Researchers compared new and renewed flood insurance policies in the wake of the storms, both in areas that experienced active flooding and those unaffected by the hurricanes. They relied on the Facebook Social Connectedness Index, which measures the probability that two individuals across two locations are connected on Facebook, to estimate “social learning” about the floods.

People in areas with stronger social ties to flooded areas purchased more new flood insurance policies in the three years after Harvey and Irma, they found. The researchers estimated an additional 250,000 policies in areas where flood emergencies were declared after the hurricanes and 81,000 more in unflooded areas in the wake of the disasters, with six times as many new policies as renewals.

“Our evidence of social learning suggests that a short episode of a regional climate disaster can stimulate persistent adaptation behavior in the entire social network up to 3 years after the disaster,” they write. As a result, they suggest, disaster awareness and climate mitigation efforts should take social learning into account, leveraging people’s tendency to learn about climate change risks from their friends’ experiences.

Social learning could be particularly important when it comes to climate change, Yilan Xu, a professor of agricultural and consumer economics at the University of Illinois at Urbana-Champaign who is a co-author of the study, said in a news release.

“It’s a very politically fraught issue for some people,” Xu said. “Even though it can be really difficult for you to convince someone else that climate change is real, if they see their friends and family experiencing its negative consequences, that’s a prime opportunity for the skeptics to update their thinking about climate change.”

Flooding is likely to increase along with climate change, climate researchers project. According to the U.N. Development Program, coastal flooding will increase fivefold this century, while National Oceanic and Atmospheric Administration scientists predict “rapid intensification” in tropical cyclones in the coming years.