In the fog of an unfolding global pandemic, policymakers and communicators need to better understand the effect of their words and actions on people’s perceptions.

As the worldwide coronavirus pandemic began to gain wide public attention, we examined data gathered from more than 12 million publicly available social media interactions over the two-month period from mid-January to mid-March, identifying the most impactful topic trends based on well-recognized drivers of risk perception and emotional impact.

This is the same analysis that is used by major corporations and government to measure group public perception for marketing and brand reputation management purposes.

What became initially apparent is that there is not one, single “national conversation” about coronavirus. Instead, there is a divergence of voices and messages that, for our study purposes, is most manageable when reduced to state-by-state analysis.

By overlaying these millions of conversations and the reactions they caused on a state-by-state basis, we learned that some states with little or no coronavirus activity ironically experienced higher perceptions of risk than those states with the highest number of recorded cases. Conversely, some states with the highest recorded number of cases had the lowest perception of risk.

This is significant to decision makers and public health officials because a state whose citizens perceive themselves at lower risk might be less inclined to follow public health guidelines. Conversely, those states with a disproportionately high perception of risk could easily be driven to hoarding and more extreme restrictions on social interaction than are recommended.

Public health communications strategies cannot be planned simply along national lines because populations at the state and local levels see things far differently than the nation as a whole. The same message will cause far different reactions in different places.

What’s more, these millions of conversations are happening constantly, so the data is changing on a minute-by-minute basis.

The state of Washington, for instance, had one of the highest number of cases early in the pandemic, but ranked only 37 out of 50 by risk perception. Just 24 hours later, however, Washington’s perception of risk had changed relative to the rest of the nation, moving its risk perception ranking from 37 to 22.

Only by constant monitoring of the data can decision makers and communicators better understand the risks and consequences of their words.

What are the topics causing people to believe they are at greater or lesser risk? To find out, we went beyond measuring the volume of a given topic and studied how much impact these discrete topics were having.

Not surprisingly, the topic that had the highest volume and impact involved discussions of updates to the situation as the pandemic unfolded.

The other topics that had more than 1 million hits and were rated to have high impact were almost all political in nature: criticisms of Donald Trump; Nancy Pelosi’s views on remote voting or leaving Washington; criticisms of Mike Pence; the developing coronavirus death toll; the pandemic’s impact on airlines; and support for Trump.

Other topics that heightened risk perception included assertions that coronavirus is a biological weapon; the appointment of Pence to lead the federal coronavirus response; the notion that any person coughing in public might have coronavirus; and the false claim that drinking bleach is a cure.

Discussions of preventive measures, meanwhile, drew 800,000 mentions and had a very calming emotional impact. Other divergent topics that lowered risk perceptions, whether true or not, included comparisons of coronavirus to other infections; calling coronavirus a “hoax”; assertions that coronavirus is under control; assertions that Christians need not fear the virus; and claims that Democratic policies kill more people than coronavirus.

Public health officials have been clear that we have still not seen the worst of this pandemic, and, in any event, we know there will be others in the years to come.

As we continue to navigate this crisis and as we plan for future ones, policymakers and public health officials should be aware that the public has a highly segmented and rapidly evolving perception of what is important and what are the risks posed by this or any other epidemic.

Only by knowing what are the issues that resonate and how these messages are changing can they effectively address this issue