Full report available at:
http://www.publichealth.hurricane.lsu.edu/ Adobe%20files%20for%20webpage/reports%20%26%20meeting%20minutes /Annual%20Interim%20Report%20(Yr3)%20Narrative%20-%20%20HEF%20(2001-06)-01.pdf [PDF 2.5 MB]

(5). New Orleans Population Survey · Hurricane Evacuation and Sheltering

Dr. Jeanne Hurlbert
LSU Research Professor
Dept. of Sociology
Dr. John J. Beggs
LSU Associate Professor
Dept. of Sociology

Status of survey work

The primary focus of this survey is to understand better how individuals in the New Orleans area would be affected by, and would respond to, a major hurricane. Our previous work has studied how individuals used their social and economic resources—particularly their social networks—to prepare for and respond to Hurricanes Andrew and Georges. That research has shown us what kinds of networks tend to be most useful in dealing with hurricanes, as well as what types of individuals are most likely to have those networks.

In the current survey, we seek to understand how residents of the New Orleans area would prepare for and respond to a major hurricane—one as severe as Andrew—in two ways. First, we are assessing the network structures and resources of these residents so that we can understand better what kinds of resources are available and gain some sense of the extent to which individuals would use formal and informal sources of help. Included in this component is a focus on social and economic resources, broadly construed, as this information helps us to understand the resources of individuals in the metropolitan area as well as how they use those resources in non-disaster situations. For example, knowing whether individuals use their networks in non-disaster situations may help us to predict whether they would use those networks in a disaster scenario. Second, we are asking residents about their past hurricane experience and how they would respond if a storm as bad as Andrew approached the New Orleans area. We include questions on past hurricane experiences; the availability of transportation; and how, under what circumstances, and in what direction they would evacuate if a hurricane approached the New Orleans area in the future. These questions included asking whether they knew someone to whom they could go and whether they would go to a motel/hotel, shelter, someone’s house, etc. if they evacuated.

In conjunction with civil engineers at LSU, we included an expanded section on hurricane evacuation. This section allows us to gain vital information on how the direction and severity of an approaching hurricane would affect the decision to evacuate and the timing of the evacuation. With all of this information, we will be able to understand well the social and economic resources of New Orleans area residents and have detailed information on evacuation response. This will provide critical data for planning for a future storm. Because the data will be geocoded (using respondents’ address information), we will know a great deal about the respondents’ locations and the characteristics of the areas in which they live. All of this will give us a much better picture of who would evacuate, under what circumstances, and in what direction if a hurricane struck this vulnerable city.

We have completed the main portion of this telephone survey. The data are cleaned and we are beginning to analyze and geocode them. We presented preliminary results at the annual advisory meetings for this project and have presented them to individuals from the Centers for Disease Control. These analyses yield interesting and important information about the likelihood and correlates of evacuation.

We are working to complete the second portion of the survey with our “supplemental sample.” Because the main portion of the survey was collected via telephone (with the sample constructed through random-digit dialing), this sample excluded households that did not have telephones. Those households are disproportionately to contain poor, minority residents. These individuals are also much more likely than those living in households that have telephones to reside in vulnerable housing. And, in some parts of the city, the proportion of households that lacks telephones is high—as much as 25%. Using 2000 Census data (Summary File 3), we have constructed maps of the city that identify areas in which the rate of non-telephone households is high. We must now send field teams into those areas with cellular telephones so that they can locate non-telephone households, secure the cooperation of respondents, and the interview can be conducted via cell phone from LSU. This allows us to represent this poorer, more transient, and probably more vulnerable portion of the population.

We have been working with individuals from the American Red Cross in New Orleans, who are assisting so that we can stretch our budget to cover the costs of interviewing in these areas. After identifying the areas with high proportions of non-telephone households, and meeting with the Red Cross personnel and individuals from the New Orleans Police Department (NOPD), we realized that we had underestimated the danger entailed in sending interviewers into the areas in which these households were concentrated. We tried to secure assistance from the NOPD: They had agreed, last fall, to donate 40 hours of police officer time to protect the field interviewers while they are conducting this portion of the survey. Unfortunately, Hurricane Ivan disrupted this process. We must now renegotiate with the NOPD to secure assistance. We hope to enter the field in the early spring, once daylight hours increase.

Preliminary results

Among our measures of likely evacuation behavior was one that asked respondents what they would do if a storm as severe as Hurricane Andrew—a Category 4 storm—threatened New Orleans. Our analyses thus far focus on that measure of probable evacuation behavior. We focus these analyses further on the proportion of individuals who would leave the New Orleans area, and on its correlates. Overall, we find that 68.8% of respondents would leave the area, 9.8% would leave their homes but remain in the area, and 21.4% would remain in their homes. That 21.4% of respondents would remain in their homes is a startling and important statistic, because it indicates that nearly 1 in 4 residents would refuse to leave their homes and 3 in 10 would refuse to leave the area.

These results provide some correlates of possible evacuation if a major storm threatened New Orleans. We described many of our results in our last report. Particularly interesting was the finding that individuals who had lived in the area longer and those who had experienced a major storm are actually less likely to leave than those with shorter tenure in the area and residents who had not “ridden through” a major storm. As we have explored those findings, anecdotal evidence suggests that a “culture” of sorts may exist in New Orleans which encourages residents not to evacuate, even in the face of a major storm.

We have extended these preliminary analyses in two ways. First, we are now exploring the health correlates of evacuation. We find that individuals who are in “good,” “excellent,” and even “fair” health are much more likely to report that they would evacuate than those who are in “poor” health: 73 % of those in “excellent” health, 70% of those in “good” health, 68% of those in “fair” health, but only 43% of those in “poor” health report that they would leave the area. We find similar patterns for mental health and disability. Of those with low levels of depression, 72% would evacuate, whereas 65% of those at middle levels and 56% of those at high levels would do so. Only 53% of those with a disability, compared to 71% of those without would leave the area. Psychological resources also appear to be consequential, with individuals with better coping skills more likely to say that they would leave the area than those with lower levels of mastery. Individuals reporting higher levels of social support are more likely to say that they would evacuate than those with lower levels of support.

The second extension of our earlier analyses was to begin estimating multivariate models. These analyses are important because they begin to show us which factors exert statistically significant effects on evacuation behavior, net of each other and net of controls. Although these results are very preliminary, they show some interesting patterns.

As one would expect, the perceived threat (e.g., the extent to which respondents believe that injury and/or damage would result from a severe storm) exerts the strongest effect of the factors we have examined. We also find, though, that the structure of individuals’ social networks exerts significant effects on the likelihood of evacuation. The higher the perceived likelihood that one’s network members would evacuate, the higher the likelihood that they would evacuate. Having a higher percentage of network members who are perceived able to provide help after a hurricane decreases the likelihood of evacuation.

Our next steps are to extend these analyses and to complete the field interviews. From there, we will begin to analyze the effects of other socioeconomic resources—including other network resources—on the likelihood of evacuation. We will also begin to model neighborhood effects, using multi-level modeling techniques. We have scheduled our first conference presentation from these data at the Southwestern Social Science meetings in March, 2005.