Edited Version July 1, 1998
EIIP Virtual Library Online Presentation

"Crying Wolf: Repeat Response to Hurricane Evacuation Warnings"

Special Presentation

Dr. Kirstin Dow
Associate Professor, Department of Geography
University of South Carolina at Columbia.

The original transcript of the July 1, 1998 online Virtual Library presentation is available on the EIIP Virtual Forum (http://www.emforum.org). The following version of the transcript has been edited for easier reading and comprehension. Typos were corrected, date/time/names attributed by the software to each were deleted but content of discussion, questions, and responses are as stated by each participant. Answers from the presenter to questions by the audience are grouped beneath the appropriate question to facilitate meaning. Related questions and discussion that occurred in the Virtual Forum session immediately following the 1-hour formal presentation are included in the edited transcript.

Amy Sebring: On behalf of the EIIP, I am pleased to welcome you to a special event in our Library. Please hold all questions and comments until we get to the Q&A portion of the program about half past the hour. We will review the instructions at that time. Before I introduce our special guest, I would like to review how to use links to display Web pages in another browser window for the benefit of our newcomers. When a full URL is typed in the message area, it becomes a hot link, so you can just click on it, and a web page will display in another browser window.

[Slide 1]


And now, it is my pleasure to introduce Dr. Kirstin Dow, Associate Professor, Dept. of Geography, University of South Carolina at Columbia. Dr. Dow is coauthor, with Dr. Susan Cutter, of the paper "Repeat Response to Hurricane Evacuation Orders," and will be sharing their research findings. Thank you so much for being with us today, Kirstin.


Kirstin Dow: Hello. I'm glad to be hear and looking forward to hearing your comments. The research I am going to discuss today, "Crying Wolf: Repeat Response to Hurricane Evacuation Warnings" was supported by the Natural Hazards Information and Applications Center, Quick Response Grant Program. The project report is available at their home page. The paper, this talk is based on a piece of that research, is forthcoming in Coastal Management.

In 1996, two hurricanes, Bertha and Fran both prompted evacuations of the South Carolina coast. Storm tracks showed both storms approaching South Carolina, but fortunately for the state, they both made landfall in North Carolina. Heavy winds were responsible for most of the damage in South Carolina. Slide 2, please.

[Slide 2]

Kirstin Dow: Based on the forecasts, the Governor of SC ordered almost identical evacuations for both hurricanes --- covering all of coastal South Carolina. (Areas east of the intercoastal waterway, all barrier islands, all beachfront properties, all low-lying areas, and all property bordering waterways in several counties.)

While the evacuations were the prudent action, the impact of false alarms on future evacuations, often referred to as the "crying wolf syndrome" (Breznitz 1984) is a widespread source of speculation and concern in the emergency management community. Hurricane planning issues have become more complicated by influx of new residents, large seasonal tourist populations, and the general rapid development of coastal areas.

Repeated false alarms reduce the credibility of warning information, yet very little research has directly studied them and their impact on evacuation behavior.

Using South Carolina as the study area we examined the impact of "crying wolf" through three specific research questions:

1) Were there differences in evacuation responses of residents for hurricanes Bertha and Fran?

2) Was there an action or a specific piece of information that convinced people to evacuate and did this vary between the two hurricane events?

3) What was the major source of "reliable" information influencing the decisions to evacuate and did this differ in the two hurricane events?

Responses to these questions also raised more general issues about the role of local disaster culture and the perceived relationships among residents, government officials at various levels, and information sources such as the Weather Channel.

We started from looking at previous findings on hurricane evacuation. These findings suggest that the definition of the risk area and the actions of public officials are the most important variables affecting public response. Over 90% of residents of high-risk barrier islands and open coasts will evacuate in response to strong, clear warnings from officials.

General knowledge of hurricanes and hurricane safety are poor predictors of hurricane evacuation behavior. Variation in evacuation is accounted for by the hazardousness of the area, action by public authorities, type of housing, prior perception of personal risk, and storm-specific threat factors (Baker 1991).

