Twitter data reveals partisan divide in understanding why pollen season's getting worse

Two things are clear from a University of Michigan analysis of nearly 200,000 Twitter posts between 2012 and 2022. 

One, people are really good at identifying peak pollen season: The largest volume of tweets about pollen often lined up with pollen counters hitting their biggest numbers. And two, liberal users on Twitter were more likely than conservatives to ascribe shifting pollen seasons over the years to climate change.

"There is a partisan gap in how we perceive the pollen seasons that are very relevant to our day-to-day life and even our health," said Yiluan Song, the lead author of the new report in the journal PNAS Nexus. Song is a postdoctoral fellow in the Michigan Institute for Data and AI in Society, or MIDAS, and the School for Environment and Sustainability, or SEAS. 

While the team understands that the divide itself might be what people latch onto most in today's political climate, it is the second part of Song's statement that has the researchers' focus. People have an intimate relationship with pollen that's different from other climate change impacts, they said, such as the increasing average global temperature.

"We already know that people's political beliefs shape how they think about climate change—that's been shown in many studies. But what's new here is that more and more people are struggling with pollen allergies," said Kai Zhu, a senior author of the new study and an associate professor at SEAS. 

"When people feel the impact of pollen in their own lives—sneezing more or cleaning pollen off their cars—it becomes personal. That experience can help connect the dots to climate change, which can feel abstract or distant. Instead of talking about a degree or two of global warming, we can focus on things people face directly in their daily lives."

The new study was supported, in part, by funding from the U.S. National Science Foundation. The study's other U-M authors included Nathan Fox, an AI scientist with MIDAS; Derek Van Berkel, an associate professor at SEAS; and Arun Agrawal, an emeritus professor at SEAS. Adam Millard-Ball, a professor at the University of California, Los Angeles, was another key contributor.

People remember pollen

As the planet warms due to climate change, pollen seasons around the country are starting earlier and lasting longer. That means not only more sneezes, runny noses and watery eyes for people with seasonal allergies, but also heightened risks for more serious reactions that can result in hospitalizations. A recent U-M study even linked high pollen counts to increased suicide risks.

Beyond the health concerns, people also express frustration over needing to spend more time and money cleaning their cars, patio furniture and other belongings thanks to more intense pollen seasons. In fact, Song said, folks who tweeted about increased pollen were often doing so in this context. And, whether or not they made the connection to climate change, a lot of people were tweeting about pollen every year. This annual nature of pollen season is another helpful feature on the communication front, she said.

For its study, the researchers analyzed a dataset containing more than 190,000 tweets posted between 2012 and 2022 within the United States (Twitter rebranded as X in 2023, when it also made changes in the availability of its data). The team first compared the count of pollen-related posts to pollen concentration data and found that tweet volume correlated with pollen counts. That is, people are reliable "social sensors" of pollen, Zhu said.

"We found that social media is surprisingly accurate at reflecting what’s really happening with pollen levels around the country," he said. "People posting about pollen closely matched actual pollen counts."

"Temperature changes day to day and it can be hard to remember exactly when a really warm day was," Song said. "But we have deep memories for annual events. For example, Atlanta had record-breaking pollen peaks this year, as well as in 2012 and people remember that."

The team next analyzed the content of sampled posts and scored users on a liberal to conservative gradient based on the accounts they follow. The team found that conservative users were more likely to talk about pollen and attribute it to changes in temperatures, while liberal users were more likely to go a step further and attribute it to climate change.

The team also grouped the tweets by their sources, categorizing whether they came from members of the media, scientific experts, organizations or other individuals. This allowed the team to follow the flow of information, which highlighted other communications insights.

"When we look at the more technical discussions about how climate change drives pollen season change, a lot of it's dominated by scientific experts and the media," Song said. "There are pros and cons to this. It shows scientists are getting messages out there about climate change, but, on the other hand, this knowledge production is very top-down. Perhaps the scientific community could do more to encourage bottom-up production."

Moving forward, the team plans to broaden the impact of their ecological research with the insights from this study. The work was also supported by Schmidt Sciences and the University of California Santa Cruz Belonging Program.