The versatility of SpAM: a fast, efficient, spatial method of data collection for multidimensional scaling

Area of Science:

• Cognitive Psychology
• Psychometrics
• Data Visualization

Background:

• Traditional multidimensional scaling (MDS) methods rely on pairwise comparisons, which become computationally intensive and time-consuming as the number of stimuli increases.
• Existing data collection protocols for MDS can lead to lengthy experiments or require scaling down stimulus sets, potentially compromising data quality.
• There is a need for efficient and robust methods to gather similarity data for constructing accurate perceptual or conceptual spaces.

Purpose of the Study:

• To review and critically evaluate existing methods for collecting similarity data used in multidimensional scaling (MDS).
• To examine the efficacy of the spatial arrangement method (SpAM) as a user-friendly and efficient alternative to traditional data collection techniques.
• To assess the quality and robustness of MDS solutions generated using SpAM compared to pairwise comparison methods.

Main Methods:

• Participants provided similarity ratings for visual stimuli (with defined perceptual dimensions) and non-visual stimuli (animal names) using four distinct methods.
• The four data collection methods included: traditional pairwise comparisons, the spatial arrangement method (SpAM), and two novel hybrid approaches.
• SpAM involves participants simultaneously viewing stimuli and arranging them spatially on a screen, with inter-stimulus distances reflecting subjective similarity.

Main Results:

• The spatial arrangement method (SpAM) generated high-quality multidimensional scaling (MDS) solutions that were comparable to those derived from pairwise comparisons.
• Monte Carlo simulations indicated that SpAM is robust to data degradation, including reduced sample sizes and lower granularity of similarity judgments.
• Coordinates obtained from SpAM solutions demonstrated predictive accuracy for object discrimination in same-different classification tasks.

Conclusions:

• The spatial arrangement method (SpAM) provides a fast, efficient, and user-friendly approach for collecting similarity data, significantly reducing experimental time.
• SpAM is a reliable method for generating accurate and robust multidimensional scaling (MDS) spaces, even with complex or less-defined stimuli.
• Utilizing a spatial medium for similarity measurement offers distinct advantages for cognitive and psychometric research.