The design of usable haptic icons (brief informational signals delivered through the sense of touch) requires a tool for measuring perceptual distances between icons that will be used together as a set. Our experiences with one potentially powerful approach, Multidimensional Scaling (MDS) analysis of perceptual data acquired using an efficient cluster sorting technique, raised questions relating to the methodology for data collection. In this paper, we review key issues relating to perceptual data collection method, describe an example data set and present its initial MDS analysis, and then examine the impact of collection method on MDS outcome through a secondary analysis of the data and the inherent structure of the algorithm components. Our analysis suggests that an understanding of these issues is important for the method?s effective use, but has not exposed any major flaws with the process.