Disaster Myths, Emergency Response and Social Media: A Case Study of Hurricane Harvey in Houston, Texas

Master thesis ‘Crisis and Security Management’ Class: February 2017

Name: Aukje Postma Student number: S1218328 Date: 09.01.2018

Word count: 24817 First reader: S.L. Kuipers Second reader: W.G. Broekema

Disaster Myths,

Emergency

Response and

Social Media: A

Case Study of

Hurricane Harvey

in Houston, Texas

1

Table of content

2.1.1.1 Citizens panic in a disaster ... 7

2.1.1.2 Panic and social media ... 7

2.1.2 Helplessness and dependency ... 9

2.1.2.2 Social media and dependability ... 9

2.1.3 Looting ... 10

2.1.3.2 Social media and looting ... 11

2.2 Anticipation by governmental authorities on citizen response ... 12

2.2.1. Social media as unconventional communication tool ... 13

2.2.2. Social media to collect information for victims ... 15

2.2.3.Cooperation between victim and emergency responder... 16

3. Research design and methodology ... 19

3.1 Research Design ... 19

3.2 Operationalization into concepts, definitions and indicators ... 21

3.1.1 Independent variables ... 21

3.1.2 Dependent variables ... 21

3.3 Sampling ... 22

3.4. Data collection ... 23

3.4.1 Scope of data collection ... 23

3.4.2 Data collection citizens of Houston ... 23

3.4.3 Data collection local governmental authorities of Houston ... 24

3.5 Data analysis ... 25

3.6 Limitations ... 26

4. Case time line ... 28

5. Analysis ... 30

5.1 Citizen behavior ... 30

5.1.1 Panic ... 30

5.1.2 Helplessness and Dependency ... 40

5.1.3 Looting ... 49

5.2 Emergency response governmental authorities ... 54

2

5.2.2 Social media to collect information for victims ... 62

5.2.3 Cooperation between victim and emergency responder... 68

6. Conclusion ... 74

7. Bibliography ... 78

Appendix 1. Social media reports citizens ... 84

3

1. Introduction

In August 2017 hurricane Harvey developed and became the first hurricane in twelve years to hit landfall in the United States. As a result major destruction was caused, which resulted in hurricane Harvey becoming the costliest hurricane the United States ever encountered. Another perspective about hurricane Harvey appeared after a conversation with a former Houston resident (Arjen van Dijk).1 _{From the perspective of Van Dijk, governmental authorities of}

Houston were more prepared in providing emergency response during hurricane Harvey compared to previous hurricanes hitting the region (e.g. Katrina in 2005).

Moreover Van Dijk emphasized the ‘Texan mentality’ of the citizens of Houston played an important role in rescue missions provided by fellow citizens. Van Dijk argued this ‘Texan mentality’ in Houston is all about citizens helping each other in providing food, shelter and rescue when needed during such a natural disaster. The ‘Texan-mentality’ to help fellow citizens in need of help, comes from the virtues of the traditional neighborhood mentality and fundamental American values of Texas which origins from the 19th _{century. Nowadays when it}

comes to hurricane’s and flooding, these virtues are accepted as an almost daily routine of Houston’s residents (as argued by Van Dijk).

From this subjective and anecdotal perspective about citizen behavior and emergency response by governmental authorities during a disaster, the question rose how the behavior of citizens and the emergency response by local governmental authorities of Houston actually played out from a scientific approach. Therefore this study aims investigating the interaction between citizen behavior and governmental authorities in relation to academic knowledge about emergency response.

This goal of this research was to investigate emergency response of citizens and governmental authorities after a natural disaster strikes. In particular the objective of this research was to evaluate citizen response based on so-called ‘disaster myths’ and how authorities anticipated upon these disaster myths in their emergency response after a natural disaster hits. Additionally the aim was to investigate whether in the current digital age, social media as communication platform changed the way in which these myths became visible in the development of a disaster. In this research the recent natural disaster in Houston as a result of hurricane Harvey (August 2017) was used as a case study. Therefore the following research question has been formulated:

1_{Arjen van Dijk, Dutch nationality, works as a Commercial Director at global oil and gas firm Boskalis Offshore Marine}

Services B.V. & Fairmount. When hurricane Harvey hit Houston, Van Dijk was at the Houston department of ‘Boskalis’ and was an eye-witness of the impact of hurricane Harvey.

4

‘How did social media influence citizen behavior (as suggested by disaster myths formulated by Helsloot & Ruitenberg 2004) and how did social media influence the emergency response of local governmental authorities during and after hurricane Harvey in Houston, Texas?’

A large amount of research has already been done on disaster management (e.g. Boin 2009; Roux-Dufort: 2007). Also research has been done on traditional emergency response during natural disasters such as hurricanes (e.g. Parker et al. 2009; Houston et al. 2016). The body of literature on disaster management and emergency response pays little attention to a more interconnected world through the cyber space (Kuipers & Welsh 2017: 9). In particular a knowledge gap is visible on how the rise of internet usage and social media are increasingly a part of the daily lives of people in relation to disaster management (Kuipers & Welsh 2017: 9).

