Research Paper

An empirical evaluation of the determinants of tourist's hurricaneevacuation decision making

Ignatius Cahyanto a,n, Lori Pennington-Gray b,1, Brijesh Thapa b,1, Siva Srinivasan c,2,Jorge Villegas d,3, Corene Matyas e,4, Spiro Kiousis f,5

a Tourism & Hospitality Management, School of Business, Black Hills State University, Meier Hall 335, 1200 University Street Unit 9007,Spearfish, SD 57799-9007, United Statesb Department of Tourism, Recreation and Sport Management, University of Florida, 325 FLG, PO Box 118209, Gainesville, FL 32611, United Statesc Department of Civil and Coastal Engineering, University of Florida, 513-a Weil Hall, Gainesville, FL 32611, United Statesd College of Business and Management, University of Illinois at Springfield, One University Plaza, MS UHB 4054, Springfield, IL 62703-5407, United Statese Department of Geography, University of Florida, 3141 Turlington Hall, Gainesville, FL 32611-7315, United Statesf Department of Public Relations, University of Florida, PO Box 118400, Gainesville, FL 32611-8400, United States

a r t i c l e i n f o

Article history:Received 13 February 2013Accepted 22 October 2013Available online 27 November 2013

Keywords:TouristsEvacuationDecision-makingCrisisFloridaProbit modeling

a b s t r a c t

Tourists are vulnerable in the event of a crisis. This article is focused on examining aspects of tourists thatpotentially influence whether or not they evacuate in the event of a hurricane. In general the results ofthis study suggest that individual characteristics (risk belief, connectedness, knowledge, and pastexperience with hurricanes), travel related variables and the socio-demographic characteristics oftourists influence their decision regarding whether or not to evacuate in the event of a hurricane, withtourists who are not local showing higher risk beliefs regarding hurricanes, with low connectedness andknowledge about hurricanes, without past experience with hurricane impacts, traveling with a largerparty, traveling with children, traveling for the first time to the destination, traveling by plane andpersonal vehicle, older age groups, female, with an annual income more than $125,000 are more likely toevacuate. Managerial implications of the findings are discussed.

& 2013 Elsevier Ltd. All rights reserved.

1. Background of the study

Over the last decade, the tourism industry worldwide hasexperienced ample crises ranging from naturally induced crisessuch as wild fires, tornadoes, earthquakes, tsunamis, and humaninduced crises such as political unrest and terrorist attacks. Theaforesaid crises have contributed to a decline in tourists' visitationand have had a negative impact on the economy of the tourismindustry in multiple places (Pennington-Gray, London, Cahyanto, &Klages, 2011; Schaper, 2012). A VISIT FLORIDAs survey found that20% of potential visitors were concerned with returning to Floridaduring 2005s hurricane season, indicating that if all those peoplestayed away, it would result in a $6.7 billion loss in expenditures(Pain, 2006).

Key West, Florida lost approximately $1.5 million a day fromcommerce and tourism when an evacuation order was issuedduring the 2008 hurricane season (Miami Herald, 2009).

Faulkner (2001) and Ritchie (2004) argued that there is ascarcity of research on the crisis phenomenon in the tourismindustry, albeit within the last five years there have been asubstantial number of research articles published on crises affectingthe tourism industry. Publications on tourism crisis managementcan be seen from two angles. The first angle focuses on the supplyside of tourism, while the second angle focuses on the demand sideof tourism. Recent publications, however, focus primarily on thesupply side of the tourism system and can be further categorizedinto two major themes. The first theme focuses on the impacts of acrisis on the tourism industry and studies on how the tourismindustry, such as destination management organizations, respondsto crises (e.g. Carlsen, 2006; Chacko & Marcell, 2007; Chandler,2004; Cheung & Law, 2006; Cooper, 2005). The second themeexamines how tourists might respond effectively to crises. Specifi-cally, research in this area concentrates on crisis managementmodels and attempts to offer the most effective model to save livesand mitigate risks to tourism businesses that can be implementedby destination management organizations (e.g. Evans & Elphick,2005; Faulkner, 2001; Hystad & Keller, 2006; Ritchie, 2004).

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/jdmm

Journal of Destination Marketing & Management

2212-571X/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.jdmm.2013.10.003

n Corresponding author. Tel.: þ1 605 642 6876.E-mail addresses: ignatius.cahyanto@bhsu.edu,

igcahyanto@gmail.com (I. Cahyanto).1 Tel.: þ1 352 294 1657.2 Tel.: þ1 352 392 9537.3 Tel.: þ1 217 206 7927.4 Tel.: þ1 352 392 0494; fax: þ1 352 392 8855.5 Tel.: þ1 352 273 1220; fax: þ1 352 273 1227.

Journal of Destination Marketing & Management 2 (2014) 253–265

Ritchie (2009) argued that research from the demand side, suchas how tourists respond to crises is still lacking and suggested thatfuture researchers examine tourist behavior during crises. Thisfocus on behavior is pivotal to understanding tourist behavior inthe event of a crisis in order to effectively design ways to developassistance to ensure the safety of tourists in the event of a crisis.Furthermore, Phillips and Morrow (2007) contend that tourists arevulnerable in the event of a crisis, with one of the main reasonsbeing that tourists may not speak and/or read the host languageand may lack knowledge of the risks that hurricanes present.Hence, tourists may experience difficulty in receiving, interpreting,and responding properly to risk messages. Tourists are oftenunfamiliar with their surroundings and lack support systems fromtheir home community (Burby & Wagner, 1996; Faulkner, 2001;World Tourism Organization (WTO), 1998). While in unfamiliarenvironments, tourists may not know who their “protectors” areas they do not know from whom to seek assistance. Thus, theimpacts to tourists in risky situations may be greater than to thosein the general resident population. Due to the need to betterunderstand tourists' behaviors in the event of a crisis, this study,unlike previous studies in the area of tourism crisis, will focus ontourists' behaviors in the event of crisis, specifically on theirbehaviors in the event of hurricanes.

Hurricanes are one of the most disruptive natural disasters todestinations, not only because of the costs associated with theimpacts, but also due to the time full recovery takes following ahurricane strike. According to the National Hurricane Center (NHC)(2006), the 2005 Atlantic hurricane season was considered themost active and harmful season in recorded history in the UnitedStates, causing approximately 2300 deaths and over $130 billion indamages. Furthermore, the economic losses associated with hur-ricanes from fishing, agriculture, commerce, and tourism are longlasting, usually taking several years from which to recover (Lindell& Perry, 2004).

An extensive body of research has identified many factors thatinfluence households' responses to hurricanes (e.g. Drabek, 1986,Lindell & Perry, 2004; Riad & Norris, 1998). The research hasexamined gender (Bateman & Edwards, 2002; Dash & Gladwin,2007; Gladwin & Peacock, 1997), wealth (Burton, Kates, & White,1993; Viscusi, 1995; Whitehead, 2003, 2005), past experiencesrelated to hurricanes (Burton & Kates, 1964; Peacock, Brody, &Highfield, 2005; Vitek & Berta, 1982), hurricane knowledge(Daniels & Loggins, 2007; Dow & Cutter, 1998) and race andethnicity (Gabe, Falk, & McCarty, 2005; Niga, Barnshaw, & Torres,2006).

Unfortunately, little attention has been focused on transientpopulations such as tourists (Phillips & Morrow, 2007). Conse-quently, while general findings in hurricane studies have facili-tated emergency managers and policy makers to develop plansthat make realistic assumptions about the general nature ofhuman behaviors related to hurricanes, such studies may not besufficient to provide the information that emergency manage-ment, policy makers, and the destination management organiza-tions need for specific predictions about the behavior of tourists intheir communities.

The few noteworthy contributions which examine touristbehavior during hurricanes were published in the early and mid1990s and largely by one author (Drabek, 1991, 1993, 1994, 1995,1996, 1999, 2000) with a focus on evacuation strategies andpolicies from a supply viewpoint. A recent study conducted byMatyas et al. (2011), Pennington-Gray, Kaplanidou, and Schroeder(2012) and Villegas et al. (2012) were the only studies focused onthe tourist population. These studies examined the interplaybetween tourists' perceived risks and their likelihood to evacuate,and found that those who perceived higher risk are more likely toevacuate in the event of hurricane warning. This study expands on

Matyas et al. (2011) by employing decision theory to examinedeterminants that influence tourists' evacuation decision makingin the event of hurricanes.

Given the paucity of academic research on tourists' evacuationbehavior and the urgency to conduct such studies, this paper isfocused on examining aspects of tourists that potentially influencewhether or not tourists evacuate during a hurricane. Specifically,this study is guided by three interrelated questions:

(1) What are the effects of a tourist's individual characteristics ondecisions regarding whether or not to evacuate?

(2) What are the effects of travel related variables on a tourist'sdecisions regarding whether or not to evacuate?

(3) What are the effects of socio-demography on decisions regard-ing whether or not to evacuate?

2. Literature review

In this section we discuss the guiding theory for this study aswell as three areas that are presumed to influence tourists'evacuation decision making: Individual characteristics, travelrelated variables, and socio-demography.

2.1. Decision theory under uncertainty

As this study intends to examine the decision making oftourists with regard to their evacuation choices, the study employsdecision-making theory as a theoretical lens. The decision theorytypically can be divided into two major parts: A description of theagent to which the theory applies and normative claims abouthow the agent should behave.

Under the typical decision theory, the agent in which thetheory applies satisfies the following conditions: First, the agent'sbelief state at the time can be represented by a probabilityfunction over a space of possibilities that indicates the agent'sconfidence that the possibility is true, with greater values indicat-ing greater confidence. Second, the agent's evaluative state at thetime that can be represented by a function which assigns positivenumbers to elements in the space of possibilities, referred to as autility. Utilities indicate the extent to which the agent values thatpossibility of obtaining the act. Higher numbers indicate higherutilities. Third, the agent's potential acts in a decision situationthat can be represented by a unique set of mutually exclusivepropositions {a1, …, an} where “a” can be considered as theproposition that the agent performs the ith available act. Typicaldecision theory also posits that the agent would only perform apotential act a if the utilities of this act are at least as large as theexpected utilities of any alternatives. This assertion is also knownas utility maximization (Jones, Boushey, & Workman, 2006;Meacham, 2010).