We conducted a survey of coastal residents. We were in the field less than 2 weeks after hurricane Fran; we went to Hilton Head, Myrtle Beach, and Wilmington. I'm going to talk about South Carolina cases today. Both cities are one the coast; Hilton Head is in the south and Myrtle Beach is in the north.

We conducted surveys with the help of graduate assistants. We concentrated on residents and completed interviews with 128 people in Hilton Head and 143 in Myrtle Beach; our over all response rate was about 75%. However, we did use a screening question to target residents who were in the area for at least one of the hurricanes. A total of 323 interviews were completed.

The surveys were:

• face to face interviews;

• held at entry of major grocery/discount stores;

• comprised of open-ended and closed questions; and,

• about 5 -10 minutes in length.

Findings: Evacuation Rates:

Sorenson and Mileti (1988) reviewed a variety of evacuation events (including hurricanes) and found that evacuation rates vary from 32% to 98% depending on the level of warning and perception of personal risk. For hurricanes, evacuation rates were around 90-95% in geographically-defined high risk areas compared to 25-35% in lower-risk areas.

We did not see those high evacuation rates, we found 42% evacuated for Bertha and 60% evacuated for Fran, the stronger storm. Evacuation rates for both events were higher in Hilton Head, the community furthest from landfall. This maybe partially explained by and a single evacuation route off the island. Responses also show consistency in evacuation decisions, 76% either evacuated for both storms or stayed for both. Slide 3, please.

[Slide 3]

Kirstin Dow: Prior experience did not differentiate among evacuees and non-evacuees

as 76% reported that they had dealt with a hurricane threat before.

Impact of Crying Wolf

To understand the evacuation decisions we asked three sets of questions.

1) What convinced you to evacuate? (Respondents who evacuated were asked to rank the top three factors that convinced them to leave.)

2) Respondents who stayed were asked why they did so.

3) Respondents who evacuated for one hurricane and not the other were asked to explain the difference.

First Question: What convinced residents to evacuate?

Our analysis of the decision-making criteria revealed no dramatic shifts in factors convincing people between Bertha and Fran. Unlike previous evacuation studies, no single factor dominated responses, rather, the residents provided multiple reasons for their decision-making.

The ranking of factors suggests less emphasis on advice and warnings from authorities and more reliance on media and assorted household (included in the "other" category).

Among those who evacuated, a basic crosstab table of the first (most) important factor in convincing them to leave for Bertha versus that for Fran identified the following tendencies: the Governor's order and media reports were almost equal in their influence on the decision to evacuate. The Weather Channel was mentioned specifically.

The large category of "other" factors included a number of individual or household considerations such as safety of children and household pets, vacation plans, work obligations, and potential flooding. Slide 4, please.

[Slide 4]

Second Question: Among people who did not evacuate.

Reasons given for non-evacuation centered on perceived probability and severity of a "hit" followed by the perception of safety of the housing unit and/or location relative to the danger. Note, there are differences between Hilton Head and Myrtle Beach.

Whether they stayed or evacuated, our data suggest that people are considering a wide variety of factors in their decision-making. More so than in previous hurricane evacuation studies, individual evaluation of risks (largely derived from media reports from the Weather Channel) seems to have played a larger role in evacuation decision-making during these two hurricane events.

The role of media coverage of storm characteristics and how individuals utilize this information in their personal risk calculations in evacuation decisions is a needed area for research. Slide 5, please.

[Slide 5]

Third Question: Reliability of Information Sources.

In addition to asking what convinced people, we asked them about the reliability of information sources for each hurricane. If "crying wolf" was a major issue, we would have expected to see a drop in reliability.

The perceived most reliable source of information was the media. Direct advice from local officials is absent from these reports although residents following the news would have heard their warnings. The percentages of people reporting all types of media as the most reliable increased between the two storms.

We were surprised by the specific and frequent mentioning of The Weather Channel as one of the most reliable sources. Despite its importance in convincing people to evacuate, the Governor's order were not evaluated as the most reliable of source of information about evacuation.

