Watching the Election Sausage Get Made: How Data Journalists Visualize the Vote Counting Process in U.S. Elections

Document Title

Watching the Election Sausage Get Made: How Data Journalists Visualize the Vote Counting Process in U.S. Elections

Document Information

• Subject Area: Data Journalism and Visualization of the U.S. Election Process
• Keywords: U.S. Elections, Progressive Visualization, Uncertainty Visualization, Dashboard Visualization, Design Space, Qualitative Study

Research Background and Issues

• Identified Challenges:

  • Vote counting in U.S. elections takes a significant amount of time due to the decentralized nature of the election process, leading to complexities in real-time digital visualization and challenges in managing public uncertainty.
  • Real-time election results may mislead the public into forming incorrect perceptions, such as interpreting vote count "surges" as evidence of election fraud.
  • Designing real-time election result charts requires balancing "uncertainty" with "data integrity" while meeting readers' demand for immediate information.

• Significance:

  • Real-time election result visualizations have become a critical window for the public to understand the U.S. election process, often coinciding with peak website traffic. These visualizations not only display interim vote counts but also bear the responsibility of explaining procedural details and addressing public concerns about election fraud.

• Research Motivation and Related Work:

  • Historically, quantitative data has gradually permeated journalistic practices, such as in opinion polling and real-time election forecasting. Real-time election result visualizations are a core case of applied data journalism, involving the design of intuitive dashboards for dynamic and uncertain data to facilitate public understanding and information access.

Solution

• Proposed Approach:

  • By studying dashboard designs from 19 U.S. news organizations and election result providers, a visualization vocabulary was created to describe the 2020 and 2022 U.S. elections. Qualitative interview studies were then conducted to gather feedback from data journalists on design goals and challenges.
  • A design space for real-time election results was constructed, showcasing how these designs address uncertainty management (e.g., critical thresholds for hiding partial data), enhance user understanding, and respond to voter skepticism about the election process.

• Innovations:

  • Systematically categorized real-time election result dashboards for the first time, integrating data journalism practices with mainstream visualization design theories.
  • Explored visualization methods that balance user understanding of uncertainty with the dynamic nature of data, particularly addressing challenges in the complex vote-counting process (e.g., time lags and incomplete data).

• Implementation Steps and Key Techniques:

  • Data Sampling: Analyzed dashboard designs from the 2020 U.S. Presidential Election and the 2022 Midterm Elections, capturing page components in real time.
  • Qualitative Interviews: Interviewed 13 data journalists and designers to discuss major challenges from design choices to final goals, as well as strategies for addressing voter identity and regional data acceptability.
  • Inductive Design Vocabulary: Developed a systematic method to describe dashboards' multi-layered data, visual encodings, and the dynamic flow of data as the election process unfolds.

Research Findings

• Specific Outcomes:

  • Created a two-part visual vocabulary summarizing how election result dashboards encode dynamic data and uncertainty. Through systematic analysis of design components, dashboard functions were categorized into two main types: displaying vote counts and illustrating the election process.
  • Provided dominant visual design principles for real-time election result dashboards, such as balancing data uncertainty with visualization clarity and selecting appropriate colors/graphic styles to represent the dynamic nature of partial data.

• Advantages:

  • Compared the design approaches of different news organizations in presenting real-time election data (e.g., progress bars, forecasts, visualizations of data flow) and summarized best practices for managing uncertainty and optimizing user interfaces.
  • Offered a reference framework to promote more intuitive and comprehensible election data publication designs, providing theoretical support for reducing misleading or confusing visual representations.

• Experimental or Evaluation Results:

  • Analysis of dashboards and qualitative interviews confirmed several methods for quantifying or minimizing public uncertainty in visualizations (e.g., using gray backgrounds to hide early vote count data rather than displaying it outright). It also revealed that many news organizations were constrained by time and resources, limiting their ability to implement more complex predictive models.

• Limitations and Future Directions:

  • Limitations: Did not include real-time election reporting formats in television broadcasts and print media, thus failing to cover all visualization design constraints. Did not comprehensively study user experiences and behaviors during interactions with dashboards.
  • Future Directions: Suggested conducting user studies to evaluate the impact of dashboard designs on audience trust and understanding of elections; exploring how dynamic data can be better integrated into real-time visualizations of other complex societal issues (e.g., pandemic data).