Within decision theory, there are two competing schools ofthought regarding individual choice. The first is the rational choicethat assumes that individuals behave as if they were acting as pureutility maximizers to deduce patterns of outputs from socialsystems (Friedman, 1953). The second is bounded rationality thatacknowledges psychological restrictions on human decision-mak-ing, premised on the assumption that individuals would maximizethe utility of alternatives that are available for them at that time.Decision theory has been widely used in order to examine variousfields from transportation behaviors (Kitazawa & Batty, 2004),insurance purchasers (Kunreuther & Pauly, 2004) as well asevacuation decision makers (Burton et al., 1993; Letson, Sutter, &Lazo, 2007; Viscusi, 1995).

In tourism literature, decision theory has long been used toexamine travel decision-making, purchase behaviors, as well as

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265254

factors that influence decisions (Moutinho, 1987; Sirakaya &Woodside, 2005; Woodside & MacDonald, 1994). Woodside andMacDonald (1994) applied decision theory to understand destina-tion choices. In their study they found that often tourist choices ofdestination are not always rational. The choices are attributed to theinteraction of individual preferences and the influences of travelparty members. Sirakaya and Woodside (2005) provide a compre-hensive development of theories of decision making by tourists.While they argue that decision theory has been frequently used tounderstand travel behavior, there is still a need to advance currentunderstanding of tourist decision-making by applying the theory toother facets of travel experiences.

Decision theory under bounded rationality expands our under-standing of tourist's decision making. During a crisis situation thatis characterized by uncertainty, high threats, and a short decisiontime, individuals often do not have the luxury of time andinformation to assist them in decision making. Thus, theseindividuals are more likely to make a decision based on limitedoptions available. Burton et al. (1993) and Vescusi (1995) arguedthat individuals make choices under the uncertainty of the threatby maximizing their expected utilities. To do so these individualsmight be willing to forgo their wealth including their income,capital, and property in order to minimize those threats. Burtonet al. (1993) elaborates further by contending that under the threatof environmental hazards, an individual hazard response is influ-enced by four major elements: Prior experience with the specifichazard, an individual's wealth, their intrinsic characteristics, andtheir interaction with society.

To date, there is no empirical evidence that the same decisionmaking processes apply to tourists differently than to residents.This is due to tourists lack of ties to the destination and that theyare transient populations compared to residents who have residedin the destination and have greater knowledge of the destination.This study specifically focuses on the variables that are assumed toinfluence tourists' evacuation choices as identified in the evacua-tion literature which is focused on residents.

2.2. Individual characteristics and evacuation

We defined individual characteristics as multiple internalvariables beyond typical socio-demographic variables. Specifically,we focused on four variables: Individual risk belief with hurri-canes, level of individual connectedness to hurricanes, hurricaneknowledge, and past hurricane experience.

Risk belief signifies personal beliefs regarding controllability,optimism bias, as well as risk propensity/aversion (Rohrmann,1995, 1999, 2000). Risk-specific belief pertains to the extent of anindividual's level of confidence to overcome uncertainty (Aldoory,Kim, & Tindall, 2010; Grunig, 1989; Major, 1993, 1998; Lee &Rodriquez, 2008; Quintal, Lee, & Soutar, 2010; Slater, Chipman,Auld, Keefe, & Kendall, 1992; Sriramesh, Moghan, & Wei, 2007).When individuals hold higher perceptions of controllability, theyare less likely to perceive that they are at risk (Burby & Wagner,1996). Similarly, individuals with lower perceptions of the con-trollability of an event are much more likely to perceive that theyare at risk. Consequently, those who are less confident with theirability to overcome risks associated with hurricanes are morelikely to perceive higher risks leading to their evacuation decision(Sorensen & Sorensen, 2007; Whitehead, 2003).

Individual risk beliefs have been found to be predictors offuture travel behaviors (Sonmez & Graefe, 1998a, 1998b) and alsoinfluence the evaluation of destination alternatives and informa-tion acquisition (Roehl & Fesenmaier, 1992). Research suggests thatspecific locations are more risky than other destinations in termsof perceptions of risk by tourists (Sonmez & Graefe, 1998a). Forexample, destinations which have experienced previous natural

disasters (i.e., Miami or Key West) may be perceived as riskier thandestinations which have not been hit by past natural disasters(Sonmez & Graefe, 1998b). In addition, Floyd and Pennington-Gray(2004) found that tourists perceived national parks, natural areasand museums to be less risky locations than theme parks. Coastalversus non-coastal areas may have varying levels of perceived risk,particularly as they relate to hurricanes, therefore, in this studyrisk belief is predicted to positively influence evacuation.

H1. Tourists with higher risk belief regarding hurricanes are morelikely to evacuate than those with lower risk belief.

The level of individual connectedness refers to how personallyconnected an individual feels to an issue (Aldoory et al., 2010;Grunig, 1989; Petty & Cacioppo, 1986; Slater et al., 1992; Srirameshet al., 2007). Typically, the level of individual connectedness can bedetermined by examining three attributes: Interest, importance,and curiosity that individuals have concerning a specific issue(Greenwald & Leavitt, 1984; Hallahan, 2000; Zaichkowsky, 1985).Past studies on individual connectedness to hurricanes and eva-cuation predictions have been inconclusive. However, severalresearchers argue that personal connectedness to hurricanes canelevate perceptions of risk, leading to higher evacuation rates(Burton & Kates, 1964; Vitek & Berta, 1982). Peacock et al. (2005)argued that this occurs because those who are personally con-nected to hurricanes will be more receptive to warning messages,compared to those who are not personally connected to hurri-canes. However, other researchers such as Lindell and Perry (2000)found that those with personal connectedness to hurricanes areresistant to warning messages as they are more likely to perceivethat they are in control of the situation, leading to a lowerevacuation rate.

H2. Tourists with low individual connectedness to hurricanes aremore likely to evacuate than those with higher individual con-nectedness to hurricanes.

Current knowledge has been found to influence the travelrelated decision making of tourists (Gursoy & McCleary, 2004;Ratchford, 2001; Vogt & Fesenmaier, 1998), who can gainprior knowledge from their experiences with the destination fromthe experiences of others, the mass media, and the Internet.Hoogenraad, Eden, and King (2004) assert that independenttourists are more vulnerable to natural hazards as they travelseparate from recognized groups and that they often take morerisks, while Murphy and Bayley (1989) argue that due to thenature of pleasure travel, tourists tend to dismiss risks and displaya low level of natural disaster awareness. Research conducted byJohnston et al. (2002 in Johnston et al., 2007) found that 46%of U.S visitors were unaware of tsunami warning systems com-pared to 28% of locals and only 19% of visitors had seen tsunamihazard maps.

Similarly, a study of backpackers in North Queensland Australiafound that this group had a low awareness of cyclones, with only30% receiving information concerning cyclones during their trip,thus leading to an increase in their vulnerability (Hoogenraadet al., 2004). These studies suggest that current hurricane knowl-edge may influence tourists' decisions regarding evacuation. Thosewho lack knowledge of hurricanes are more likely to experience ahigher risk perception preceding their evacuation choice. Conver-sely, those who possess a sufficient level of hurricane knowledgetend to be better able to decide in an appropriate manner (Griffin,Dunwoody, & Neuwirth, 1999). Therefore, it is predicted that thelevel of hurricane knowledge has a negative correlation withchoosing to voluntarily evacuate during hurricanes.

H3. Tourists with low hurricane knowledge are more likely toevacuate than those with higher hurricane knowledge.

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265 255

In the area of hazards, past studies have used past experience as apredictor of several dependent variables ranging from individualperceptions of risk (Griffin, Dunwoody, & Zabala, 1998), informationseeking regarding the hazard (Johnson & Meischke, 1993; Lenz,1984), and evacuation choices (Baker, 1991; Lindell, Lu, & Prater,2005; Phillips & Morrow, 2007; Whitehead, 2003). Whitehead(2003) contends that the main goal of hurricane evacuations is toalleviate the risk of injury or death and that people who are in floodprone areas have demonstrated a higher likelihood of evacuationthan those who are not in at risk areas. Phillips and Morrow (2007)found that having past experience with hurricanes affect residents'decisions as to whether or not they evacuate. Thus, residents ofhurricane-prone regions who have experienced the impacts ofhurricanes tend to have a greater familiarity and thus bettercomprehension of hurricane-related terminology than tourists whohave not had prior hurricane experience. It is also important to notethat another study by Lindell et al. (2005) found that past experiencewith hurricanes do not significantly influence evacuation choices.While several studies regarding residents' evacuation behavior haveaddressed the relationship between past experience to currentevacuation behavior, little analysis has been done within the contextof tourism. One study by Matyas et al. (2011) on tourists' evacuationbehavior found a significant relationship between tourists' pastexperience with hurricanes and their likelihood to evacuate in theevent of hurricane warnings with those who experienced hurricaneimpacts in the past demonstrating a much lower propensity toevacuate than those who have never experienced a hurricane in thepast. They did not, however, specifically ask respondents about theseverity of hurricanes that they had experienced in the past andwhether or not they were asked to evacuate in the past. Nonetheless,the Matyas et al. (2011) study has helped shed light on the need toexamine the relationship between past experiences and evacuationbehaviors. In this study it is predicted that tourists with pastexperience with hurricanes have a positive association with thedecision to voluntarily evacuate.

H4. Tourists with past experience with hurricane impacts are lesslikely to evacuate than those without such experience.

2.3. Travel related variables

This study focuses on the size of the travel party, the number ofvisits to the destination, the composition of the travel party, andmode of transportation. The tourism literature has found that thetravel party size influences travel related decisions such as whereto eat, where to stay, and what to do (Decrop & Snelders, 2005). Inthe context of evacuation, familial size and having a united familyare relevant factors, with a larger family size having been found topositively correlate with a higher propensity to evacuate(Sorensen, 2000) with one possible reason being the desire toprotect the entire family. Therefore it is predicted that the size ofthe travel party is positively associated with tourists' decisions tochoose to voluntarily evacuate in order to safeguard all membersof the travel party.

