Enhancing public health communication and decision-making: evaluating the efficacy of bivariate choropleth mapping using eye tracking

Imagine making decisions during a crisis with only one piece of information. During the COVID-19 pandemic, we often saw single metrics like total cases or deaths on choropleth maps, which use different shades or colors to show values of a variable. The novelty and spread of the disease highlighted the need for better visualizations of public health data. In that vein, I developed a multivariate visualization web app for COVID-19 data in nursing homes. Initially an interactive map, it became cumbersome with added data and visuals. Users preferred information directly on the map, leading me to research and develop bivariate choropleth maps.

Bivariate maps combine two different types of information on one map, using colors and patterns to represent multiple variables simultaneously. In contrast, traditional maps show one piece of information at a time.

In my research, I compared bivariate maps with traditional maps. Using eye-tracking technology, we observed how 48 people interacted with these maps. Half used bivariate maps, and the other half used traditional ones. We found that those using bivariate maps finished tasks faster and with more straightforward search patterns for specific regions or anomalies. However, understanding detailed information about specific areas was similar for both types of maps.

In a second study with 52 participants, we explored adding interactive features to these maps. Interactive maps allow users to engage with the data, such as hovering on regions for more information, or highlighting data displayed. On the other hand, static maps do not have these interactive features and display information in a fixed format. Interactive maps helped people detect anomalies more accurately and improved their search strategies without taking extra time. Both bivariate interactive and static map users showed better accuracy and efficiency over time.

These findings suggest that bivariate maps can make it easier and quicker to extract information from complex data, especially with interactive features. This research helps us understand how to improve tools for visualizing complex geospatial data, their relationships, and differences, making it more accessible and useful for everyone.