUN) Database: A collection of novel images for use in experimental research

“…It is clear from these plots that the data are comprised of more than just two primary featural Configuration. In order to choose the "best" solution to report, we correlated the interitem distance vectors across solutions derived from each of the three starting configurations (see Ferguson, 2012 andGoldstone, 1994 for similar approaches). This provides a metric to gauge the extent to which the arrangement of points in one MDS space "agrees" with the others (e.g., a pair of points that is located close together in one space should be close together in the others, and vice versa).…”

“…Thus, this provided a full similarity matrix comparing the ratings of each image to all of the other images (i.e., all 2016 comparisons) for each participant. This took participants under an hour to complete; similar rating procedures have been used by other researchers (Goldstone, 1994;Hout, Papesh, & Goldinger, 2012;Kriegeskorte & Marieke, 2012).…”

“…Participants were shown all 64 objects across 13 trials, and provided similarity ratings using the spatial arrangement method (SpAM, Hout, Goldinger, & Ferguson, 2013, see also Goldstone, 1994). On each trial, twenty different pictures were shown to the participant, arranged in 4 discrete rows (with 5 items per row), with random item placement.…”

Uncovering mental representations with Markov chain Monte Carlo

“…MDS has been applied to many natural categories, including Morse code (Rothkopf, 1957) and colors (Nosofsky, 1987). Traditional MDS does not scale well to a large number of objects because all pairwise judgments must be collected, but alternative formulations have been developed to deal with large numbers of stimuli (Hadsell, Chopra, & LeCun, 2006;de Silva & Tenenbaum, 2003;Goldstone, 1994).…”

Optimizing stimuli order in marketing experiments: A comparison of four orders using six criteria

“…30 With this approach, all brands are available to consider simultaneously. Subjects may refocus their attention as desired, in whatever way best allows them to complete their judgments.…”