Furthermore research has been done on disaster myths. Research into disaster myths has focused on the belief of society into disaster myths (e.g. Wenger et al. 1975), the influence of disaster myths into popular culture (Mitchell et al. 2000), how disaster myths influence decision making (e.g. Fischer 1998) and research into particular disaster myths such as looting (e.g. Tierney et al. 2006). A knowledge gap in the literature on emergency response and disaster myths in relation to the cyber space (e.g. social media and internet) is however visible.

This study is academically relevant, because it has got as objective to examine whether these disaster myths remain standing in a society which is increasingly intertwined with the cyber space. This study is furthermore academic relevant, because it aims at examining emergency responses during and directly after hurricane Harvey in Houston, Texas. An in-depth study of the actual presence of these disaster myths in relation to developing communication tools, might contribute to the knowledge about expected behavior of citizens in the emergency response by governmental authorities.

This study is moreover relevant from a societal point of view, because it examines the use of emerging and evolving communication technologies in disaster communication. Knowledge about emerging communication technologies might aid and inform policymakers responsible for emergency response during a similar disaster and in its direct aftermath. Furthermore the influence of social media on the rejection or confirmation of disaster myths visible in citizen behavior in Houston, informs policymakers on steps to undertake in similar disasters in the future. The results from this research could influence at a local level how social media is applied as operational and strategic tool in emergency response of governmental authorities and how emergency-decision-making develops.

5

2. Theoretical framework

In the scholarly field of disaster research it is argued that both citizens and actors such as governments believe in the occurrence of ‘disaster myths’ during a disaster (Tierney et al. 2006). The research into disaster myths developed over the course of time, starting with research in the 1940’s and 1950’s in the United States (Tierney et al. 2006). This research focused on how citizens could respond to a nuclear attack and whether this respond would result in panic and social breakdown (Quarantelli 1987). Although these studies showed that victims of a disaster did not act in a panicked and unmoral matter, research into the visibility of such behavior continued. For example research into the myth that people panic after a disaster increased after the terrorist attacks on 9/11 (Tierney 2003). Other examples of such disaster myths are looting and corresponding with that social lawlessness (Tierney et al. 2006). Tierney et al. (2006: 59) argue that although the scholarly field of disaster research recognizes ‘panic’ and ‘helplessness’ as mythological beliefs in disaster response, governments and international organizations such as the Red Cross base their policies of emergency response on the actual occurrence of such citizen behavior during a disaster. Tierney et al. (2006: 58) discuss in their research of hurricane Katrina that traditional media played an important role in such persistent belief in specific disaster behavior.

Helsloot and Ruitenberg (2004) put citizen response to disasters and how governmental authorities organize emergency response together in one literature review. Therefore in this theoretical framework the disaster myths as formulated by Helsloot and Ruitenberg (2004) is elaborated on. The research into disaster myths in this literature review focusses on the following topics. Firstly how citizens respond to a disaster, during and after the event. Secondly whether governmental authorities perceive that citizen will act upon the disaster myths and therefore base their emergency response on the occurrence of such behavior (Helsloot & Ruitenberg 2004; Tierney et al. 2006).

Traditional media and communication tools influence how citizens respond to a disaster (Helsloot & Ruitenberg 2004; Tierney et al. 2006). The internet and social media developed as new communication tools for both citizens and (governmental) authorities (e.g. Guth 2008). Therefore disaster myths are put in relation to social media use in the event of a natural disaster. In particular the focus of this research is how research into social media use after a disaster strikes can be put in relation to ‘disaster myths’. Firstly myths relating to citizen response are discussed. Secondly citizen response is linked to social media usage. Thirdly traditional emergency response by governmental authorities is elaborated on. Fourthly institutional based

6

observations regarding emergency response is linked to social media usage. Disaster media usage is elaborated during and after disaster (Houston et al. 2016: 7).

Helsloot and Ruitenberg (2004) identify three types of myths in the academic literature of citizen response during a disaster. Helsloot and Ruitenberg (2004) argue that these myths are still the foundation of governments in preparing for disaster management. The first disaster myth Helsloot and Ruitenberg (2004) describe is that ‘citizens panic in a disaster’. The second disaster myth is that ‘citizens are helpless and dependent’. The final disaster myth regarding citizen response during a disaster Helsloot and Ruitenberg (2004) identify is ‘looting occurs during and after a disaster’. In the following paragraphs these particular disaster myths are put in relation to literature on social media usage during a disaster.

2.1 Citizen Response

The literature on disaster myths indicates that behavior such as ‘panic’ is not a normal response of citizens in case of a disaster (Tierney et al. 2006). More likely behavior of victims of a disaster, is citizens independently providing help for themselves and for others (Tierney et al. 2006). However, as mentioned in the introduction, emergency response by authorities is still predominantly focused on the assumptions that these disaster myths are real. Anticipation on the occurrence of these ‘disaster myths’ such as ‘panic’ amongst citizens by authorities responsible form emergency response, might be problematic (Tierney et al. 2006). Belief in these ideas could influence the response of authorities during and after a disaster. Tierney et al. (2006) argue that wrong assumptions about citizen response might cause incorrect use of public resource by authorities to provide emergency response.