H5. Larger travel parties are more likely to evacuate than thosewith smaller travel parties.

In the resident evacuation literature, the length of residencehas also been found to influence evacuation decisions, with longerlengths of stay being correlated with a lower likelihood ofevacuation in the event of a potential hurricane strike (Gladwin& Peacock, 1997). Likewise, the tourism literature has also foundthat those who have never visited the destination before exhibithigher perceptions of risk than those who have been in thedestination before (Sonmez & Graefe, 1998b). One possible reasonis that those who have previously been in the destination have

more familiarity with regard to the destination and may also havedeveloped support systems there. Thus, first time tourists aremore vulnerable in the event of a crisis as they lack the requisitehurricane related knowledge. This lack of familiarity may causehigher perceptions of risks associated with hurricanes and lead toan effort to alleviate the uncertainty by evacuating. Therefore, inthis study it is predicted that those who have never been in thedestination are more likely to evacuate than those who have beenin the destination before.

H6. First time tourists are more likely to evacuate than those whoare not.

Tourism research has documented the roles of travel partycomposition in travel related decision making (Decrop & Snelders,2005; Egelhoff & Sen, 1992; Fesenmaier & Jeng, 2000; Hyde, 2004).The role of children in household may significantly affect decision-making with regard to travel related decisions (Nickerson &Jurowski, 2001). Similarly, the presence of children and the elderlyhave been found to significantly influence evacuation choices.Dash and Gladwin (2007) and Solis, Thomas, and Letson (2010)found that households with children display a higher propensityfor evacuation. One of the reasons is that the social expectation toprotect children from any possible danger increases the likelihoodof evacuation. However, it is also important to note that the studyby Matyas et al. (2011) found that although the presence ofchildren increased perceptions of risk, it did not lead to a higherlikelihood of evacuation by tourists. However, they argued thatthis finding might be caused by the unfamiliarity of the destina-tion and not knowing where to go when an evacuation order isissued. Thus, they called for further examination on the effect ofchildren on evacuation decisions. Several studies such as Soliset al. (2010) and Dash and Gladwin (2007) also found thathouseholds with elderly persons showed a lower probability toevacuate. They argued that health issues that might limit mobilityor over reliance on personal experience or knowledge mightdiscourage the elderly to evacuate. For those who travel withelderly, other travel members may need to accommodate eldersneeds that might decrease the likelihood of evacuation. Based onprior studies, this study predicts that the presence of children inthe travel party will increase the likelihood of evacuation, whilethe presence of elderly will decrease the likelihood of evacuation.

H7. Those who travel with children are more likely to evacuatethan those without children.

H8. Those who travel with elderly are less likely to evacuate thanthose without.

Access to transportation has consistently been determined tobe a factor that influences hurricane evacuation decision makingamong residents, with those who have access to transportationbeing more likely to evacuate than those without access (Soliset al., 2010). With regard to tourists however, Matyas et al. (2011)found that tourists who fly to the destination and those who rentvehicles in the destination exhibit lower rates of evacuation thanthose who do not fly or rent a vehicle in the destination. In thisstudy, those who fly to the destination and those with personalvehicles are predicted to have a positive association with evacua-tion likelihood, while those with rental vehicles are predicted tohave a negative association.

H9. Tourists who fly to the destination are less likely to evacuatethan those who do not fly.

H10. Tourists who rent a vehicle in the destination are less likelyto evacuate than those who do not.

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265256

H11. Tourists who take personal vehicles to the destination aremore likely to evacuate than those who do not.

2.4. Socio-demography and evacuation

Past studies on hurricane evacuation have consistently utilizeddemographics as a factor in predicting evacuation choices rangingfrom ethnicity, age, gender, education, and wealth, with conflictingresults (Baker, 1991; Dow & Cutter, 1998; Lindell et al., 2005; Soliset al., 2010; Whitehead et al., 2000). Dow and Cutter (1998) foundno differences in factors such as age, race, and gender betweenevacuees and non-evacuees, while other researchers (Bateman &Edwards, 2002; Dash & Gladwin, 2007; Gladwin & Peacock, 1997)found the aforesaid factors to be significant in influencing evacua-tion decisions.

With regard to the influence of tourists' socio-demographicvariables to their evacuation choices, Matyas et al. (2011) foundthat international tourists in coastal destinations are more likely toevacuate than domestic tourists. One plausible explanation wasthat international tourists tend to be less familiar with thedestination compared to domestic tourists and coastal destina-tions are considered risky destination with regard to hurricanes.Furthermore, the study also found a significant relationshipbetween a tourist's age and the likelihood of evacuation withyounger tourists being more likely to evacuate than tourists aged50 or higher, although these tourists demonstrated the highestperception of risks, which is consistent with previous studies(Baker, 1979; Drabek, 1986; Eisenman, Cordasco, Asch, Golden, &Glik, 1997).

The same study also found that female tourists have a relativelyhigher risk propensity and are more likely to evacuate than malesin the event of a hurricane warning. In addition, Matyas et al.(2011) also found that tourists traveling in their own vehicleindicate a higher evacuation likelihood and perception of riskthan those who did not travel in their own vehicle. These findingsare parallel with previous studies regarding residents' evacuationdecisions (Dow & Cutter, 2002; Lindell et al., 2005). Nonetheless,as Matyas et al. (2011) were focused on seeking relationshipsbetween tourists' socio-demographic characteristics and theirevacuation choices, further study needs to be conducted toexamine the causality of tourists' socio-demographic variableswith respect to their evacuation choices as relationship does notassume causality (Agresti & Finley, 2009).

The effect of race on the evacuation decisions of residents havebeen examined by several researchers (e.g. Bateman & Edwards,2002; Solis et al., 2010). The studies consistently found that AfricanAmericans are less likely to evacuate than Caucasians. One reasonis that African Americans often demonstrate a lack of trust towardgovernments that might discourage them from evacuating (Soliset al., 2010). Likewise, as evacuation might trigger unexpectedexpenses such as rebooking flights, transportations, and meals,having discretionary income that can be accessed during evacua-tion might increase the likelihood of evacuation. Additionally, weposit that international tourists are more likely to evacuate thandomestic tourists due to unfamiliarity and knowledge regardingthe destination and hurricanes in general.

Based on the literature, the hypotheses for this study werespecified as followed:

H12. Female tourists are more likely to evacuate than maletourists.

H13. Younger tourists are more likely to evacuate than oldertourists.

H14. African American tourists are less likely to evacuate thanCaucasian tourists.

H15. Tourists with higher income are more likely to evacuate thanthose with lower income.

H16. International tourists are more likely to evacuate thantourists from Florida.

3. Methods

3.1. Sample and sampling frame

The population of this study was tourists who were currentlyvisiting Florida. This study utilized an intercepting approach indata collecting to sample tourists. In this study, a tourist wasdefined as “any person who participates in trade or recreationactivities outside the county of his or her permanent residenceor who rents or leases transient living quarters or accommoda-tions” (Florida Statue 125.0104(3)(a)). Screening questions wereutilized to ensure the eligibility of the respondents. To maximizethe randomness of participants, the survey administration usedevery nth formula. Every third eligible tourist in each site wasapproached and asked to complete the survey after verifying theireligibility.

3.2. Study design

This study used a stated preference survey to elicit tourists'evacuation decisions. The use of stated preference surveys allowedelicitation of evacuation decision-making choices based on multi-ple hypothetical hurricane scenarios (McKelvey & Zavoina, 1975;Smith, 1999; Solis et al., 2010). Data collection for this study wasconducted in August-September 2011 during the Atlantic hurri-cane season based on the preposition that during the hurricaneseason, people are more likely to cognitively ponder and seekinformation about hurricanes. Surveys were administrated in twosites in Orlando and Fort Lauderdale Beach, Florida.

Students were hired to administer and intercept surveys ineach site. Prior to the actual survey administration, training wasconducted for student surveyors. A pilot study in the form of afocus group and qualitative interviews was administered tovalidate the findings and the design of the survey instrument aswell as the time needed to complete the survey to minimize anypotential systematic errors (Dillman, Smyth, & Christian, 2009).The survey took approximately 25 min to complete.

3.3. Operationalization

There were sixteen independent variables and one dependentvariable in this study. The independent variables were: Risk-specific beliefs, individual connectedness to hurricanes, currentknowledge, hurricane past experience, size of travel party, priorvisit to destination, presence of children, presence of elderlypersons, the use of an airplane, rental vehicles, personal vehicles,gender, age, ethnicity, income, and place of residency. The evacua-tion decision was the dependent variable. Table 1 summarizes thevariables and their measurements.

To elicit the evacuation choice, eight hypothetical hurricanescenarios were developed, with each containing a combination ofthree attributes, each with two levels. The attributes and the levelswere: Projected hurricane path (passing through the destinationand offset the destination), projected intensity at landfall (category1 and category 4) and time to the destination (48 h and 36 h). Eachrespondent was given 4 scenarios to evaluate, then they wereasked to state their choice of evacuation for a given scenario usinga 5 point Likert scale with 1¼most likely to stay to 5¼most likely toevacuate. Fig. 1 illustrates the scenario used to elicit respondents'likelihood of evacuation.

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265 257

Table 1Variables and measurements.

Variable Measurement Adapted from

Individual characteristicRisk specific beliefs Four statements using a 5-point Likert scale

(strongly disagree to strongly agree) asking respondents'perception on hurricanes.

Rohrmann (2000)

Individual connectedness to hurricanes Three statements measuring the level of individuals' interest,importance, and curiosity with regard to hurricaneswith a 5-point Likert scale.