The broader response of listening to the media, coupled with specific references to the Weather Channel, again indicated people considered a variety of types of information. Slide 6, please.

[Slide 6]

Kirstin Dow: These responses highlight a distinction between the "reliability" of information sources and actions and information that "convinced" people to evacuate. The most reliable sources (media) provided data on hurricane characteristics such as magnitude and direction of landfall. Advice or orders from officials sometimes convinced people to evacuate, but the advisories were not reported as particularly reliable.

These findings suggest that access to basic data and experts' evaluations through sources such as the Weather Channel and the Internet, provided a necessary confirmation for a more independent and personally relevant evaluation of hurricane risks by local residents.

We also tried to look at the impact of these "unnecessary" evacuations by asking a hypothetical question, "If another hurricane approached the coast, would you evacuate? And why?".

While hypothetical questions can pose difficulties, in the context of this discussion, asking residents about future actions was based on their recent, past decision-making in the face of potential danger.

The majority of our respondents (48%) answered "it depends" to the question while 21% responded "no." The remaining third said "yes" they would evacuate if another hurricane approached.

For those who replied yes, concerns over personal and family safety were paramount. Statements such as "I'm afraid," "better safe than sorry," or "I'd be crazy not to evacuate" illustrate this view. Other respondents gave specific reasons including past experience with hurricanes, and living in a mobile home or near water. Experience with "false alarms" did not seem to sway perception of risk.

Among those respondents who replied "no," only a few mentioned they would not evacuate because the last few forecasted hurricanes did not hit their area. Comments included, "Government leaders cried wolf too often" and "the last few didn't hit" as reasons for on-evacuation.

The largest group, however, was residents who believed their home was a safe place or at least safer than anywhere else. Others cited job requirements as a reason for staying. There were a few respondents who mentioned concern about being able to return home as a reason for not leaving. The majority of these future non-evacuees (81%) also did not evacuate for either Bertha or Fran.

Among the group who said "it depends" they explain that their evacuation decisions would depend on 1)The severity of the storm; 2) Storm path; 3)Probability of a direct hit; 4)Probability of flooding; 5) Less Than 2% would depend on the Governor's Order or orders from local officials.


Based on this research, we have concluded that:

1) The influence of "crying wolf" played only a minor role in evacuation decision- making. South Carolina residents anticipate little change in their evacuation behavior, although there is some semblance of an evacuation-resistant population forming with almost 1/3 of our residents saying that they would not evacuate under any circumstances.

2) "Premature evacuations" did not result in dramatic changes in reasons for evacuation or sources respondents found reliable;

3) The role of official advisories and orders was limited as people sought a wider variety of information;

4) Reliance on the media, the Weather Channel in particular, was more pronounced than in previous studies;

5) Personal assessment of the storm characteristics and its risk played a larger than anticipated role in evacuation decisions as people considered quality of home construction, family needs and safety, and data on storm tracks, strength, and probabilities in their decision-making.

Amy, I will be happy to take questions now.


Amy Sebring: While we are waiting, does South Carolina have a mandatory evacuation law, Kirstin?

Kirstin Dow: Yes, it does, the Governor can order AND compel evacuation in South Carolina.


Kevin Farrell: Of the people you interviewed, did you ask about whether they sustained significant property damage?

Kirstin Dow: There was mostly wind damage in South Carolina. We did not ask people specifically.


Cindy Rice: Any idea on how evacuations are perceived in the nursing homes/hospitals, since you couldn't poll them?

Kirstin Dow: We did a lot of interviews. While we didn't do polls of the nursing homes, we did speak with many people who worked in them. They felt like the evacuations were handled well. People generally felt that the evacuations were handled well, they just sometimes didn't feel that the advice suited them.


Terry Storer: Did the people feel that the Weather Channel was more valid in their expertise than the Governor's office?