Furthermore research shows that traditional media are important elements in portraying a disaster (Tierney at al. 2006). This research shows that traditional promote the presence of certain citizen response on disasters (e.g. panic and looting). The anticipation of such behavior through the media might therefore influence certain emergency response by the governmental authorities. This paragraph provides a framework of the disaster myths based on the research of Helsloot and Ruitenberg (2004). Furthermore will be assessed how social media as new media platforms might reinforce or weaken the development of ‘disaster myths’ during a disaster. This theoretical framework provides tools to analyze the presence of these ‘disaster myths’ of the particular case study in this research. Moreover this framework provides tools to measure whether social media as new communication platform reinforces or holds back on such behavior from occurring amongst victims of a disaster.

7 2.1.1 Panic

In the following paragraph ‘panic’ as disaster myth is discussed. Furthermore the possible influence of social media on ‘panic’ as disaster myth is elaborated on.

2.1.1.1 Citizens panic in a disaster

The first ‘disaster myth’ discussed is panicked behavior of citizens in a disaster. There is consensus amongst scholars, that ‘panic’ as response of citizens during a disaster is not the most likely behavior of victims (Quarentelli 1954; Helsloot & Ruitenberg 2004; Tierney et al. 2006). Helsloot and Ruitenberg (2004: 102) however assess that no consensus in the literature is made about what ‘panic’ is.

Quarantelli is an important scholar in the field of panic of citizens during crises and disasters. Quarantelli (1954: 272) argues that in general ‘panic’ can be defined as ‘an acute fear reaction marked by a loss of self-control which is followed by non-social and non-rational flight’. Perry and Lindell (2003: 52) assess that observers of disasters interpret fleeing from a threat as evidence for ‘panic’. Perry and Lindell (2003: 52) however question whether such behavior could be seen as ‘panic’ when put in relation to Quarentelli’s (1954) definition. Fleeing from a threat to increase protection does not accord with Quarantelli’s criteria of being ‘non-rational’ or ‘non-social’. Helsloot & Ruitenberg (2004: 102) and Verhees (2008: 12) argue that although authorities assume ‘panic’, citizen response in case of a disaster hardly results in ‘panic’. Moreover citizens quickly act to secure themselves and others (Helsloot & Ruitenberg 2004).

In order to research what situations do lead to panic amongst victims of a disaster, Perry and Lindell (2003) identified four factors which could result in panic. These factors are based on the perceived threat individuals believe they face (Perry & Lindell 2003: 52). 1. Perception of immediate and serious danger, 2. Perception of a few escape routes, 3. Perception that escape routes are closing which make instant escape imminent. 4. Lack of communication. These four factors provide a framework to measure whether and how ‘panic’ did or did not take place in a particular disaster (see paragraph 3.2 ‘operationalization’).

2.1.1.2 Panic and social media

As discussed before, traditional media have the opportunity to influence emergency response by governmental authorities after a disaster strikes (e.g. Tierney et al. 2006). Academics (e.g. Houston et al. 2016; Ferris et al. 2016) focused their research on disaster social media usage

8

and identified that social media and the use of mobile technology potentially influences the behavior of victims during a disaster. Social media as communication tool might impact ‘panic’ as disaster myth in various ways.

First of all, social media is a tool potentially weakening ‘citizens panic in a disaster’ as disaster myth. Houston et al. (2016: 8) and Ferris et al. (2016: 207) indicate that in practical manners social media is a tool which can aid at connecting individuals amongst each other in a disaster. Ferris et al. (2016: 207) argue that during a disaster, communication through social media greatly influences fear and hope of civilians. Fear and panic can be reduced by highlighting evacuation routes through mobile devices (Ferris et al. 2016: 207). In various case studies into disaster management, different results are visible concerning the usage of social media and internet by citizens. Overall, Twitter is perceived as an important information tool. For example Gleason (2013: 967) discusses that people use Twitter as platform to acquire the most recent news and information about an event.

Secondly, social media is a tool potentially strengthening ‘citizens panic in a disaster’ as disaster myth. Ferris et al. (2017: 207) argue that social media as communication platform could besides positively, also negatively influence fear and hope and instill panic. For example through receiving false information, and de-motivating attitudes of an individual’s community. Furthermore Guth and Alloway (2008: 17) researched internet usage among civilians in the event of hurricane Katrina. They identified that respondents only see ‘moderate’ value in using internet as information and communication platform during crisis conditions (Guth & Alloway 2008: 17). Concerning the communication with emergency responders, Veil et al. (2011: 120) add to this observation. They argue that it is important to take both traditional and social media into account as information platform for citizens and emergency responders. Eriksson and Olsson (2016: 200) however argue that social media is primarily used to contact friends and family. Traditional media are perceived as a more credible source of information on the disaster (Eriksson & Olsson 2016: 200). Based on the theory in relation to the first disaster myth the following propositions are formulated:

Proposition 1.1: Citizens show panicked behavior during and directly after a disaster strikes, when Perry & Lindell’s (2003) four factors apply.

Proposition 1.2: Social media weakens the possibility of panicked behavior of citizens during and directly after a disaster strikes.

9 2.1.2 Helplessness and dependency

In the following paragraph ‘helplessness and dependency’ as disaster myth is discussed. Furthermore the possible influence of social media on ‘helplessness and dependency’ as disaster myth is elaborated on.

2.1.2.1 Citizens are helpless and dependent

From the research into the reaction of the American public towards a nuclear attack in the 1950’s, the United States government intended to find out whether citizens are helpless and how they adapt during such a disaster (Tierney et al. 2006: 58). Tierney et al. (2016: 59) however argue that traditional media influenced the general public and organizations to believe that citizens are helpless when a disaster strikes.

Research into citizen response however indicates that citizens do not wait for emergency aid, but help each other in a proactive manner (e.g. Quarentelli 1954; Tierney et al. 2006). Therefore the second ‘disaster myth’ elaborated on is that citizens are helpless and dependent during and after a disaster strikes. Helsloot and Ruitenberg (2004: 102) argue that citizens are the first to start rescue activities, take care of victims and start reconstruction. Moreover Helsloot and Ruitenberg (2004: 103) assess that citizens do not primarily dependent on help from the government and international organizations. Victims of a crisis often go to friends and family to find shelter and help (Helsloot & Ruitenberg 2004: 103).

2.1.2.2 Social media and dependability

As discussed in paragraph 2.1.1.2 about ‘panic’, social media usage and mobile technology could influence the behavior of victims when a disaster strikes (e.g. Houston et al. 2016; Ferris et al. 2016). In this paragraph is assessed in what ways social media as communication tool might influence ‘helpless and dependent’ behavior of citizens after a disaster strikes.

First of all, social media is a tool which potentially strengthens ‘helplessness and dependent’ behavior of citizens as disaster myth. Keim and Nooij (2011: 52) argue social media empowers citizens, increases adaptability of local residents and makes information accessible for everyone. Social media is in particular a helpful tool for people living in remote locations, because the collecting and assessing of information is no longer restricted by their geographical location (Ferris et al. 2016: 224; Keim & Nooij 2016: 5). Increasing accessibility of information allows citizens to share information with each other about a disaster (Keim & Nooij 2011: 52).

10

As a result victims can adapt according to this information without being dependent on journalists and governments (Houston et al. 2016: 5).

Houston et al. (2016: 10) furthermore argue that social media is more reliable in the event of a crisis, which might explain why requests for help are communicated through social media platforms. Therefore accessibility for everyone makes social media platforms an important tool for policymakers to communicate with citizens (Ferris et al. 2016: 224). Ferris et al. (2015: 224) for example conclude from their research into hurricane Sandy, that social media and the usage of mobile devices are vital in the preparation, planning and evacuation phases in case of an hurricane.

Secondly, social media is a tool potentially weakening ‘helplessness and dependency’ of citizens in a disaster as disaster myth. As argued by Keim and Nooij (2011: 52) accessibility of information through social media platforms is important to ensure communication between victims of a disaster. Moreover when social media is accessible, usage of media platforms such as Twitter and Instagram increases in case of a disaster, while the usage of traditional communication tools (e.g. phone) decreases (Ferris et al. 2016: 208). The technologies social media platforms rely upon are however fragile (Veil et al. 2011: 119). When communication networks stop functioning, because of damage caused by the disaster, accessibility to social media platforms by citizens ceases (Veil et al. 2011: 119). Consequently empowerment of citizens, information sharing and adaptability of local residents through social media decreases. Based on the theory in relation to the second disaster myth, the following propositions are formulated:

Proposition 2.1: Citizens are not helpless and dependent on emergency response from the government during and directly after a disaster strikes.

Proposition 2.2: Social media prevents helpless and dependent behavior of citizens during and directly after a disaster strikes.

2.1.3 Looting

In the following paragraph ‘looting’ as disaster myth is discussed. Furthermore the possible influence of social media on ‘looting’ as disaster myth is elaborated on.

11 2.1.3.1 Looting occurs during and after a disaster

Research into citizen response during and after a disaster strikes moreover focusses on possible changing behavior of citizens. Focus is on changing behavior of citizens, because law enforcement structures might be absent as a result of a disaster (Tierney et al. 2006: 58). Such behavior is for example related to illegal activities such as looting. The scholarly field of disaster research however assesses that less indecent behavior occurs after a disaster then prior to a disaster. Tierney et al. (2006: 59) argue this is due to social cohesiveness and mechanisms of social control. Therefore the third disaster myth that is discussed is the occurrence of ‘looting’ during and (shortly) after a disaster (Helsloot & Ruitenberg 2004: 103).

`Various scholars argue that ‘looting’ after a disaster strikes does occur (Helsloot & Ruitenberg 2004: 103; Tierney et al. 2006; Lapierre & Moro 2001). Helsloot and Ruitenberg (2004: 103) asses that looting is only done by lone individuals from outside the community. Helsloot and Ruitenberg (2004: 103) furthermore argue that media report that looting takes place during and after a disaster. In particular occurs amongst aid workers. Tierney et al. (2006: 64) however discuss that looters operate from their community and are not ‘lone individuals’ as argued by Helsloot and Ruitenberg (2004: 103).

The literature furthermore discusses possible characteristics of the individuals turning to looting behavior. No dominant reason for looting behavior of individuals during and after a disaster appears from the literature. Possible explanations for looting behavior ranges from retaliation to simply getting a hand on free merchandise (Tierney et al. 2006: 64).

2.1.3.2 Social media and looting

As discussed before, traditional media have the opportunity to influence emergency response by authorities during and after a disaster (e.g. Tierney et al. 2006). Tierney et al. (2006: 57) and Helsloot and Ruitenberg (2004: 103) argue that traditional media in describing events of looting after a disaster strikes do not portray the situation to its reality. Traditional media put emphasis on lawlessness and need more strict social control (Tierney et al. 2006: 64). Helsloot and Ruitenberg (2004: 103) argue that governmental authorities take measures to prevent looting from occurring based on the information from these media platforms. Such measures are an increased role of the military in disaster management (Tierney et al. 2006: 57). The military are perceived as an actor bringing order into a situation of chaos (Dynes 1994) (see paragraph 2.2). Academics (e.g. Houston et al. 2016; Ferris et al. 2016) researching disaster social media usage, identified that social media and the use of mobile technology can influence the behavior of victims during a disaster. First of all, social media is a tool potentially strengthening ‘looting’

12

as disaster myth. Social media is used as a tool to enable a state of order and prevent a situation of lawlessness after a disaster. In particular social media is used as a tool by governmental authorities to prevent looting. Firstly to monitor social media platforms during and after a disaster and to identify possible ‘chaotic’ behavior (Houston et al. 2016: 9). Secondly social media is used as a tool to build community resilience against ‘lawless behavior’ (Procopio & Procopio 2007; Houston et al. 2016: 15). Procopio and Procopio (2007: 15) argue that connections made through social media stimulates proper behavior during and after a disaster. Secondly , social media is a tool potentially weakening ‘looting’ by citizens in a disaster as disaster myth. The scholarly field of disaster management and emergency response does not discuss disadvantages of social media as tool to monitor citizens by authorities and stimulate proper behavior to prevent looting after a disaster strikes. Similar to the second disaster myths (citizens are helpless and dependent) dangers might lie in depending on a fragile technology to maintain a state of order. Based on the theory in relation to the third disaster myth, the following propositions are formulated:

Proposition 3.1: Citizens do not turn to looting related behavior during and directly after a disaster strikes.

Proposition 3.2: Social media decrease the possibility of looting related behavior during and directly after a disaster strikes.

2.2 Anticipation by governmental authorities on citizen response

As discussed in the previous paragraph, governmental authorities in charge of emergency response after a disaster strikes, base their policies on the belief that citizen response formulated in the academic world as ‘disaster myths’ are likely to occur (Tierney et al. 2006: 59). Moreover is discussed that traditional media played an important role in the persistent belief of such citizen response in case of a disaster. Tierney et al. (2006) argue that wrong assumptions about citizen response potentially results in misuse of public resources to provide emergency response. Therefore how a disaster is portrayed in the media and how authorities use this frame for emergency response, influences how the general public perceives citizen response after a disaster strikes.

Traditionally emergency response by governmental authorities is based on the ‘C3’ (chaos, command and control) military doctrine (Dynes 1994). This doctrine assumes that

13

disasters cause ‘chaos’ and therefore ‘command’ and ‘control’ are necessary to stabilize the situation in case of a crisis (Helsloot & Ruitenberg 2004: 204). Disaster management by governmental authorities has expanded into cooperation with other actors such as aid organizations and advising institutions. Helsloot and Ruitenberg (2004: 104) identified elements in which this C3 structure is still visible in emergency response by authorities. For example the reluctance to use conventional means of communication of disasters and the assumption that victims are passive in their behavior and cannot take care of themselves (Helsloot & Ruitenberg 2004: 104). Tierney et al. (2006: 76) moreover argue that as a result of failed response by governmental authorities to hurricane Katrina in 2005, entities able to use force (such as military institutions) are perceived as most capable of halting civil unrest and re-establishing order in similar disasters.

In this paragraph focus is on the emergency response by governmental authorities when a disaster strikes. The different elements in which the C3 structure is visible are assessed. Moreover social media as new media and communication platform is put in relation to these different C3 structures. This literature review provides a framework for the analysis to assess to what extent governmental authorities take ‘disaster myths’ into account in their emergency response.

2.2.1. Social media as unconventional communication tool

An element of traditional emergency response by governmental authorities in which the C3 structure is visible, is that governmental authorities only make use of conventional means of communication (e.g. walkie-talkies and phones) in the case of disasters (Helsloot and Ruitenberg 2004: 104; Eriksson & Olsson 2016). Various scholars (e.g. Yates & Paquette 2010; Wagner & Majchrzak 2010) discuss that authorities identified the potential of social media in disaster communication. Do governmental authorities however make use of social media platforms as communication tool in emergency response after a disaster strikes?

Overall no consensus amongst scholars is visible about social media use in emergency response by governmental authorities. Denis et al. (2014: 737) assess that relatively little adoption of social media by public actors has been made. Guth (2008) and Eriksson & Olsson (2016) however argue based on interviews with emergency responders for their research into disaster management, that actors responsible for communication in a disaster value the advantages of social media. Especially how easy and fast information can be distributed and delivered is valued (Yates & Paquette 2010: 7). For example during communications with civilians (Guth 2008: 4).

14

Moreover no uniformity on the accommodation of social media into emergency aid is visible amongst public officers (Eriksson & Olsson 2016: 200). Emergency responders do not know how to accommodate social media into emergency response procedures, because a broadly accepted body of knowledge about the advantages of social media in relation to disaster management does not exist. Another explanation for governmental authorities to not adopt social media in disaster management, is that using these social media platforms without uniformity on its accommodation might result in political embarrassment (Liu 2014: 235).

Furthermore an extensive body of literature exists about the changing role of traditional news media in relation to new social media platforms (e.g. Ferris et al. 2016; Houston et al. 2016; Veil et al. 2011). In particular how these media platforms influence policy response to citizens in need of help as a result of a disaster. Social media has got a bigger capacity, dependability and interactivity. These are advantages for actors involved in crisis communication, both citizens and authorities. Moreover these advantages might weaken ‘panic’ and ‘dependency and helplessness’ as disaster myths (Helsloot & Ruitenberg 2004: 103).

Incorporating social media in policy response therefore changes the role of public information officers (Ferris et al. 2016: 208). Veil et al. (2011: 119) argue public actors have concerns about technological failure, uncertain information and how quickly misinformation can be distributed. The risk of these problems is they might create more issues for actors managing a crisis. These problems in relation to social media use by authorities in emergency response might strengthen ‘panic’ and ‘dependency and helplessness’ as disaster myths. Uncertain and slowly distributed information might result in ‘panic’ as proper information about the event and activities of the governmental authorities are key elements for citizens not to panic after a disaster strikes (Helsloot & Ruitenberg 2004: 103). Technical failures of platforms in emergency response on which victims of a disaster depend, could result in increased ‘helplessness’ of citizens’ (Helsloot & Ruitenberg 2004: 103). Based on the theory in relation to disaster management the following proposition is formulated:

Proposition 4.1: Authorities make limited use of social media to communicate with victims after a disaster strikes.

Proposition 4.2: Social media use in the emergency response by authorities after a disaster strikes strengthens ‘panic’ and ‘helplessness and dependency’ as disaster myths.

15 2.2.2. Social media to collect information for victims

Another element of traditional emergency response by authorities in which the C3 structure is visible, is the assumption of governmental authorities that after a disaster strikes citizens are not capable of collecting correct information, and therefore need to be informed (e.g. press conferences, official statements of authorities) (Helsloot & Ruitenberg 2004: 104).

The literature on emergency response and social media describes that social media platforms offer citizens in a disaster the opportunity to collect their own information and to inform each other (e.g. Eriksson & Olsson 2016; Houston et al. 2016; McCormick 2016). Social media use during emergency response by governmental authorities potentially results in more accessible information for victims to collect (Keim & Nooij 2011: 52). Social media use by governmental authorities for victims of a disaster to collect information, could strengthen ‘helplessness and dependency’ of citizens as disaster myth (c.f. Helsloot & Ruitenberg 2004).

McCormick (2016: 2014) however identified that local governmental authorities did not use social media to inform the victims about formal response procedures of hurricane Irene, but made use of traditional media platforms (e.g. television and radio). Local governmental authorities perceived traditional media platforms as a more efficient tool to inform the entire (local) audience (McCormick 2016: 2014). Therefore social media might not be a tool for governmental authorities to inform the victims and for to victims to collect information about the disaster. Consequently social media is possibly not of influence on the weakening or strengthening of disaster myths during and after a disaster.

Informal emergency procedures do rely on social media such as Twitter and Facebook (McCormick 2016: 2014). Veil et al. (2011: 110) for example address that disaster social media potentially motivate people to volunteer and become more involved in helping victims of a disaster. Volunteers often conduct search and rescue activities (Eisman et al. 2016: 3). Helsloot and Ruitenberg (2004: 104) assess there is a general lack of trust by governmental authorities on independent actions of volunteers or groups which are not accounted for in the original contingency plans. Spontaneous actions are not perceived as controllable, and as such seen as irrelevant of disruptive (Helsloot & Ruitenberg 2004: 104).

To a certain extent the observation of Williams et al. (2012) who assess that community based social media tools are a useful tool for governmental emergency responders, are met. Community based social media tools might be useful for emergency responders in informal response procedures. In formal response procedures, social media is not a tool used by governmental emergency responders to collect information for victims about a disaster. This observation weakens the disaster myth about citizens being helpless and dependent of

16

governmental authorities after a disaster strikes (c.f. Helsloot & Ruitenberg 2004). ‘Helplessness and dependency’ is weakened as disaster myth, because in order to collect information about formal response procedures, victims depend on information from governmental authorities instead of information from community-based platforms (Helsloot & Ruitenberg 2004: 103). Based on the theory in relation to disaster management, the following propositions are formulated:

Proposition 5.1: In a disaster victims are not dependent on correct information from governmental authorities.

Proposition: 5.2: Victims solely collect information about a disaster from community-based forums instead of authority-community-based social media platforms.

2.2.3.Cooperation between victim and emergency responder

Another element of traditional emergency response by authorities in which the C3 structure is visible, is the assumption of governmental authorities that victims are passive and cannot take care of themselves (Helsloot & Ruitenberg 2004: 104). Helsloot and Ruitenberg (2004: 105) argue that a large part of disaster management in its acute stages is attributed to problem solving of victims.

Ferris et al. (2016: 225) argue that effective communication between governmental authorities and citizens is necessary to solve the problems of victims in a disaster. Also Ferris et al. (2016: 225) discuss that effective communication and cooperation between victim and emergency responder is based on multiple strategies of which social media is one. Solely focusing on social media by emergency responders is not perceived as effective, because social media is one of the many tools which might be used for effective cooperation between victims and emergency responder.

Denis et al. (2014: 737) argue that social media is a useful tool for victims and emergency responders to raise awareness about the disaster and what measures to take to provide help for those in need. Therefore both groups are involved in solving problems and they both share responsibilities to coordinate disaster management. For example Houston et al. (2016: 8) identify that authorities could spread disaster warnings via social media such as Twitter.

Furthermore social media could be beneficial for the cooperation between victim and emergency responder in the facilitation of donations (e.g. money, food and clothing) and

17

arranging shelter locations (Keim & Nooij 2011: 51). As a result social media increases the actors directly involved in helping the victims of a disaster. Based on these notions it can to a certain extent be stated that ‘panic’ and ‘helplessness’ are strengthened as disaster myths. Through social media, victims are informed by emergency responders (citizens and governmental authorities) about the possible threat they face by (Perry & Lindell 2003: 52). Disaster myths are however weakened, because the cooperation between victim emergency responder through social media to a certain extent requires a dependency of victims upon the governmental authorities. Based on the theory in relation to disaster management the following propositions are formulated:

Proposition 6.1: Authorities make use of social media in emergency response procedures based on the assumption that victims cannot take care of themselves.

Proposition 6.2: Social media usage during disaster management is a tool for both emergency responders and victims to work together in solving problems and emergency response.

18

No. Propositions

1 1. Citizens show panicked behavior during and directly after a disaster strikes, when Perry & Lindell’s (2003) four factors apply.

2. Social media weakens the possibility of panicked behavior of citizens during and directly after a disaster strikes.

2 1. Citizens are not helpless and dependent on emergency response from the government during and directly after a disaster strikes.

2. Social media prevents helpless and dependent behavior of citizens during and directly after a disaster strikes.

3 1. Citizens do not turn to looting related behavior during and directly after a disaster strikes.

2. Social media decrease the possibility of looting related behavior during and directly after a disaster strikes.

4 1. Authorities make limited use of social media to communicate with victims after a disaster strikes.

2. Social media use in the emergency response by authorities after a disaster strikes strengthens ‘panic’ and ‘helplessness and dependency’ disaster myths.

5 1. In a disaster victims are not dependent on correct information from governmental authorities.

2. Victims solely collect information about a disaster from community-based forums instead of authority-based social media platforms.

6 1. Authorities make use of social media in emergency response procedures based on the assumption that victims cannot take care of themselves.

2. Social media usage during disaster management is a tool for both emergency responders and victims to work together in solving problems and emergency response.

19

3. Research design and methodology

In the following paragraph the research design is discussed. Secondly the operationalization of the theoretical framework is elaborated on. Thirdly the sampling strategies used in this research are assessed. Fourthly the data collection is discussed. Fifthly the data analysis is elaborated on and finally the limitations applicable to this research are assessed.

3.1 Research Design

To answer the research question, this research is based on an explorative holistic single case study design (see figure 2, p. 20). The case researched in this study is hurricane Harvey in Houston, Texas and the results are based on qualitative research methods. The research is based on an inductive method of reasoning, because from the theory on ‘disaster myths’ is tested whether social media influence citizen response and emergency response by governmental authorities after a disaster strikes (Neuman 2014: 95; Yin 2011: 94). In particular the objective of this research is to gain insights about the influence of social media on citizen response through ‘disaster myths’ during and after a disaster strikes. Furthermore the goal of this research to generate knowledge about the influence of anticipated citizen behavior after a disaster strikes (disaster myths) on the emergency response of governmental authorities. An inductive approach is moreover visible, because the research is a constant reflection on theory, empirical case and gathered data (Neuman 2014: 95; Yin 2011: 94).

An holistic single case study design is the best way to answer my research question, because emphasis is on in-depth knowledge into a specific topic through a qualitative research approach (Rigan 2008: 82-83; Yin 2011: 6). An in-depth case study provides more information about a particular case and therefore gives more detailed insight on how a case relates to the identified concepts (Rigan 2008: 83). For several reasons a holistic single case study design is the best fit for this research: 1. In general the research is about a specific topic and the main objective of this research is to provide in-depth knowledge about a particular case (Yin 1984: 25). 2. The specific context of hurricane Harvey offers insights in a natural disaster with functioning communication networks. Therefore this specific case study allows research into the influence of social media. 3. Houston as case study during hurricane Harvey is relevant, because a large set of local actors responsible for emergency response is visible. Therefore an in-depth study into the disaster management of hurricane Harvey is possible.

The goal of the qualitative approach in this research is to advance the theory on ‘disaster myths’. Careful attention must however be paid to the danger of generalization of incidents. An

20

incident such as the disaster as a result of hurricane Harvey, is an unique event and can therefore not be generalized (Onderzoeksraad voor Veiligheid 2014: 1; Yin 2011: 98). To a certain extent a generalization can be made about the possible relationship between facts (Onderzoeksraad voor Veiligheid 2014: 2; Yin 2008: 98). For example a relationship between panic of citizens in a disaster and the absence of communication through social media platforms. Multiple similar cases must however be researched for such assumptions to be generalized.

