Understanding the Role of Visualizations on Decision Making: A Study on Working Memory

Sung-Hee Kim

doi:10.3390/informatics7040053

Abstract

The use of data visualization is increasing; however, there is little empirical explanation for how it supports users. Our goal in this paper is to deepen our understanding of the role of interactive visualizations in a particular context of decision making. Specifically, we attempt to understand the role of the working memory system, which is a concept to understand the mechanism of the processing and temporary storage of information in variety of cognitive tasks. We compared two interfaces, SimulSort and its non-visual counterpart Typical Sorting, with a multi-attribute decision-making problem. Because decision outcomes are known to be affected by the limitations of a person’s working memory, we conducted a crowdsourcing-based user study using SimulSort to understand how working memory, especially the phonological loop, can benefit from the using visualizations. We examined the impact on working memory with a well known dual-task methodology by designing a concurrent task to tap into the main decision-making task. The experiment was conducted with a total of 137 participants and an ordered logistic regression using a proportional odds model was applied to analyze the decision quality. The results supported the hypothesis that when using SimulSort, participants required less working memory than they required with Typical Sorting to accomplish the multi-attribute decision-making task even though SimulSort outperformed Typical Sorting in terms of decision quality. We also provide methodologies to conduct working memory studies by implementing an articulatory suppression task on crowdsourcing platforms in which experimenters have less control over the participants.

Highlights

Data visualization is widely used for understanding public data through dashboards [1,2,3], utilized in mobile services to understand personal data [4,5], and is gaining interest even for K-12 education [6,7]
Working memory consists of various components that can hold a limited amount of transformable information for a finite period of time [21]
A well-known model suggests that the working memory consists of two temporary memory systems, a phonological loop and a visuo-spatial sketchpad [22]

Summary

Introduction

Data visualization is widely used for understanding public data through dashboards [1,2,3], utilized in mobile services to understand personal data [4,5], and is gaining interest even for K-12 education [6,7]. Numerous studies [9] have demonstrated that visualizations help people find novel insights and serendipitous findings. These findings could be results of amplified cognition, but it is difficult to explain how visualizations amplify cognition. Multi-attribute decision making occurs when the decision maker has to select one option from a set of alternatives, each of which has a set of attributes [30] This is a common situation in which one can organize the data in tabular form [31]. The consumer may want a car with a lower price, lower fuel economy, higher horsepower, and higher consumer ratings They need to compare all the attributes and one can collect more information about the car to consider. Giving decision-makers more information to consider does not mean that they will make better decisions

Objectives

Methods

Results

Discussion

Conclusion