Testing Data Visualization as Scientific Communication

Two people in a dark room standing in front of a large data display

Problem

People may not read data charts the way designers think they do.

These days, one cannot look at the news or social media without coming across a data visualization or infographic because they are increasingly being used across many fields and platforms to share information or tell a story—scientific or otherwise. While most data designers are likely doing their best to present the data accurately, there is not enough consistent, empirical evidence to inform data visualization best practices.

While studies have examined the utility of different types of charts and visualization methods for many years, these studies frequently rely on subjective theories, lack scientific rigor, or both. Moreover, samples are often limited by size or method, leading to limited and possibly biased conclusions about the best way to present data visually and accurately.

Solution

NORC is testing how people perceive and interpret various data visualization methods.

Since 2022, we have been using a nationally representative sample of US adults to test data visualization methods. As of 2024, we had conducted more than 15 rounds of data collection from 1,000+ respondents per round, studying their perceptions, understanding, and accuracy. We present them with many different types of charts to gauge how well color, visual elements, annotated cues, and other components help with judgement and interpretation.

Our goal is to understand the utility of data visualization as a form of scientific communication and measure the depth of users’ understanding about what the charts actually mean.

Result

We’re finding significant variation in how people understand data graphics.

Our research to date suggests that to better communicate data through graphics, we must consider the needs of various users. One of our most notable findings so far is that the level of understanding differs by demographic subgroup. Specifically, those who are less educated are less able to accurately interpret some charts. Considering the preponderance of data graphics in daily life, this speaks to issues of data access, equity, and the ability of people to make decisions that affect their lives.

We also found that user accuracy is also affected by color, chart orientation, size, and other visual elements. Our data also illustrate differences in both the level of engagement with graphics and attitudes towards them as a form of communication.