Assessing the Use of Sound in Graphical Data Analysis

• Activity Number: 465
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Computing
• Author(s): Jessica Orth*+ and Kate Cowles
• Companies: University of Iowa and University of Iowa
• Keywords: Auditory Display ; Regression Diagnostics ; Sonification

Abstract

There are many advanced and sophisticated graphical displays in common use by statisticians today, but there is little in the area of representing graphical displays through sound. We aim to introduce sound to graphical displays in the simple linear regression setting. Our goal is to use sound to introduce another dimension to complement visual tools and at the same time create an alternative to visualization that has a potential to help users with visual disabilities. Using sounds generated from the audiolyzR package, we analyze an online survey we designed to test users' perceptions of sound in simple linear regression diagnostics by means of the current software. Based on the analysis of our survey, we can conclude that users are able to correctly identify different diagnostic plots in simple linear regression through the sound medium, though this was more challenging for certain regression concepts than others. Areas of difficulty include identifying a non-constant variance relationship and outliers, based on our survey sample.