Overview

During the COVID-19 pandemic, visualizations became one of the most important tools for understanding the world. From dashboards tracking cases to animated charts in the New York Times, these graphics became part of our daily lives—sometimes shaping behavior as much as the underlying data itself. This project explores how visual storytelling evolved under crisis conditions, and how that evolution impacted how we viewed the pandemic. We analyze 173 data-driven articles from the “What’s Going On in This Graph?” series published between 2015 and 2024. Our goal was to explore how the urgency, complexity, and volume of pandemic data reshaped visual design practices—both in terms of graphic form (e.g., interactivity, color use, complexity) and thematic focus (e.g., health and risk communication). We also wished to explore how these changes under pressure may have affected how we viewed the pandemic. By combining historical analysis, literature review, and our own empirical study, we provide a timeline of this transformation and reflect on its broader implications.

Skills and Tools Used

Combined R-based data wrangling, visualization, and content analysis to examine 173 New York Times articles from 2015–2024. Merged article-level and graphic-level data using shared identifiers, applied time-based filtering (pre/post-COVID), and used packages like ggplot2, dplyr, and tidyverse to track changes in graph types, design elements (color, motion, Gestalt principles), interactivity, and perception tasks. Explored topical trends using keyword classification to evaluate shifts in visual storytelling, especially in pandemic-related reporting. Above is a comprehensive toolkit for our data exploration, visualization, and reporting in R. We also focused on qualitative coding of design principles and topics.

Data Process

Stat Nuggets

We began by collecting all available “What’s Going On in This Graph?” articles published by the New York Times Learning Network in collaboration with the American Statistical Association. These weekly publications were designed to promote visual literacy in classrooms, but they also provide a consistent, publicly accessible record of data visualizations in media from 2015–2024.

Check it out!

Data Collection and Cleaning

After compiling 173 articles, we manually coded metadata for each graphic: type (bar, line, scatter, etc.), topic (e.g., health, economy, sports), presence of interactivity, design complexity, and use of Gestalt principles. Each graphic was also analyzed for its interpretive demands—such as number of perception tasks required and whether uncertainty (e.g., confidence intervals) was visually communicated. This data was managed through two dataframes as seen here. The articles dataframe contains metadata about each article, while the graphics dataframe focuses on the individual graphics within those articles.

Data Analysis and Coding

To assess changes over time, we split the dataset into pre-COVID (before March 2020) and post-COVID subsets. We are using R to compute trends and cross-tabulations—looking not only at which graphic forms became more common, but also how their design adapted in response to pandemic-related content.

Results

Although this project is still ongoing, it has already been a significant opportunity for growth. Being involved in every stage—from early development to strategic decision-making—has deepened my understanding of what it means to lead data-driven research and infrastructure. My goal is always to make data more accessible and to let it tell meaningful stories. The insights we uncover have the power to inform, inspire, and drive real-world impact—and that continues to motivate my work. Through this project, I gained a deeper understanding of how data visualizations evolve under the pressure of real-world crises. Our analysis of 173 New York Times graphics revealed clear shifts in both the form and function of visual storytelling during the COVID-19 pandemic. Post-2020 visualizations were more likely to feature interactive elements, greater visual complexity, and design choices that emphasized uncertainty and risk communication—often using color, motion, and Gestalt principles more intentionally than before. Thematically, the dataset showed a marked rise in health-related topics and multi-variable comparisons, reflecting the increasing need to communicate layered, time-sensitive data to the public. As a data scientist, this project strengthened my skills in data wrangling, visualization, and mixed-method analysis, while also reinforcing the importance of context in data communication. Working across both quantitative trends and qualitative coding sharpened my ability to extract insight from messy, narrative-rich datasets. I also learned how thoughtful design can shape perception—how a graph is read, understood, and acted upon—especially during moments of crisis. This reinforced a core takeaway: data storytelling is not just about clarity; it’s about responsibility. This project not only broadened my technical and analytical toolkit but deepened my commitment to ethical, human-centered data communication.