Category Archives: Social Networks

Social Networks and Evacuation

This blog post is related to a recently accepted paper titled “Simulating the effects of social networks on a population’s hurricane evacuation participation” in the Journal of Geographical Systems. Link to come soon. 

It’s trite to say that the world is becoming an increasingly connected place. But all of these interconnections, many enabled by popular social media technologies like twitter and facebook, can impact how we communicate and spread information throughout our environments – and subsequently affect behaviors.

This is exemplified by a recent, playful ad from Twitter claiming they’re faster than an earthquake.

While the video is largely tongue-in-cheek, this blog entry at Social Flow provides a more analytical breakdown of how the twittersphere reacted to the August 2011 quake in Virginia, where tweets from DC beat the earthquake to New York City.

In my mind, the reaction of many people to tweet in the face of a relatively rare and potentially threatening natural event raises a couple of questions. Is social media becoming so ubiquitous that people will turn to their computers and mobile devices for information about a natural or man-made disaster?  And perhaps more importantly – how might information from social media communications that originate from connections in very different geographic contexts influence a person’s behavior in an evacuation situation?

In regards to the latter question, some disasters happen so quickly that there may not be enough time to significantly change a person’s behavior – more than, say, simply raising your coffee cup to avoid it spilling during a surprise earthquake. However, if you consider potentially hazardous events like hurricanes, where there is usually sufficient amounts of time to implement region-wide evacuations, social media communications could play an important role in influencing people’s behaviors.

I teamed up with Mark Horner at Florida State University and Sara Metcalf at the University at Buffalo – SUNY to build a computer simulation, where households in Bay County, FL considered evacuation given demographic traits (like owning a vehicle), environmental context (like having a home in a known flood zone), and social network information.

Bay County, FL. Red dots represent individual households that have decided not to evacuate at the end of the simulation (the time at which the hurricane strikes).

We explored how evacuation participation varies across the region with three different social network structures – spatial proximity networks, small world networks, and spatially random networks (the structure that most readily replicates a network with many social media connections). After running the simulation using a number of different assumptions about important parameters (like the weight households place on social networks and the average number of connections of each household) we found that random networks (the type of networks social media technologies help create) encourage the diffusion of information across larger spatial regions more quickly, and can greatly increase the number households participating in evacuation.

While more research is needed to understand how seriously people consider information from social media versus person-to-person communications, there is an opportunity here to devise new strategies for getting good information out to the people who need it most. However, emergency management agencies need to tread carefully, as passing information through such a complex network with disaggregated actors could just as easily result in the spread of misinformation.

Dr. Michael Widener is an Assistant Professor at the University of Cincinnati’s Department of Geography. Dr. Mark Horner is an Associate Professor at Florida State University’s Department of Geography. Dr. Sara Metcalf is an Assistant Professor at the University at Buffalo’s Department of Geography. Please contact Dr. Widener with any questions.