My Master's thesis treated the usefulness of animated transitions for storytelling with data. The research was conducted in collaboration with Dr. Nicole Jardine and Dr. Steven Franconeri from the Visual Thinking Lab at Northwestern University.
Research on data visualization has traditionally focused on maximizing the throughput of information through the reader’s visual system. This is useful for researchers who need to explore large amounts of data but it is very strenuous. Narrative visualization, on the other hand, is a new approach to visualization that uses storytelling techniques to reduce the mental burden of the reader.
Charts as a story
One of these techniques is to guide the reader by organizing multiple charts in a sequential, narrative order. While this type of presentation splits information processing into manageable chunks, the reader now needs to understand how the individual charts are related. In practice, animated transitions are often used for this. But existing research indicates that animation might not be an effective way to present abstract relationships.
Do animated transitions help?
This thesis aims to find out if animated transitions have a positive impact on the readers understanding of narrative visualization. Based on a review of the literature on transition understanding and animation, we introduce the concept of characters to narrative visualization. We propose that readers understand transitions by comparing how characters and their setting changes between two charts. The strengths of animation were evaluated based on how they would serve this process. This has led to the formulation of several hypotheses.
- Animation makes understanding transitions easier Based on the widespread use of animation in
practice and based on the discussion in this chapter, we hypothesize that animation would reduce the
cognitive load of readers. - Animation implies a (causal) relationship
- Animation highlights the important characters
- Animation leads to object constancy
- Animation leads to higher engagement
Experiment
An experiment presented 8 transitions, half of them animated, to 56 participants to test these hypotheses. We have used Amazon Mechanical Turk to recruit the participants. Each participants saw 2 narratives with 4 transitions each. In one of the narratives, the transitions were static, in the other they were animated.
After each narrative we asked the participants to write a short summary of what they understood and rate their engagement.
Results
Animated transitions have indeed highlighted the important characters as they were mentioned more frequently by participants. Animated transitions also reduced back-and-forth-scrolling and thus support object constancy. No significant difference could be found in perceived engagement and understanding. The results thus confirm existing research on the use of animation within the context of narrative visualization.
Technical implementation
To create the stories, a small Typescript-library called Morphinchart was implemented. Also a logger that tracks the participants scrolling behavior as can be seen in one of the narratives.