Carnegie Mellon University
Carnegie Mellon University
Sample size: 537
Field period: 3/13/2008-03/19/2008
The study examines the effectiveness of textual information dissemination to the general population. When it comes to natural disasters, people need to be aware of their risks and hold correct assumptions of the probability of personal loss or damage. This study explores whether images accompanying text can stimulate greater perceptions of risk. Additionally, this study explores how the formulation of risk perceptions is closely attached to notions of self. If participants show an increased sense of risk surrounding events that are familiar to them, this shows a realistic and common sense approach to determining risk but also highlights potential gaps in risk mitigation and prevention.
Hypothesis 1. Imagery accompanying descriptive text of natural disasters will result in higher perceptions of risk.
Hypothesis 2. The most relevant natural disasters of a participant based on their location of residence will result in higher perceptions of risk.
Hypothesis 3. Imagery accompanying descriptive text of natural disasters will result in higher perceptions of risk even when the natural disaster is not relevant to a participant based on their location of residence.
The study calls for a between subjects factor (text versus text plus image) and a within subjects factor (location relevance). The first independent variable, information presentation is manipulated via two conditions. One set of participants will be given only textual descriptions of natural disasters (Condition 1) while the second set will be given textual descriptions accompanied by images of damage created by the associated natural disaster (Condition 2). Participants will be randomly assigned to one of these two conditions.
The second independent variable, location relevance, will be manipulated via natural disaster type. Each group of participants will be given the descriptions of the same two natural disasters: hurricanes, and wildfires. Each disaster type is more common in various regions of the country. Those disaster types typical of the participant’s region will serve as the most relevant disaster type. The presentation order of the two natural disaster descriptions was randomized for each participant. The sample of participants came from primarily the Southeast for hurricane relevant locations and the West for wildfire relevant locations. We obtained zip codes for each participant and coded each zip code for wildfire and hurricane prevalence.
Participants will be asked four questions with respect to each natural disaster. The first two questions ask participants to judge the safety of their own life and the likelihood personal property will be damaged by each natural disaster. These two questions represent the participant’s risk perceptions with respect to each natural disaster. The next two questions ask participants to judge the likelihood of purchasing insurance against each natural disaster and to assign each natural disaster a priority level for federal natural disaster relief funding. These questions reflect prospective decisions. An online or television survey will be administered to obtain this data.
We had several control variables as well. First we asked whether the participant or someone close to them had been previously in a natural disaster. Second, we asked whether participants had insurance (renter's or home).
We found support for our first hypothesis and second hypothesis, however these effects were not statistically strong. We did not find support for our third hypothesis.
Having previously been in a natural disaster had a large impact on individual responses as well. Individuals felt more likely to be affected by a natural disaster and more scared by the narratives.