WHERE DID THE VOTERS GO?

This presents an interactive geo-visualization of my original static design. Visit my behance page to explore the static version. Download your free copy of the static version here. The data for these visualizations comes from the University of Florida Election Lab.

In the 2024 US Presidential Election, 3.63 out of 10 voter-eligible Americans did not vote. Another way of saying this is voter turnout was 63.7%. Believe it or not, this is considered high for a US Presidential Election. It was, however, lower than 2020, when the national turnout was 66.6%. This series of visualizations explores the 2024 average voter turnout for each state, and highlights the difference(+/-) in turnout rate from 2020 to 2024.

We geo-visualization experts excel at designing data-driven charts, graphs, and - of course - maps. As spatial scientists, it's hard to abandon the map when the data itself is spatial, even if some data stories are best conveyed through non-map visualizations. At the very least, we can admit that charts can enhance a story—even when the data has a spatial component.

The following three visualizations began as a static visualization. In that first version, I created only charts, since a map would not have provided additional insights into the data’s story. That is to say, although states are geographic, the pattern of voter turnout rates by state is not inherently spatial. However, most people feel a personal draw to their own geography, and thus maps can help pull people into a dataset’s story. For this reason among others, I decided to move forward with this d3.js map.

Choosing the best visualization for your purpose

Below are three separate visualizations. Although each visualization tells the same story, they are quite different from one another.

1. Map. Maps pull us in, and spatially situate this dataset’s story. The interactivity enhances comprehension by allowing users to explore details dynamically through tooltips, reducing clutter while preserving clarity.

2. Graduated symbols. The simple graduated symbology minimizes the real estate of the visualization, which may be ideal in certain circumstances. Again, interactivity helps this minimal visualization tell a deeper story.

3. Toggleable vertical bar chart. This is my preference for this particular dataset, especially with the capacity to toggle between the two variable. In doing so, viewers get a visual of each state's ranking.

For each visualization, the categories are identical, and the color symbolizing these categories remains the same throughout.

These visualization is best viewed on desktop and tablet.

Presidential candidates typically concentrate their final campaign efforts on "swing states," which are critical battleground states, hosting multiple rallies to mobilize voters. In contrast, non-swing states often receive little to no candidate visits during the election cycle. Campaign effors, such as rallies, are intented to get voters to the polls. For this reason, it makes sense that in 2024, nearly all swing states had above-average voter turnout. The exception was Arizona, which fell below the national average.

This bar chart allows users to toggle between two data perspectives: changes in voter turnout from 2020 to 2024 and each state's variance (+/-) from the national average.

Because these visualizations focus on state-by-state ranking, the original static view prioritized vertical bar charts to depict voter turnout changes and differences from the national average. Without interactivity, a map would not have provided additional insights into these patterns, leading to its omission from the static view. Although I could have treated these four multi-variate categories as quantitative, I deliberately chose a qualitative color scheme. This is not a matter of right versus wrong in symbolization, but rather a deliberate design decision rooted in the nature of the data and the goals of the visualization. While some may prefer rigid frameworks for defining "best practices," effective visualization requires adaptability to context and audience needs. In the case of the above d3 charts, interactivity enhances comprehension by allowing users to explore details dynamically through tooltips, reducing clutter while preserving clarity.

Part of the excitement of geo-data visualization is uncovering patterns. Like democracy itself, the patterns can be messy. In the best way, of course. State-level policies vary widely: some states do not offer online registration, while others make registration accessible online and in person. Some states provide voters with a four-week window to cast their ballot, compared to states that require most voting to occur only on Election Day. These and other differences in voting access across states correlate with turnout rates.

However, this visualization is designed to highlight two specific characteristics for each state: its increase or decrease in voter turnout percentage and whether turnout per state was above or below the national average. The map categorizes states based on these factors, as reflected in the legend. While it does not delve into the underlying causes of these clusters, the cluster of states per each category suggest that the Electoral College plays at least a partial role. Every state that saw an increase in voter turnout while also exceeding the 2024 national average turnout is a swing state. Among them, Michigan, Wisconsin, Pennsylvania, and Georgia stand out, topping both categories in voter engagement. The existence of swing states itself is a direct result of the winner-take-all Electoral College system. Again, this is a side project because I enjoy exploring our democracy landscape with data visualization. If it were a fully funded project with time for futher research, I presume I would find that the vast majority campaign events took place in swing states as well.