Journal of Public Health Management & Practice:
Research Brief Report
Hasan, Samiul MSc; Ukkusuri, Satish V. PhD
School of Civil Engineering, Purdue University, West Lafayette, Indiana.
Correspondence: Samiul Hasan, MSc, School of Civil Engineering, Purdue University, 550 Stadium Mall Dr, West Lafayette, IN 47907 (email@example.com).
The authors declare no conflicts of interest.
Individual evacuation decisions are often characterized by the influence of one's social network, referred to as informal warning network. In this article, a threshold model of social contagion, originally introduced in the network science literature, is proposed to characterize this social influence in the evacuation decision-making process, in particular the timing of evacuation decision. Simulation models are developed to investigate the effects of community mixing patterns and the strength of ties on timing of evacuation decision.
Complex decision-making processes at different levels (ie, individual, household, and community) of influence are involved in hurricane evacuations. Important decisions related to evacuation process include the following: whether to evacuate or not; when to evacuate; where to evacuate to; and which route to take. In addition to several individual and household-level characteristics, evacuation orders, and storm threat, the personal risk perception is the most important factor in determining the evacuation decision.1 Although the role of social influence on individual risk perceptions is not extensively studied, one can suspect that risk perceptions can be socially influenced as evacuation warnings spread through social networks. Official warning messages sometimes provide vague information that is usually confirmed through other sources (ie, through one's social network).2 Gladwin et al3 conclude that informal networks based on neighbors, coworkers, family members, and friends can influence the initiation of decision-making processes and the role of warning dissemination. In addition to that, strengths of social relationships may also play an important role in the social influence process of evacuation decision making.
The contagion model of warning propagation4 mimics a binary decision-making context where an agent (ie, an individual or a household) has to decide between 2 choices of whether to evacuate or not because of a hurricane warning threat. When a hurricane warning is issued for a population of agents, the contagion of this hurricane warning and the related evacuation decisions due to the warning are observed. Each agent follows a simple binary decision rule observing the current states (either 0 or 1, i.e., either not evacuated or evacuated, respectively) of κ other agents, which we call its neighbors, and adopts state 1 if at least a threshold fraction (φ) of its κ neighbors are in state 1, else, it adopts state 0. The neighbors represent the members of the agent's social network consisting of friends, relatives, and colleagues. To account for variations in risk perception and access to resources required to evacuate, the threshold value of an agent may be treated as heterogeneous. Each agent belongs to a particular community having a specific degree distribution. This mixing pattern in a population can be characterized by a quantity eij, which is defined as the fraction of neighbors in a network that connect an agent of type i to the agent of type j.
Results from simulation models suggest that as we decrease the intracommunity edges, which are equivalent to increasing intercommunity edges, warning propagates faster. As a result, the more the intercommunity edges, the faster the propagation of the warning. These intercommunity edges can play the role of bridges between the communities and let the warning information and the influence propagate faster across the communities. The propagation curve has different stages; the initial stage consists of the propagation of cascade within the community of the initial seed; in the subsequent phases, the cascade starts to propagate to the other community. When the proportion of intra- and intercommunity edges are equal, the whole network acts like a single community and, in this case, the rate of propagation is the highest. Given that communities are common in our societies, this analysis shows the benefits of having more connections among communities in the social network regarding the contagion process.
It is likely that different relationships will have different level of influence on an individual's decision. Here, we assign each tie a weight, using 2 categories of ties (ie, strong tie and weak tie). A strong tie (weight = 3) represents the relationship with a kin (eg, spouse, parent, sibling, child, or other family member), and a weak tie (weight = 1) represents the relationship with a nonkin (eg, coworker, friend, advisor, neighbor, or group member). In general, we observe a direct relationship between the strength of ties and the average cascade size. That is, the stronger the ties are, the greater the average cascade size is. This result is contrary to the finding of “the strength of weak ties”; weak ties may provide greater cascade size for simple contagion, but, when weights are introduced in the network, the strength of weak ties is no longer found. Results indicate that when the network has higher proportion of strong ties (eg, discussion with kin about evacuation decisions), the size of the evacuated agents will be larger than the network with smaller proportion of strong ties. Evacuation managers can expect a higher level of compliance behavior if the community at risk has a greater proportion of strong ties. An assessment parameter can be developed on the basis of the proportions of households with extended family or close friends living in the same community. This finding has also important implications about the role of social media. We observe that the decision to evacuate is strongly influenced not only by whom an agent is “linked” to but also on the trust (weight) that the agent places on the connected nodes.
1. Dash N, Gladwin H. Evacuation decision making and behavioral responses: individual and household. Nat Hazards Rev. 2007;8(3):69–77.
2. Mileti DS, Beck EM. Communication in crisis: explaining evacuation symbolically. Commun Res. 1975;2(1):24–49.
3. Gladwin H, Lazo JK, Morrow BH, Peacock WG, Willoughby HE. Social science research needs for the hurricane forecast and warning system. Nat Hazards Rev. 2007;8(3):87–95.
4. Hasan S, Ukkusuri SV. A threshold model of social contagion process for evacuation decision making. Transp Res Part B. 2011;45(10):1590–1605.
disaster preparedness; hurricane evacuations; social networks
© 2013 Lippincott Williams & Wilkins, Inc.