DIMENSIONAL INTERPRETATION AND CONFIGURATION INVARIANCE IN MULTIDIMENSIONAL SCALING: AN EMPIRICAL STUDY.

Area of Science:

• Psychology
• Marketing Science
• Data Analysis

Background:

• Multidimensional scaling (MDS) is a technique used to visualize perceived relationships among stimuli.
• Interpreting the axes of MDS configurations can be challenging for researchers.
• Understanding the stability of MDS solutions across varying stimulus domains is crucial for reliable analysis.

Purpose of the Study:

• To investigate the invariance of nonmetric multidimensional scaling (NMDS) solutions when the stimulus domain changes.
• To explore methods for interpreting the axes of NMDS configurations using external data.
• To compare the effectiveness of similarity judgments, semantic differential scales, and preference data in constructing MDS configurations.

Main Methods:

• The study employed nonmetric multidimensional scaling (NMDS) with 17 popular automobile brands as stimuli.
• Similarity and preference judgments were collected from 37 subjects, divided into two groups with overlapping stimulus sets.
• Semantic differential scales were used to gather property information for each stimulus.

Main Results:

• NMDS solutions demonstrated stability, with interpoint distances for a core set of stimuli remaining consistent across different subject groups and stimulus set compositions.
• Semantic differential ratings successfully identified directions in the configurational space that correlated highly with external property vectors.
• Configurations derived from semantic differential data closely matched those from direct similarity judgments.

Conclusions:

• NMDS solutions exhibit invariance to changes in stimulus domain, ensuring robust results.
• Semantic differential data provides a reliable method for interpreting MDS axes, enhancing the understanding of stimulus configurations.
• Preference data, when used alone for unfolding, did not yield configurations as congruent with similarity-based analyses.