Major (1998)

Hurricane knowledge Four T/F questions about hurricanes. Matyas et al. (2011)Past hurricane experiences One Y/N question about past experiences with hurricane impacts. Solis et al. (2010)

Travel – relatedSize of travel party One question asking the number of travel party. Decrop and Snelders (2005)First visit to destination One Y/N question asking whether or not a respondent has been

to the destination before.Sonmez and Graefe (1998b)

Presence of children One Y/N question asking the presence of children in the travel party. Nickerson and Jurowski (2001)Presence of elderly One Y/N question asking the presence of elderly in the travel party. Dash and Gladwin (2007)Plane One Y/N question asking whether or not the respondent flew

to the destination.Matyas et al. (2011)

Rental vehicle One Y/N question asking whether or not the respondent rentsa vehicle in the destination.

Matyas et al. (2011)

Personal vehicles One Y/N question asking whether or not the respondent usespersonal vehicles in the destination.

Matyas et al. (2011)

Socio-demographyGender One question asking the respondent's gender. Solis et al. (2010)Age One question asking the respondent's age. Solis et al. (2010)Ethnicity One question asking the respondent's ethnicity. Solis et al. (2010)Income One question asking the respondent's income. Whitehead et al. (2000)Place of residence One question asking the respondent's place of residence. Matyas et al. (2011)

Dependent variableEvacuation decision A 5-point Likert scale on the likelihood to evacuate.

(most likely to stay to most likely to evacuate)Whitehead et al. (2000)

Fig. 1. Example of hypothetical hurricane scenario.

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265258

3.4. Data analysis

Data analysis for this study involved two stages. First, descrip-tive statistics were performed on the variables, which allowed fora closer look at the nature of the data pattern. Next, in the secondstage, multivariate analysis was performed. This study employedan ordered response model (McKelvey & Zavoina, 1975) procedureto relate all individual characteristics and evacuation choices. Asthe responses to evacuation choices were measured using a Likertscale (1–5), an ordered probit model was employed. The orderedprobit model recognizes the inherent ordering in the outcomevariables of interest and allows for calculation of the probability ofeach level of outcome as a function of explanatory factors. In theordered probit model, a positive parameter indicates that thecorresponding factor is associated with a higher likelihood ofevacuation and a negative parameter indicates the opposite effect.The beta values and the odd ratios are the two main ways in whichpeople typically report the main effects of the ordered responsemodel (McKalvey & Zavoina, 1975). However, unlike logisticregression, the ordered probit model does not produce odd ratios.The odd ratio of ordered probit model was obtained by taking theexponent of the beta coefficient. The parameters of the modelwere estimated using the maximum likelihood estimation.

We focused on one model with all potential explanatoryvariables rather than multiple models with subsets of variables,as our goal was to develop a descriptive model and not necessarilya predictive model. Therefore, assessing accuracy in prediction wasnot a focus. The next step of the study would be further datacollection with more samples. Future studies could then use themodel to do predictive assessments with the new data sets. It isbelieved that this provides a better sequence of testing rather thansplitting the data randomly into two samples and conducting crossvalidation with the same dataset.

We used a main effect model to calculate the effect of eachvariable while holding the other variables constant. Using onemodel with all potential explanatory variables reduces the risk ofone variable picking up the effect of another variable due tointernal correlations. For instance, if people with children aregenerally more risk averse, and one of the two is excluded in themodel, then this relationship is not accounted for in the finalmodel. However, if both variables are included in the model andboth variables are significant, it is expected that each variable haspicked up the “true” effects of each variable. Prior to estimatingthe model, the dataset was converted from a person-based dataset to a scenario-based data set to better reflect choices based onhypothetical hurricane scenarios. A statistical analysis was per-formed using the SPSS 18 package.

4. Results

4.1. Profile of the respondents

A total of 632 eligible tourists were approached in all sites. Fivehundred and forty four agreed to participate (response rate¼86%).Out of the total of 544 completed surveys, 533 were deemedusable due to the completeness of the responses, and thereforewere used for this study. Out of 533, females encompassed 56%.Domestic tourists from outside of Florida encompassed 50% of thesample, followed by international tourists (38%) and domestictourists from Florida (12%), with a median travel party of threepeople in the group. The youngest respondent's age was 20 yearsold and the oldest was 88, with a median age of 44. More than halfof the respondents (54%) indicated that they have been in thedestination before. A majority of respondents did not travel withchildren (64%) and without elderly persons (79%) in their travel

party. Caucasians represented the majority of the sample (69%)followed by Hispanic (12%) and African Americans (8%). Forty fourpercent of respondents earned $50,000 to $99,000 annually.A comparison with the actual profiles of Florida's tourists indicatedthat while the sample was compatible, females were overrepre-sented by 4% (Florida Department of Transportation, 2012) andinternational tourists by 22% (Visit Florida@, 2012). These discre-pancies were considered in the managerial implication. Table 2outlines the descriptive information about the respondents.

A reliability test was employed for “risk belief” and “individualconnectedness to hurricanes” items to ensure the consistency ofthe construct measurements. The overall Cronbach alpha for riskbelief was 0.91, while the overall Cronbach alpha for “individualconnectedness to hurricanes” was 0.90, higher than a minimumvalue of 0.70 (Zinbarg, Revelle, Yovel, & Li, 2005) that indicated ahigh consistency among items in each scale. Consequently, a singlevalue was calculated from each scale to represent individualaggregate “risk belief” and “individual connectedness to hurri-canes.” The mean for aggregate risk belief was 3.2 (S.D.¼1.0),while the mean for overall individual connectedness to hurricaneswas 2.4 (S.D.¼1.0) in a 5-point Likert scale to indicate that inaverage respondents had medium risk belief regarding hurricanesand low individual connectedness to hurricanes with hurricanes.With regard to knowledge of hurricanes, an aggregate score wascalculated based on respondents' responses to four hurricanerelated questions. The overall mean score for hurricane knowledgewas 1.5 (S.D.¼1.5) with 4 as a maximum score, which indicatedoverall low hurricane knowledge among participants.

4.2. Results of ordered probit model

As one person responded to four different scenarios, the samplefor the ordered probit model was 2137. The �2 Log likelihood atconvergence was 4246.393 (χ2¼524.080, df¼27, sig. 0.001)

Table 2Percentage of variables.

Variables Percentage

GenderFemale 56Male 44ResidenceInternational 38Domestic outside Florida 50Domestic inside Florida 12

Income$125,000o 17$100,000–$124,999 20$75,000–$99,999 22$50,000–$ 74,999 22$35,000–$ 49,999 9$24,000–$ 34,999 5o$24,000 5

EthnicityCaucasian 69Hispanic 13African 8Other 10

Past hurricane experience No (49), Yes (51)First visit to destination No (54), Yes (46)

TransportationAirplane No (37), Yes (63)Rental vehicle No (74), Yes (26)Personal vehicle No (65), Yes (35)

Travel party compositionChildren No (64), Yes (36)Elderly No (80), Yes (20)

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265 259

indicating a significant improvement from the baseline model. Themodel with all independent variables accounted for 23% of thevariance in the evacuation likelihood. As our intent was to develop adescriptive model, the percentage of variance explained reflectedhow the aggregate performance of the model is likely to be shouldwe undertake a predictive assessment. Table 3 outlines the resultsof the ordered probit model.

4.2.1. Individual characteristicsThe ordered probit model indicated that there were significant

effects of individual risk belief with regard to hurricanes, indivi-dual connectedness to hurricanes, hurricane knowledge, and theirpast hurricane experiences in predicting evacuation decisions.Those with higher risk beliefs with regard to hurricanes weremore likely to evacuate than those with lower risk belief (β¼0.07)that upheld Hypothesis 1. The odd ratio of risk belief was 1.07suggesting that for one unit increase in risk belief, the likelihood ofevacuation would increase by 1.07.

Two individual characteristics, individual connectedness tohurricanes with hurricane topics and hurricane knowledge alsonegatively affected the likelihood of evacuation, β¼�0.09 andβ¼�0.09 respectively. Thus, it upheld Hypothesis 2 and 3. Foreach one unit increase in the individual connectedness to hurri-canes scale the evacuation likelihood declined by.92. Likewise for

one unit increase in the hurricane knowledge scale, the evacuationlikelihood declined by.92.

Past hurricane experience contributed to the decrease ofevacuation likelihood (β¼0.49). The contribution of this variablein predicting likelihood of evacuation was found to be verysignificant, with having past hurricane experience decreasing theevacuation likelihood by .61. As past hurricane experience showednegative association, Hypothesis 4 was supported.

4.2.2. Travel related variablesWith regard to travel related variables, the ordered probit model

indicated that the size of travel party, first visit, presence of children,presence of the elderly, and modes of transportation used all hadsignificant effects on evacuation likelihood. The size of travel partywas found to have a positive effect on evacuation likelihood (β¼0.05)with odd ratio of 1.05, that upheld Hypothesis 5. This indicated thatthe larger the travel party, the higher the likelihood to evacuate.

Those who have never been in the destination before showed ahigher propensity to evacuate than those who visited the destina-tion in the past (β¼0.24). The odd ratios indicated that those whowere first time tourists were 1.27 times more likely to evacuatethan those who were not first time visitors. With this finding,Hypothesis 6 was upheld.

Table 3Result of the ordered probit model.

Variables Parameter estimate Odd ratio Sig.

Individual characteristicH1. Risk belief 0.07 1.07 0.008n

H2. Individual connectedness to hurricanes with hurricane topics �0.09 0.92 0.003n

H3. Hurricane knowledge �0.09 0.92 0.001n

H4. With past hurricane experience (ref.¼without) �0.49 0.61 0.001n

Travel – relatedH5. Number of travel party 0.05 1.05 0.001n

H6. First visit to destination (ref.¼no) 0.24 1.27 0.001n

H7. With children (ref.¼without) 0.16 1.17 0.007n

H8. With elderly (ref. ¼without) �0.2 0.82 0.002n

H9. Plane (ref.¼no) 0.5 1.65 0.001n

H10. Rental vehicle (ref.¼no) �0.15 0.86 0.014n

H11. Personal vehicle (ref.¼no) 0.25 1.28 0.002n

Socio-demographyH12. Female (ref.¼male) 0.24 1.27 0.001n

H13. Age 0.01 1.01 0.001n

H14. Ethnicity (ref.¼Caucasian)African American �0.02 0.97 0.787Hispanic 0.12 1.13 0.119Other 0.07 1.07 0.426

H15. Income (ref¼o24)125o 0.32 1.37 0.019n

100–124.9 0.17 1.18 0.20475–99.9 0.04 1.03 0.76750–74.9 �0.04 0.96 0.74935–49.9 0.13 1.14 0.34624–34.9 �0.07 0.93 0.658

H16. Place of residence (ref¼Florida)International 0.36 1.43 0.001n

Domestic outside Florida 0.24 1.28 0.003n

ThresholdsEvacuate¼1 �0.54 – 0.007Evacuate¼2 �0.02 – 0.898Evacuate¼3 0.75 – 0.001Evacuate¼4 1.46 – 0.001

�2 Log likelihood at convergence(n¼2137) 4246.393 (χ2¼524.080, df¼27, sig. 0.001)Pseudo R2 Negelkerke 0.23

Ref¼reference group.n po0.05.

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265260

A composition of travel party yielded interesting results. Thosewith children were 1.17 times more likely to evacuate than thosewithout children (β¼0.16) which found support for Hypothesis 7.The presence of elderly individuals in the travel group affectedevacuation likelihood with those with elderly being.82 times lesslikely to evacuate than those without the presence of elderly intheir travel group (β¼�0.20). Therefore Hypothesis 8 was upheld.

Mode of transportation was also found to significantly affectevacuation likelihood with airplane and personal vehicle usagespecified to have positive effects β¼0.50 and β¼0.25 while the useof rental vehicles was specified to have a negative effect β¼�0.15.The odd ratio indicated that those who flew to Florida were 1.65times more likely to evacuate than those who did not fly to Florida.Those who utilized personal vehicles were 1.28 times more likelyto evacuate than those who did not use personal vehicles. Inter-estingly, those who rented vehicles were 0.86 times less likely toevacuate than those who did not rent vehicles. Thus Hypothesis 9was not supported, while Hypothesis 10 and 11 were supported.

4.2.3. Socio-demographyThe results of the ordered probit model indicated that females

were 1.27 times more likely to evacuate than males (β¼0.24),which provided support for Hypothesis 12. However, the modelindicated that the older the tourist was the more likely this touristwas to evacuate (β¼0.01). The odd ratio indicated that for one unitincrease in age, the evacuation likelihood increased by 1.01. Thisfinding contradicted the findings in Hypothesis 13.

The result of the model did not find support for Hypothesis 14.In this study, race and ethnicity were not found to significantlyaffect the likelihood of evacuation. With regard to income, onlythose who earned more than $125,000 in annual householdincome were found to vary significantly such that they were 1.37times more likely to evacuate than those who earned less than$24,000 annually. Therefore, Hypothesis 15 was upheld.

Likewise, the residence of origin of tourists was found topositively affect evacuation likelihood. International tourists were1.43 times more likely to evacuate than those who resided inFlorida (β¼0.36), while those who were from outside Florida were1.28 times more likely to evacuate than those who were fromFlorida (β¼1.28). This finding provided support to upholdHypothesis 16.

5. Discussion

The use of decision theory under uncertainty and boundedrationality allowed for a detailed examination of the determinantsof tourists' evacuation. By applying the theory to the estimatedmodel, it can be argued that the decision by the tourist regardingwhether or not to evacuate is a dynamic complex phenomenon.The ordered probit model yields interesting results. Most hypoth-eses were upheld. Three hypotheses were not upheld: Hypothesis9, 13, and 14. Those were the use of plane, effects of age, and raceand ethnicity on evacuation likelihood. Variables associated withindividual characteristics (i.e. risk belief, individual connectednessto hurricanes with hurricane, hurricane knowledge, and pasthurricane experience) all have significant effects, with risk beliefwith hurricanes having a positive association with evacuationlikelihood, while other variables including past experience, indi-vidual connectedness to hurricanes, and hurricane knowledge allhaving negative associations.

The positive association of individual risk belief to evacuation isexplicable as risk belief reflects personal belief on controllabilityand optimism bias (Rohrmann, 1995, 1999). Risk belief also reflectsan individual's level of confidence to overcome uncertainty (Quintalet al., 2010). Therefore, it is understandable that individuals with

higher risk belief will feel more vulnerable and perceive lesscontrollability when exposed to hurricane risks. As these indivi-duals feel more threatened with hurricane risks, they would thenprefer to leave the destination when hurricane warnings are issuedto alleviate uncertainty (Sorensen & Sorensen, 2007). Likewise,higher individual connectedness with a certain topic has beenconsistently found to increase one's self confidence (Aldoory et al.,2010; Grunig, 1989), and therefore makes one less susceptible toexternal threats. As predicted, tourists with low individual con-nectedness to hurricanes showed a higher likelihood to evacuatethan those with greater individual connectedness to hurricanes.

Additionally, the negative association of hurricane knowledgeand evacuation likelihood indicated that those with higher hurri-cane knowledge are less likely to evacuate than those with lowhurricane knowledge. One possible explanation is that level ofknowledge of hurricane and its risk help the person in theirdecision making with regard to evacuation. Sufficient knowledgeabout hurricanes will allow the individual to make a betterdecision, instead of allowing him or herself to be susceptible toexternal threats (Hoogenraad et al., 2004). Individuals are oftenbased on their own knowledge to make a decision in a riskysituation (Griffin et al., 1999). The samples for this study indicatedthat most of them have low knowledge with regard to hurricanesthat might be resulted due to the nature of the tourist population.The effect of knowledge was consistent with what was found inpast research, with leisure travelers demonstrating a low level ofnatural disaster awareness (Murphy & Bayley, 1989; Johnston et al.,2007). This study only used four questions that were designed tomeasure the level of hurricane knowledge that may affect thequality of responses. A better set of questions may be needed tofully capture the effect of hurricane knowledge on evacuation.

The result of the order probit model on risk belief, individualconnectedness to hurricanes, and hurricane knowledge suggeststhe need to increase tourists' self-confidence. Increasing individualconfidence can be done in various ways. One possible way is toincrease one's knowledge level of hurricanes. As most tourists inthis study indicated that they have low individual connectednessto hurricanes and minimal hurricane knowledge, it is suggestedthat activities aimed at increasing hurricane awareness amongtourists is needed. This may include creating hurricane informa-tion in multiple formats that are easy to access.

In this study past hurricane experience shows that thosewithout past experience with hurricane impacts are more likelyto evacuate than those that experienced hurricane impacts in thepast. This finding was consistent with the Matyas et al. (2011)study. This is an interesting finding as most studies on residentsand their evacuation behaviors indicate that past hurricaneexperience has no influence (Lindell et al., 2005) or positiveassociation (Riad, Norris, & Ruback, 1999; Solis et al., 2010). Onepossible explanation is that people often used past experiences asa reference to make a decision in the same situation (Johnson &Tversky, 1983). For those without past experiences regarding theimpacts of a natural disaster, past studies have found that theseindividuals would create a reference based on social cues on whichto base their decisions to alleviate uncertainty (Major, 1998) andthat those without past experiences have consistently indicatedhigher perceived vulnerability as they are uncertain about what todo in the event of natural disasters (Griffin et al., 1998). Therefore,it is understandable that tourists without past experiences withhurricane impacts are more likely to evacuate than those withhurricane experience as they felt that they were more vulnerablein the event of hurricanes.

Results of an ordered probit model also indicate the roles oftravel related variables in predicting the likelihood of evacuation.As predicted, the larger the travel party the more likely tourists areto leave the destination in the event of a hurricane. One plausible

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265 261

explanation is that with more people in the group, one will notonly consider his or her own safety but also other individuals inthe group. With regard to the result of the presence of children,the findings support the pre-specified hypothesis. As predicted,the presence of children was found to increase the likelihood ofevacuation. Moreover, it was interesting to note that those whotravel with elderly persons are more likely to remain in thedestination, which is parallel with prior studies regarding resi-dence evacuation (Gladwin, Gladwin, & Peacock, 2001). Onepossible explanation is seniors are at greater risk for healthproblems which could be exacerbated during evacuation. Localdestination management organizations need to establish a systemto respond to tourists who may have disabilities or are not asmobile. Contrary to widely held expectations, those who traveledby air and those who used their own vehicle were more likely toevacuate than those who rented a vehicle. One possible explana-tion is that those who come to Florida by plane feel morevulnerable than those who do not take an airplane. These aremore likely to be international tourists and tourists from outside ofFlorida.

The findings with regard to owning a personal vehicle areparallel with Matyas et al. (2011), such that those with personalvehicles are more likely to evacuate than those without a personalvehicle. In addition, the findings also confirm prior studies ofresident evacuation behaviors that argue that access to vehicles isan important factor in dictation evacuation likelihood, with thosewith access demonstrating a higher likelihood of evacuation thanthose without access (Eisenmann et al., 1997). One possibleexplanation is that concern over potential traffic associated withevacuation may motivate tourists who have access to personalvehicles to leave the destination as soon as possible in order toavoid traffic. As perception over potential traffic may play a role toboth motivate tourists to leave the destination or to stay in thedestination, it is essential to inform tourists of potential trafficinformation, main evacuation routes as well as secondary evacua-tion routes. One notable issue with hurricane evacuation is“shadow evacuation” (Wolshon, Urbina, Wilmot, & Levitan,2005), where those who have not been ordered to evacuate decideto evacuate, contributing to major traffic in the interstate system.While the state of Florida has done well in posting evacuationroute signs along evacuation routes, tourists who are not familiarwith the routes and procedures may still need more informationregarding evacuation routes.

The findings related to socio-demography were also note-worthy. With regard to gender, the findings of this study wereconsistent with several past studies that found females to demon-strate a higher propensity to evacuate than males (Lindell et al.,2005; Riad et al., 1999). This suggests that males and females viewevacuation differently. In addition, we also believe that thedifference is also due to socially constructed gender roles in otherfactors that influence the intention and capacity to evacuate. Thismay include but not limited to care-giving roles and perception ofsubjective risk. Due to the limitation of our data, ensuing researchshould examine more thoroughly gender differences in riskexposure and perceptions as well as their corresponding influ-ences on the decision to evacuate.

The significance of the age of the tourist in predicting thelikelihood of evacuation is parallel with Matyas et al. (2011). Onepossible explanation is that younger tourists have greater access to“evacuation incentives” (Perry, Lindell, & Greene, 1981). Evacua-tion incentives are factors that would seem to increase theprobability that a threatened individual will comply with evacua-tion warnings, for instance greater mobility and greater access totransportation. Nonetheless, further research needs to exploremore thoroughly the effect of age on evacuation decision. Thefindings indicated that race and ethnicity were not significant

factors of the likelihood to evacuate, which is consistent with thefindings by Dow and Cutter (1998) and Baker (1991). More recentstudies however have found that race and ethnicity does affectevacuation rates with Caucasian being more likely to evacuatethan African Americans (Elder et al., 2007; Pennington-Gray et al.,2012). This discrepancy may be due to the nature of our sample;however further research is needed to confirm this finding. As atransient population, tourists have different connections with thedestination compared to residents who reside in the destinationand that other variables better explain the intention to evacuatethan ethnicity.

With regard to income, this finding partially supports thefindings of previous studies on residents' evacuation behaviors(Solis et al., 2010) with higher incomes being associated with agreater likelihood to evacuate. This study also found that onlytourists with incomes of more than $125,000 are more likely toevacuate than those who earned less than $24,000. One possibleexplanation is that changing travel plans often requires a sub-stantial amount of money for related expenses such as changingflights and accommodation plans which could exceed theirallotted budget. Therefore, it is understandable that those withsubstantially higher incomes are more likely to leave.

With regard to residence of origin, this study found thatinternational tourists demonstrated a higher propensity to evac-uate than those from other states and from Florida. This isconsistent with the findings of Matyas et al. (2011). One possibleexplanation is that the proximity of residence has an effect on thefamiliarity of hurricanes. International tourists may not be familiarwith hurricanes and perceive the threat differently than those whoreside in Florida. Therefore, international tourists are more likelyto leave the destination to alleviate uncertainty. Since mostinternational tourists flew to Florida, it is understandable thatthe findings also reflect the finding on the use of airplanes withthose who flew to Florida showing a higher likelihood of evacua-tion than those who do not travel by air. This suggests the need toprovide educational information regarding hurricanes and evacua-tion orders. Such information should be made available in multiplelanguages to accommodate international tourists. Internationaltourists are very vulnerable in the event of a hurricane evacuation,as they may not know what to do and how to seek informationregarding the evacuation due to language and cultural barriers.Therefore, local Destination Management Organizations need towork with accommodations, policy makers and local emergencymanagement organizations to ensure their awareness and safety.

6. Managerial implications

With regards to managerial implications, the results obtainedin the estimated model may be a useful tool to identify thewillingness to evacuate by broad tourist groups. This informationmay help destination management organizations and emergencymanagement organizations in Florida to target resources moreefficiently, focusing not only on groups of tourists with higherrisks but also on those groups with a lower probability to evacuate.

Due to the background of tourists, who are not from thedestination, the design of messages could be improved. Forexample, from this study, age was found to be a significantpredictor of evacuation decision, thus, choosing age-appropriatewarning characteristics by tailoring the physical characteristics ofmessages to compensate for age-related changes in perception isrecommended.

For visual messages targeted at the elderly, typographicalcharacteristics of text can be enhanced, such as utilizing san seriffonts such as Helvetica which has been found to increase textlegibility for elderly (Hartley, 1999). Likewise, for auditory

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265262

messages, they need to be broadcast at frequencies that are notaffected by noise-induced loss so that elderly can comprehend themessage better. In addition, sometimes fast moving text-crawlerson the television screen may be subject to glare and use small textsize and as such, international tourists who may not speak Englishmay find it difficult to comprehend the content of the message.Slower movement and the use of larger fonts may be needed to aidinternational tourists to perceive the messages.

Another approach to improving tourists' text comprehensionfor written warnings is to increase the readability by making thetext message simple, direct and easy to understand. To do so,Mileti and Sorensen (1995) suggests the use of specific pieces ofinformation anchors: Hazard, location, time and guidance. There-fore, messages directed at tourists need to have a simplifiedsentence structures with non-technical jargon. The content ofthe message needs to be written to lessen the need for inferentialprocessing.

When new information is received, individuals typically inter-pret it in the context of their preexisting knowledge. As found inthis study, when individuals are presented with hurricane advi-sory messages, they will likely tap into their past experience ortheir knowledge with hurricanes and other hazards and reactbased on the outcomes of those actions. Therefore, warningmessages should be presented in a fashion that is consistent withwhat they already know. By doing so, the credibility of thewarning massage will also be amended. To assess what touristsalready know about hurricane risk and evacuation procedures,efforts should be made to regularly review the effectiveness oftourists hurricane awareness programs. Because much of tourists'knowledge about hurricanes, especially repeat tourists will bedetermined by exposure to general information materials such aspamphlets that are encountered during a hurricane season, it iscritically important to determine what information is most com-prehensible and memorable.

To assess tourists' hurricane knowledge, emergency plannerscan collaborate with local destination management organizationsto conduct surveys, structured interviews, or focus groups. Oneadvantage of such programs is a better understanding of howtourists' conceptualize hurricane risks. If tourists have misconcep-tions regarding the complexity of refund policies associated withhotels or flights, the hurricane awareness and warnings could beredesigned to focus and eradicate these misunderstandings.

Additionally, it is recommended that tourism organizations atthe local level (e.g. hotels, CVBs) in Florida allocate funds andresources to respond to the threat or impact of a hurricane. Thesefunds require policies which address who can access the fund,when the fund will be accessed (under what conditions), whomakes the decision on when and how to use the funds, etc. Inaddition, resources need to be used specifically to meet tourists'needs. This may include but is not limited to training employeesand staff to be ready through drills, mock evacuations, as well asdesigning a shelter on the property or creating a brochure, whichshows the location of the nearest shelters for the guests. By havingreserved funds and resources, the tourism organizations will beable to provide a quicker response to meet the need of tourists. Itis very challenging to address tourist needs in the event of a crisiswithout sufficient funds and resources.

Next, tourism organizations also need to establish collaborationand mutual aid agreements with the emergency agencies, otherdestination management organizations, and other key partners.Communicating hurricane risks to tourists is often overwhelming,especially for small organizations that may not have adequateresources. Collaboration and mutual agreements with other orga-nizations allow organizations to gain access to resources that maynot be available in their inventory. Tourism organizations alsoneed to communicate regularly with the emergency agencies,

especially at the county level to get current information duringhurricane season. Assistance from local emergency agency may beneeded if the situation worsens, and guests and tourists need to beevacuated. Collaboration with other organizations is crucial toallow resource sharing and speedy communication flow, whichwill be needed in the event of hurricanes.

Next, the tourism industry in Florida needs to develop guide-lines to accommodate tourists. This may include a refund policy,local shelter information, and assistance, and other such measures.In addition, the tourism industry needs to provide mechanisms toinform hurricane risks to guests and tourists. These mechanismscan be in the form of hurricane brochures that can be placed inseveral major locations such as airports, rest areas on the FloridaTurnpike and Florida Welcome Centers when evacuations areissued. The accommodation industry can also provide hurricanebrochures in the hotel lobby or in the guest's room whenhurricanes are in the horizon. Adequate information about hurri-cane risks will help guests and tourists make well-informeddecisions and help alleviate their perception of risk. In addition,the guidelines will help to ensure tourists' needs are met and theyare kept safe as well as send a signal to tourists that the hosts arecaring and concerned for their well-being.

7. Conclusion

The study analyzes the determinants of tourist hurricaneevacuation decision making based on the decision theory ofbounded rationality. This study contributes to the literature byaccounting for two issues normally neglected in previous evacua-tion studies. First, it focused on transient populations of touristswho are in the destination when a hurricane evacuation issued.Secondly, this study utilizes several variables that were not used inprevious resident evacuation studies such as risk belief andindividual connectedness to hurricanes with hurricanes.

This study also contributes to the practical implications byhighlighting several areas that can be improved to strengthen thehurricane evacuation strategy for tourists in Florida. As theprevious section indicates, there are a myriad of ways the touristhurricane evacuation procedures in Florida can be improved, forinstance, by improving the content of the messages, establishingpartnerships between local CVBs with local emergency agencies,allocating budgets, and disseminating information. We believethat the findings can be generalized to other storm-prone destina-tions. Nevertheless, given that the study was conducted in thestate of Florida that is considered to have a better hurricaneinformation system compared to other locations, further researchis needed to test the predictive assessment of the model bycollecting larger samples and its application across differenttourism destinations.

As tourists may already be in the destination for several daysbefore they learn that a hurricane is coming, further study isneeded to measure the effect of the amount of time betweenlearning that a hurricane is coming and when the evacuation orderis issued on their evacuation decision. Those who have been in thedestination longer may be more likely to evacuate than those whohear that a hurricane is coming and an evacuation order is issuedon the first day of their stay after they have experienced most ofthe destination. However, there is also a possibility that thesegroups are less likely to evacuate as the longer they stay the morefamiliar they are with the destination and its support systems.Thus, further study is needed to fully understand the effect of timewhen a tourist becomes aware of the hurricanes threat on theirevacuation decision. Finally, while this study does not specificallyaddress the effect of tourists' social interaction in the process ofevacuation decision, the significance of information gathering in

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265 263

the process of deciding to evacuate while facing a hurricane threatis certainly complex and should be the subject of future studies.

References

Agresti, A., & Finley, B. (2009). Statistical methods for the social sciences. New Jersey,NJ: Pearson Prentice Hall.

Aldoory, L., Kim, J., & Tindall, N. (2010). The influence of perceived shared risk incrisis communication: Elaborating the situational theory of publics. PublicRelation Review, 36(2), 134–140.

Baker, E. J. (1979). Predicting response to hurricane warnings: A reanalysis of datafrom four studies. Mass Emergencies, 4, 9–24.

Baker, E. J. (1991). Hurricane evacuation behavior. International Journal of MassEmergencies and Disasters, 9, 287–310.

Bateman, J., & Edwards, B. (2002). Gender and evacuation: A closer look at whywomen are more likely to evacuate for hurricanes. Natural Hazards Review, 3,107–117.

Burby, F. J., & Wagner, F. (1996). Protecting tourists from death and injury in coastalstorms. Disasters, 20(1), 49–60.

Burton, I., & Kates, R. (1964). The perception of natural hazards in resourcemanagement. Natural Resource Journal, 3, 412–441.

Burton, I., Kates, R., & White, G. (1993). The environment as hazard (2nd edition).New York, NY: The Guidord Press.

Carlsen, J. (2006). Post-Tsunami tourism strategies for the Maldives. Tourism ReviewInternational, 10(1/2), 69–80.

Chacko, H. E., & Marcell, M. H. (2007). Repositioning a tourism destination: The caseof New Orleans after hurricane Katrina. Journal of Travel & Tourism Marketing,23(2/3/4), 223–235.

Chandler, J. A. (2004). An analysis of the economic impact of hurricane Dennis,Floyd, and Irene on North Carolina's lodging industry. Journal of Hospitality &Tourism Research, 28(3), 313–326.

Cheung, C., & Law, R. (2006). How can hotel guests be protected during theoccurrence of a tsunami? Asia Pacific Journal of Tourism Research, 11(3),289–295.

Cooper, M. (2005). Japanese tourism and the SARS epidemic of 2003. Journal ofTravel & Tourism Marketing, 19(2/3), 117–131.

Daniels, R. J., & Loggins, G. M. (2007). Conceptualizing continuous coverage: Astrategic model for wall-to-wall local television weather broadcasts. Journal ofApplied Communication Research, 35, 48–66.

Dash, N., & Gladwin, H. (2007). Evacuation decision making and behavioralresponses: Individual and household. Natural Hazards Review, 8(3), 69–77.

Decrop, A., & Snelders, D. (2005). A grounded typology of vacation decision-making. Tourism Management, 26, 121–132.

Dillman, D., Smyth, J., & Christian, L. M. (2009). Internet, mail and mixed-modesurveys: The tailored design method. Hobeken, NJ: Wiley.

Dow, K., & Cutter, S. L. (1998). Crying wolf: Repeat responses to hurricaneevacuation orders. Coastal Management, 26, 237–252.

Dow, K., & Cutter, S. L. (2002). Emerging hurricane evacuation issues: HurricaneFloyd and South Carolina. Natural Hazards Review, 3(6), 12–18.

Drabek, T. E. (1986). Human system responses to disaster. New York, NY: Springer.Drabek, T. E. (1991). Anticipating organizational evacuations: Disaster planning by

managers of tourist-oriented private firms. International Journal of MassEmergencies and Disasters, 9(2), 219–245.

Drabek, T. E. (1993). Variations in disaster evacuation behavior: Public responsesversus private sector executive decision-making processes. Disasters, 16(2),104–118.

Drabek, T. E. (1994). Disaster evacuation and the tourist industry: Environment andBehavior Monograph No. 57. Boulder, CO: Natural Hazards Research andApplication Information Center.

Drabek, T. E. (1995). Disaster responses within the tourist industry. InternationalJournal of Mass Emergencies and Disasters, 13, 7–23.

Drabek, T. E. (1996). Disaster evacuation behavior: Tourist and other transients.Environment and Behavior Monograph No. 58. Boulder, CO: Natural HazardsResearch and Application Information Center.

Drabek, T. E. (1999). Disaster-induced employee evacuation. Boulder, CO: Institute ofBehavioral Science, University of Colorado.

Drabek, T. E. (2000). Disaster evacuation behavior: Tourist-business managers rarelyact as customers expect. Cornell Hotel and Restaurant Administration Quarterly,August, pp. 48–57.

Egelhoff, W. G., & Sen, F. (1992). An information-processing model of crisismanagement. Management Communication Quarterly, 5, 443–484.

Eisenman, D. P., Cordasco, K. M., Asch, S., Golden, J. F., & Glik, D. (1997). Disasterplanning and risk communication with vulnerable communities: Lessons fromHurricane Katrina. American Journal of Public Health, 97, 109–115.

Elder, K., Xirasagar, S., Miller, N., Bowen, S. A., Glover, S., & Piper, C. (2007). AfricanAmericans' decisions not to evacuate New Orleans before Hurricane Katrina: Aqualitative study. American Journal of Public Health, 97(Suppl. 1), 124–129.

Evans, N., & Elphick, S. (2005). Models of crisis management: An evaluation of theirvalue for strategic planning in the international travel industry. InternationalJournal of Tourism Research, 7, 135–150.

Faulkner, B. (2001). Towards a framework for tourism disaster management.Tourism Management, 22(2), 135–147.

Fesenmaier, D., & Jeng, J. (2000). Assessing structure in the pleasure trip planningprocess. Tourism Analysis, 5(1), 13–27.

Florida Department of Transportation (2012). Travel demand: visitors and tourists.Office of Policy Planning of the Florida Department of Transportation (Trendsand Conditions Report-2012).

Floyd, M., & Pennington-Gray, L. (2004). Profiling risk perceptions of tourists.Annals of Tourism Research, 31(4), 1051–1054.

Friedman, M. (1953). Essays in positive economics. Chicago, IL: University of ChicagoPress.

Gabe, T., Falk, G., & McCarty, M. (2005). Hurricane Katrina: Social demographiccharacteristics of impacted areas. Washington, DC: Congressional ResearchService.

Gladwin, C. H., & Peacock, W. G. (1997). Warning and evacuation: A night for hardhouses. In: W. G. Peacock, B. H. Morrow, & C. H. Gladwin. (Eds.), HurricaneAndrew: Ethnicity, Gender, and Sociology of Disaster (pp. 52–74). London:Routledge.

Gladwin, C. H., Gladwin, H., & Peacock, W. G. (2001). Modeling hurricane evacua-tion decision with ethnographic methods. International Journal of Mass Emer-gency and Disasters, 19, 117–143.

Greenwald, A. G., & Leavitt, C. (1984). Audience involvement in advertising: Fourlevels. Journal of Consumer Research, 11.

Griffin, R. J., Dunwoody, S., & Neuwirth, K. (1999). Proposed model of the relation-ship of risk information seeking and processing to the development ofpreventive behaviors. Environmental Research Section A, 80, S230–S245.

Griffin, R. J., Dunwoody, S., & Zabala, F. (1998). Public reliance on risk communica-tion channels in the wake of a cryptosporidium outbreak. Risk Analysis, 18,367–375.

Grunig, J. E. (1989). Sierra club study shows who become activists. Public RelationReview, 15(8), 3–24.

Gursoy, D., & McCleary, K. W. (2004). An integrative model of tourist informationsearch behavior. Annals of Tourism Research, 31(2), 353–373.

Hallahan, K. (2000). Inactive publics: The forgotten publics in public relations.Public Relation Reviews, 26(4), 499–515.

Hartley, J. (1999). What does it say? Text design, medical information, and olderreaders. In: D. C. Park, R. W. Morrell, & K. Shifren (Eds.), Processing of MedicalInformation in Aging Patients: Cognitive and Human Factors Perspectives (pp.233–248). Mahwah, NJ: Lawrence Erlbaum Associates.

Hoogenraad, W., Eden, R. V., & King, D. (2004). Cyclone awareness amongstbackpackers in Northern Australia. The Australian Journal of Emergency Manage-ment, 19(2), 25–29.

Hyde, K. (2004). A duality in vacation decision-making. Tourism Analysis, 8(2–4),183–186.

Hystad, P., & Keller, P. (2006). Disaster management: Kelowna tourism industry'spreparedness, impact and response to a 2003 major forest fire. Journal ofHospitality and Tourism Management, 13(1), 44–58.

Johnson, E. J., & Tversky, A. (1983). Affect, generalization, and the perception risk.Journal of Personality and Social Psychology, 45, 20–31.

Johnson, J. D., & Meischke, H. (1993). A comprehensive model of cancer-relatedinformation seeking applied to magazines. Human Communication Research, 19,343–367.

Johnston, D., Becker, J., Gregg, C., Houghton, B., Paton, D., Leonard, G., et al. (2007).Developing warning and disaster response capacity in the tourism sector incoastal Washington, USA. Disaster Prevention and Management, 16(2), 210–216.

Jones, B. D., Boushey, G., & Workman, S. (2006). Behavioral rationality and thepolicy processes: Toward a new model of organizational information proces-sing. In: G. Peters, & J. Pierre (Eds.), Handbook of Public Policy. Thousand Oaks,CA: SAGE Publications.

Kitazawa, K., & Batty, M. (2004). Pedestrian behavior modeling – An application toretail movements using genetic algorithms. In J.P. van Leeuwen & H.J.P.Timmermans (Eds.), Developments in Design & Decision Support Systems inArchitecture and Urban Planning (pp. 130–150). Eindhoven, Netherlands: Eind-hoven University of Technology.

Kunreuther, H., & Pauly, M. (2004). Neglecting disaster: Why don't people insureagainst large losses? Journal of Risk and Uncertainty, 28(1), 5–21.

Lee, S., & Rodriquez, L. (2008). Four publics of anti-bioterrorism informationcampaigns: A test of the situational theory. Public Relation Review, 34, 60–62.

Lenz, E. R. (1984). Information seeking: A component of client decisions and healthbehavior. Advanced Nursing Science, 6, 59–72.

Letson, D., Sutter, D., & Lazo, J. (2007). The economic value of hurricane forecasts:An overview and research needs. Natural Hazards Review, 8, 78–86.

Lindell, M. K., & Perry, R. W. (2000). Household adjustment to earthquake hazard.Environmental Behavior, 32, 590–630.

Lindell, M. K., & Perry, R. W. (2004). Communicating environmental risk in multi-ethnic communities. Thousand Oaks, CA: Sage Publications.

Lindell, M. K., Lu, J., & Prater, C. S. (2005). Household decision making andevacuation in response to hurricane Lili. Natural Hazards Review, 6(4), 171–179.

Major, A. M. (1993). Environmental concern and situational communication theory:Implications for communicating with environmental publics. Journal of PublicRelations Research, 5(4), 251–268.

Major, A. M. (1998). The utility of situational theory of publics for assessing publicresponse to a disaster prediction. Public Relations Review, 24(4), 489–509.

Matyas, C., Srinivasan, S., Cahyanto, I., Thapa, B., Pennington-Gray, L., & Villegas, J.(2011). Risk perception and evacuation decisions of Florida tourists underhurricane threats: A stated preference analysis. Natural Hazards, 59, 871–890.

McKelvey, R. D., & Zavoina, W. (1975). A statistical model for the analysis of ordinallevel dependent variables. Journal of Mathematical Sociology, 4, 103–120.

Meacham, J. G. (2010). Building and its consequences. Philosophical Studies, 149(1),49–71.

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265264

Miami Herald (2009). Keys May Ease hurricane evacuation policy. Retrieved on June3, 2011 from: ⟨http://www.miamiherald.com⟩.

Mileti, D. S., & Sorensen, J. H. (1995). Warning systems: A social science perspective.In: D. Golding, J. X. Kasperson, & R. E. Kasperson (Eds.), Preparing for nuclearpower accidents (pp. 336–375). Boulder, CO: Westview Press.

Moutinho, L. (1987). Consumer behavior in tourism. European Journal of Marketing,21(10), 3–44.

Murphy, P. E., & Bayley, R. (1989). Tourism and disaster planning. GeographicalReview, 79(1), 36–46.

National Hurricane Center (NHC). (2006). Tropical cyclone report: 2005 AtlanticHurricane Season. Washington DC: NOAA.

Nickerson, N. P., & Jurowski, C. (2001). The influence of children on vacation travelpatterns. Journal of Vacation Marketing, 7(1), 19–30.

Niga, J. M., Barnshaw, J., & Torres, M. R. (2006). Hurricane Katrina and the floods ofNew Orleans emergent issues in sheltering and temporary housing. The Annalsof the American Academy of Political and Social Science, 604, 115–128.

Pain, J. (2006). Florida tourism flying high despite '04 hurricanes. Associated Press.Retrieved on February 15, 2012 from: ⟨http://www.usatoday.com/travel/news/2005-06-14-floridatourism_x.htm⟩.

Peacock, W., Brody, S., & Highfield, W. (2005). Hurricane risk perceptions amongFlorida's single family homeowners. Landscape and Urban Planning, 73(2–3),120–135.

Pennington-Gray, L., London, B., Cahyanto, I., & Klages, W. (2011). Expanding thetourism crisis management planning framework to include social media:Lessons from the deepwater horizon oil spill, 2010. International Journal ofTourism Anthropology, 1(3/4), 239–253.

Pennington-Gray, L., Kaplanidou, K., & Schroeder, A. (2012). Drivers of social mediause among African Americans in the event of a crisis. Natural Hazards, 66(1),77–96.

Perry, R. W., Lindell, M. K., & Greene, M. R. (1981). Evacuation planning in emergencymanagement. Lexinton, MS: Lexington Press.

Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central andperipheral routes to attitude change. New York, NY: Springer-Verlag.

Phillips, B., & Morrow, B. (2007). Social science research needs: Focus on vulnerablepopulations, forecasting, and warnings. Natural Hazards Review, 8(3), 61–68.

Quintal, V. A., Lee, J. A., & Soutar, G. N. (2010). Tourists’ information search: Thedifferential impact of risk and uncertainty avoidance. International Journal ofTourism Research, 12(4), 321–333.

Ratchford, B. (2001). The economic of consumer knowledge. Journal of ConsumerResearch, 27, 397–411.

Riad, J. K., & Norris, F. H. (1998). Hurricane threat and evacuation intentions: Ananalysis of risk perception, preparedness, social influence and resources. Newark,DE: University of Delaware Disaster Research Center.

Riad, J. K., Norris, F. H., & Ruback, R. B. (1999). Predicting evacuation in two majordisasters: Risk perception, social influence, and access to resources. Journal ofApplied Social Psychology, 29, 918–934.

Ritchie, B. (2004). Chaos, crises, and disasters: A strategic approach to crisismanagement in the tourism industry. Tourism Management, 25(6), 669–683.

Ritchie, B. (2009). Crisis and disaster management for tourism. Bristol, UK Channel:View Publications.

Roehl, W. S., & Fesenmaier, D. R. (1992). Risk perceptions and pleasure travel: Anexploratory analysis. Journal of Travel Research, 30(4), 17–26.

Rohrmann, B. (1995). Effective risk communication for fire preparedness: Aconceptual framework. The Australian Journal of Emergency Management, 10,43–48.

Rohrmann, B. (1999). The appraisal of information material on disaster prepared-ness. In EMA (Ed.), Proceedings of the Australian Disaster Conference (pp. 251–256).

Rohrmann, B. (2000). A socio-psychological model for analyzing risk communica-tion processes. Australasian Journal of Disaster Studies, 2000–2, 1–20. ⟨http://www.massey.ac.nz/�trauma/issues/2000-2/rohrmann.htm⟩.

Schaper, D. (2012). Post-Sandy: Atlantic City wants its tourists back. Retrievedon July, 26, 2013 from: ⟨http://www.npr.org/2012/12/06/166627881/post-sandy-atlantic-city-wants-its-tourists-back⟩.

Sirakaya, E., & Woodside, A. G. (2005). Building and testing theories of decisionmaking by travellers. Tourism Management, 26, 815–832.

Slater, M. D., Chipman, H., Auld, G., Keefe, T., & Kendall, P. (1992). Informationprocessing and situational theory: A cognitive response analysis. Journal ofPublic Relations Research, 4(4), 189–203.

Smith, K. T. (1999). Estimating the cost of hurricane evacuation: A study of evacuationbehavior and risk interpretation using combined reveal and stated preferenceshousehold data. Greenville, NC: Department of Economics, East CarolinaUniversity.

Solis, D., Thomas, M., & Letson, D. (2010). An empirical evaluation of thedeterminants of household hurricane evacuation choice. Journal of Developmentand Agricultural Economics, 2(3), 188–196.

Sonmez, S. F., & Graefe, A. R. (1998a). Determining future travel behavior from pasttravel experience and perceptions of risk and safety. Journal of Travel Research,37, 171–177.

Sonmez, S. F., & Graefe, A. R. (1998b). Influence of terrorism on foreign tourismdecisions. Annals of Tourism Research, 25(1), 112–144.

Sorensen, J. H., & Sorensen, B. V. (2007). Community process: Warning andevacuation. In: H. Rodriquez, E. L. Quarentelli, & R. R. Dynes (Eds.), Handbookof Disaster Research (pp. 182–199). New York, NY: Springer.

Sorenson, J. H. (2000). Hazard warning systems: Review of 20 years of progress.Natural Hazards Review, 1(2), 119–125.

Sriramesh, K., Moghan, S., & Wei, D. L. (2007). The situational theory of publics in adifferent cultural consumer publics in Singapore. Journal of Public RelationsResearch, 19(4), 307–332.

Villegas, J., Matyas, C., Srinivasan, S., Cahyanto, I., Thapa, B., & Pennington-Gray, L.(2012). Cognitive and affective responses of Florida tourists after exposure tohurricane warning messages. Natural Hazards, 66(1), 97–116, http://dx.doi.org/10.1007/S11069-012-0119-3.

Viscusi, W. (1995). Fatal tradeoffs: Public and private responsibilities for risk. Oxford,NY: Oxford University Press.

Visit Florida@. (2012). Florida visitors statistics. Retrieved July 30, 2013 from: ⟨http://www.visitflorida.com/en-us/media/research.html⟩.

Vitek, J. D., & Berta, S. M. (1982). Improving perception of and response to naturalhazards: The need for local education. Journal of Geography, 81(6), 225–228.

Vogt, C. A., & Fesenmaier, D. R. (1998). Expanding the functional information searchmodel. Annals of Tourism Research, 25(3), 551–578.

Whitehead, J. C. (2003). One million dollars a mile? The opportunity costs ofhurricane evacuation. Ocean and Coastal Management, 46, 1069–1089.

Whitehead, J. C., Edwards, B., Willigen, M. V., Maiolo, J. R., Wilson, K., & Smith, K. T.(2000). Heading for higher ground: Factors affecting real and hypotheticalhurricane evacuation behavior. Environmental Hazards, 2, 133–142.

Whitehead, J. C. (2005). Environmental Risk and averting behavior: Predictivevalidity of revealed and stated preference data. Environmental and ResourceEconomics, 32, 301–316.

Wolshon, B., Urbina, E., Wilmot, C., & Levitan, M. (2005). Review of policies andpractices for hurricane evacuation. I: Transportation planning, preparedness,and response. Natural Hazards Review, 6(3), 129–142.

Woodside, A. G., & MacDonald, R. (1994). General system framework of customerchoice processes of tourism services. In: R. V. Gasser, & K. Weirmair (Eds.),Spoilt for Choice: Decision-Making Processes and Preference Change of Tourists:Intertemporal and intercountry perspectives (pp. 30–59). Thaur, Germany:Kulturverlag.

World Tourism Organization (WTO) (1998). Handbook on natural disaster reductionin tourist areas. Madrid: World Tourism Organization.

Zaichkowsky, J. L. (1985). Measuring the involvement construct. Journal of Con-sumer Research, 12, 341–352.

Zinbarg, R., Revelle, W., Yovel, I., & Li, W. (2005). Cronbach's, Revelle's, andMcDonald's: Their relations with each other and two alternative conceptuali-zations of reliability. Psychometrika, 70, 123–133.

I. Cahyanto et al. / Journal of Destination Marketing & Management 2 (2014) 253–265 265