Kirstin Dow: The Weather Channel was important in people's decision-making. It appeared as the single most "reliable" information source. The Governor's office was at about 6-8 % in comparison to 15-18%, when we asked people to rank the reliable sources.


Shari Coffin: The International Association of Emergency Managers (IAEM) is commenting on an FCC proposal to prohibit cable overrides for local EAS alerts. Does this have implications on whether that's a good idea?

Kirstin Dow: This study has a lot of implications for the way we think about communication. People are discussing two paths in response. One is to aim for a consistent message and try to get people to focus on one message. The other is to provide more information so that people are making well-informed decisions that take into account the real risks.


Tom Wahle: How would you assess businesses? Have they been a help or barrier to effective evacuation?

Kirstin Dow: The business people I interviewed were involved in facilitating evacuation and others were expected to stay at work. About 7% of the population did not evacuate because of work expectations. Some of these were emergency management related, but not all of them were. Some managers, for instance, reported a commitment to the store. I can't tell you about tourism issues, unfortunately.


Tom Wahle: It seems that business is willing to play along and let people go, right?

Kirstin Dow: In general, that is what we heard from people. I did many of those interviews and did not hear any complaints about that but about other things, but not about being required to stay.


Amy Sebring: Kirstin, were expectations of traffic congestion ever mentioned as a factor?

Kirstin Dow: Traffic congestion was a factor particularly in Hilton Head. There is only one route off of that island and confusion during the Bertha evacuation lead to a large traffic jam. Some people decided to wait to leave and eventually it became clear that the landfall would be much further north, so they did not leave at all.


Isabel McCurdy: Any feedback on what advice should be given for evacuation compliance?

Kirstin Dow: For evacuation compliance, I am most concerned about what is starting to look like the formation of an evacuation-resistant group. About 30 % were not planning to leave under any circumstances, I would focus some attention on those people.

Final Question:

Terry Storer: Would it be fair to say that evacuation decisions remain a personal decision rather than a reaction to public mandate?

Kirstin Dow: As far as decisions, yes. I think it is more of a personal decision than past research has reported. We were also interested in the differences in the response rates between communities. Hilton Head, the southern one, had higher evacuation rates. This led us to wonder about local evacuation response cultures. How much difference can we expect among communities and why?

Amy Sebring: Thank you very much Kirstin for sharing your work with us today. Transcript will be posted with background material in a few days.

Kirstin Dow: Thank you.

Amy Sebring: Before we adjourn to the Virtual Forum room for some follow up, I would like Avagene to say a brief word about tomorrow evening.

Avagene Moore: Tomorrow night is our first evening discussion called "Mutual Aid". This session will be an open dialogue to encourage emergency professionals to bring any issues or problems they wish to discuss; a time and place where experience and expertise can be shared; and a means of networking with others who have mutual interests and concerns. Please tune in tomorrow night July 2 at 8 PM EDT. On Tuesday, we have a discussion in the Round Table on XII. Lois Clark McCoy will lead the discussion.

Amy Sebring: Also next week in Virtual Classroom we will feature FEMA's EENET.

Thank you audience, and since our time is up, we will close down the Library for today, but we will be in the Virtual Forum room for a few minutes longer, and you are welcome to join us there for open discussion. Thank you for your cooperation.

Break: Further Discussions and Questions in the Virtual Room.

After additional expressions to Dr. Kirstin Dow, a few participants stayed for a few minutes with additional comments and discussion related to the library presentation; the following are excerpts that convey additional information about "Crying Wolf: Repeat Response to Hurricane Evacuation Warnings".

Kirstin Dow: Thanks, this kind of interaction is fun.

Amy Sebring: Shhhh...the fun part is supposed to be a secret.

Amy Sebring: Kirstin, what did you want to ask about Internet?

Kirstin Dow: Several of the people we talked with said that they had logged on the Internet. The local emergency responders later told us that they were called for help with interpretation of technical pages. Has anyone else had that kind of experience?

Amy Sebring: Not yet!

David Crews: Some states and local governments have used the Internet for response and recovery. Ft. Collins and North Dakota are two examples.

Amy Sebring: Presumably they are going to National Hurricane Center, that is, straight to the source. I find it very plausible that if individuals have greater direct access to information on which to make their decision, they will rely less on others.

Kirstin Dow: I believe that they are going straight to the source. Another indication of bypassing more local experts.

Amy Sebring: This direct access certainly has grown over the past decade.

David Crews: Florida is working on a system. Also FEMA is moving more into response and recovery via an "intranet" know as NEMIS.

Kirstin Dow: I'm concerned that their information is partial.

Amy Sebring: I tell people that we (local officials) do not have any secret info!

Probability statements can also be easily misinterpreted.

Kirstin Dow: Judgments about the safety of their house also concerned me.

David Crews: The NWS has made vast improvements on their Internet usage to the public via IWIN.

Amy Sebring: Also, following on David's comments, ERLink offers Hurricane Center mirror restricted to EM use.

Amy Sebring: Of the evacuation resistant group, do they perceive themselves as being relatively safe where they are?

Kirstin Dow: Mostly, yes. They thought they were fine.

Kirstin Dow: When you say, there have been improvements. What have they done?

David Crews: Immediate Warnings, Zone Weather and many other features.

Kirstin Dow: I think that those will be getting more use in the future.

David Crews: I use the NWS IWIN all the time in FEMA DFOs.

Amy Sebring: Did anyone else have a follow up for Kirstin or should we let her go with our appreciation?

Kirstin Dow: I did schedule a 1:15 meeting. I've enjoyed "chatting" with you.

Amy Sebring: Again, thank you for the research and the presentation today, Kirstin. Go to your meeting!


What are the Implications ?

A first step is to re-examine the role of hazard information in the decision-making process. We found that evacuation decisions were based on multiple sources of risk information. Additional research is needed on the role of increased media coverage, improved technological access, and greater breadth of information in public decision-making.

A second avenue for continued research is based on the false alarms. While precautionary evacuations are necessary, the timing and longer-term implications require detailed study in other locales. These cautious evacuation orders do not seem to influence the decision-making of residents as to whether to evacuate or not. However, we see significant reduction in the credibility of governmental officials, so much so, that official advisories are not perceived as reliable or personally relevant.

One practical implication is a change in how resident's obtain risk information and more importantly, their view of emergency management officials.

In the past, emergency management officials had a paternalistic relationship with the residents--it was their job to provide hazard information and help protect the community from harm. Perhaps due to more aggressive educational programs, this paternalistic relationship is weakening and people are building on their understandings of hurricane threats to collect and evaluate relevant information.

Emergency managers seem to have little effect on the evacuation decision of local residents according to our survey. Instead, they are now being viewed as traffic control officers, caretakers for tourists, and major obstacles to re-entry after the storm they ultimately influence a decision to stay and ride out the storm. It is clear that the electronic media information is often inconsistent with official warnings from the Governor.

One alternative derived from our results is to have coordinated advisories so that there is "one voice" and thus one set of information that is being transmitted to the public. The placement of official hurricane evacuation advisories and orders on The Weather Channel (for example) may help alleviate some of the confusion and improve the credibility of state-elected and emergency response officials.

A second alternative is to provide more detailed explanations of the rationale behind evacuation orders and their implementation so that those considerations are more fully incorporated into decision-making processes.

A final area for future research involves the notion of a local disaster culture.

Despite more aggressive coastal and hurricane education programs, we still see a significant evacuation-resistant population varying among locations. A more detailed study of these reluctant evacuees might help determine the existence of a local disaster culture that either enhances the movement of people out of harm's way or thwarts attempts to do so.

The differences between our study sites suggests that geographic site and situation may also be important in the formation of local disaster culture and the perceived relationships among residents, authorities at different levels of government, and information sources such as the Weather Channel. Clearly, more localized studies of evacuation decision-making and behavior will help determine either the uniqueness of the South Carolina experience or its generalization to all coastal communities and residents.