A Comparison of Animated Maps with Static Small-Multiple Maps for Visually Identifying Space-Time Clusters

Abstract

Although animated maps are widely promoted as ideal vehicles for learning and scientific discovery, there has been little empirical work that demonstrates their relative effectiveness in relation to static small-multiple alternatives. In this article, we attempt to clarify the issues related to the potential of animation from an explicitly geographic perspective, but one that is also grounded in broader cognitive science and human-computer interaction considerations. We compared the effectiveness of animated with static small-multiple maps, specifically in relation to map readers' ability to identify clusters that move over space and through time. In this study, we focused on several factors that might impact (or help explain) map readers' ability to correctly identify clusters. These factors included animation pace, cluster coherence, and gender. We found that map readers answer more quickly and identify more patterns correctly when using animated maps than when using static small-multiple maps. We also found that pace and cluster coherence interact so that different paces are more effective for identifying certain types of clusters (none vs. subtle vs. strong), and that there are some gender differences in the animated condition. This study is one of a small number of controlled experiments directed to the relative advantages of animated and static small-multiple maps. It provides the basis for further research that is needed to better understand the cognitive load involved in reading animated maps, to better describe and understand gender differences, and to investigate the efficacy of animated maps for other types of map reading tasks.

Key Words:

• map animation
• small-multiple maps
• visual cluster detection

Acknowledgments

Funding for this work was provided by National Science Foundation Digital Government Grants 9983451, 9983459, and 983461. The authors would like to thank Dr. Mark Harrower for his comments on the initial experimental design; Brian Pacheco, Geoff Hatchard, and Derek Swingley for their assistance with data collection; the Penn State University students who participated in the research; and the four anonymous referees who provided helpful and constructive critiques of earlier drafts of this article.

Notes

1. Interactivity itself has been shown to increase students' ability to learn, so comparing interactive animations with noninteractive static small-multiple visual representations would confound interactivity and animation. For this reason, Tversky, Morrison, and Betrancourt (2002) decided to exclude studies of interactive animations from their review.

2. Clusters can be considered to be either (1) concentrations of events surrounded by nonevents or (2) areal concentrations within which the mapped phenomenon exists at either lower or higher intensities than surrounding areas.