Understanding the role of interactive visualizations on consumer multi-attribute choice making

Sung-Hee Kim, Purdue University

Abstract

Multi-attribute decision making is the process of selecting an alternative from a set with several associated attributes. The challenging aspect of this type of decision making is that humans have limited cognitive ability to process large amounts of data. Information visualization techniques have been used to develop visual aids to help consumers process the given information, thus improving decision outcomes by increasing decision quality. However, not much empirical explanation exists regarding the process by which visual aids help the decision-making process. Therefore, the goal of this dissertation is to provide evidence as to how visual aids enhance decision outcomes by examining their role in information processing in addition to showing that it enhances decision outcomes in a real context. To achieve this goal, a series of experiments were conducted with a visual aid named SimulSort. Conducting studies with SimulSort provides an advantage over its nonvisual counterpart Typical Sorting. Three studies were developed to trace visual aid performance using different technological tools and scientific measures. Study 1 examined how SimulSort impacted decision-making strategies. An eye-tracking study revealed that SimulSort supports more efficient browsing behavior with shorter fixation duration and higher fixation count than Typical Sorting. This quick browsing behavior might have promoted compensatory decision-making strategies, giving an advantage to the more efficiently designed visual aid. An additional crowdsourcing-based user study yielded results supporting that participants using SimulSort could perceive more information throughout all the attributes. Study 2 investigated how decision outcomes improved among participants who demonstrated the behaviors revealed in Study 1. Because task performance is known to be affected by the limitations of a person's working memory, two crowdsourcing-based studies were conducted to understand the role of working memory through SimulSort. The results supported the hypothesis that when using SimulSort compared to Typical Sorting, participants relied on added visual information which required less working memory to accomplish a particular decision-making task. The experiments for Studies 1 and 2 used context-free data to minimize the impact of participants' preference structures and focus the study solely on information-processing behaviors. For example, attributes were labeled as A, B, C, while data with context might label the same attributes as price, weight, and star rating. Therefore, Study 3 was conducted with data with context to see how visual aids could have an impact in a more practical setting. The results supported the supposition that SimulSort leads to a higher chance of making a good choice, by selecting a nondominated alternative over a dominated alternative. Mouse interaction data revealed that this may have been possible because participants could evaluate a larger number of alternatives using SimulSort.

Degree

Ph.D.

Advisors

Yi, Purdue University.

Subject Area

Industrial engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS