CSIM Master Thesis on Interactivity in Data Visualization

Carlos Ordóñez, who works for Bestiario, submitted his CSIM (Cognitive Science and Interactive Media) master Thesis on Interactivity in Data Visualization. The GTI accepted to supervise the research and offered Carlos the opportunity to be part of the group of data visualization research.
Carlos’ thesis deals with a missing topic in the fast growing research about Data Visualization, Interactivity. Current research about Interactivity is focused on studying the effects produced by changes in the User Interface. Carlos’ thesis is focused on the component-to-component interactivity instead.
Let’s explain what component-to-component interaction is with an example. Suppose that there is a data visualization layout where the countries GDPs are represented by choropleths (colors coded on a political map), and that those values are represented by another component, a sorted bar chart in the same layout. Furthermore, the user is able to highlight a country on the choropleth by clicking on it. If, when the user highlights a country on the choropleth, the same country is highlighted as well on the bar chart, we have Component-to-Component Interaction (CCI).
The CCI of a data visualization layout can be thought of as a network of interactions and this abstraction is called the Component-to-Component Interaction Network (CCIN). The thesis proposes a way of visualizing the CCIN, to analyze it.

Figure 1: Component-to-Component Interaction network of the popular data visualization layout Gapminder

The thesis explores whether the effectiveness and efficiency of a data visualization application can be enhanced through an experiment which compares two versions of the same application only differing on the respective CCIN. Two versions of an experimental application were designed and developed to find information about the Syrian Civil War.
26 people took part in the experiment, and the results indicate that introducing Interactivity enhances effectiveness, as the users were able to understand better the information, while the efficiency might be decreased, as the users needed longer time to explore the information.

Version without interactivity
Version with interactivity
Figure 2: Component-to-Component Interaction network of two versions of experimental application, the main difference is the extra connection between the news component and the range selector component

Adding the extra interactivity feature allows users to be more accurate but they have to invest more time in understanding how the application works and how they can take advantage to the feature. Thus, adding an interactive link between components, is interesting if the users could be considered as experts or they will be using the application during an extended time, but it is not convenient for a mainstream audience.