For the qualitative part of this research, document analysis has been done on social media messages from for example Twitter. Furthermore case study research has been done on ‘hurricane Harvey’. Finally literature analysis has been done to define concepts which are used as analytical frames to study the research objectives (Rigan 2008: 83).

21

3.2 Operationalization into concepts, definitions and indicators

In the following paragraph the research question is operationalized into independent and dependent variables. The concepts central in this research are defined based on the theoretical framework. Furthermore these concepts are operationalized into indicators.

3.1.1 Independent variables

‘Social media’ and ‘conventional media’ as concepts are the independent variables of this research. Based on the indicators from these concepts is explained whether the independent variables could influence the dependent variables.

Concept Definition Indicators Data source

Social media “… a broad term for a variety of web-based platforms and services that allow users to develop public or semi-public profiles and/or content, and to connect with other users’ profiles and/or content”

- Web based platforms - Public/semipublic content - Connection with at least 1 other profile/content on social media platforms Houston et al. (2016: 3) Conventional/ Traditional media - - Print media - Television - Radio Ferris et al. (2016); Houston et al. (2016); Veil et al. (2011) 3.1.2 Dependent variables

‘Panic’, ‘helplessness and dependency’ and ‘looting’ as concepts are the dependent variables of this research. In this research is measured whether the dependent variables such as looting are influenced by social media as communication tool. Moreover is measured how these dependent variables are visible in messages of both citizens and local governmental authorities.

Concept Definition Indicators Examples data Data source

Panic “...an acute fear reaction marked by a loss of self-control which is followed by social and non-rational flight” - Perception of immediate and serious danger - Perception of a few escape routes - Perception that escape routes are closing which make

- Report from individual about leaking gas station.

- The Mayor of Houston advising citizens to stay in their houses, instead of fleeing by car.

- Reports of individuals being stuck in their

Quarantelli (1954: 272); Perry and Lindell (2003: 52)

22 instant escape imminent - Lack of communication

houses while their houses are flooding.

- Reports from

individuals not able to reach ‘9-1-1’ to ask for emergency assistance. Helplessness

and

dependency

Condition in which the threat of a disaster just about to happen or its actual impact does paralyze those affected. Passivity in the face of danger is almost non-existent - Victims of a crisis go to friends and family to find shelter and help

- Citizens are the first to start rescue activities, take care of victims and start reconstruction.

- Reports of individuals requesting help for family members in immediate danger.

- Reports of individuals using their own boats to independently start rescuing people. - Quarantelli (1993); Helsloot and Ruitenberg (2004: 102-103)

Looting “A form of anti-social behavior in the chaos of a disaster” - At least 1 shop is raided - Reports about lawlessness in a particular crisis - Reports about social unrest

- Reports of citizens about shops being raided. - Reports about

individuals imposing as police officers at people’s homes. Telling them to evacuate, in order to raid their houses.

- Reports about people fighting.

Helsloot and Ruitenberg (2004: 103)

3.3 Sampling

In this research the social media messages for document analysis are based on both purposive and convenience sampling. Purposive sampling is used to select social media messages specifically about ‘hurricane Harvey’ (Yin 2011: 311). Convenience sampling is used to select messages which are accessible to the researcher and is moreover a common sampling method used in explorative research (Bryman 2012: 710). A limitation of convenience sampling is however that the data gathered might not represent the entire population relevant for this research, as not every individual might make use of social media (Bryman 2012: 424).

23

3.4. Data collection

In the following paragraphs the data collection of this particular research is elaborated on. Firstly, the scope of the data collection is narrowed down by identifying the units of analysis and the specific timeframe in which data has been collected. Secondly, the data collection of ‘local governmental authorities’ is discussed and finally the data collection of ‘citizens’ is assessed.

3.4.1 Scope of data collection

The units of analysis in this research follow from the selected case (hurricane Harvey Houston, Texas) and the research objective, namely researching both the citizens and the authorities involved during and after hurricane Harvey in Houston. The units of ‘analysis’ in this research are therefore ‘the citizens of Houston’ and ‘the local governmental authorities Houston’.

In the literature no exact definition is visible about the ‘immediate aftermath’ of a natural disaster. Overall ‘immediate aftermath’ is however operationalized as ‘the first hours and days’ of a disaster (Boin & McConnel 2007: 54; Kemmelmeier et al. 2008: 220). In the case of hurricane Harvey this is difficult to operationalize, because not only the wind, but rainfall resulting in flooding added to the scope and scale of the disaster (see chapter 4 ‘Case time line’). Therefore the period researched is from the 24th _{of August 2017 until the 30}th _{of August 2017.}

In this research a triangulation of methods is used (case study analysis, document analysis, literature analysis). Through this triangulation, the research subject is analyzed from different perspectives though multiple data sources (Neuman 2014: 166; Bowen 2009: 28). The methods used to collect the data are desktop research, data from social media platforms and data from traditional media platforms.

The criterion for data collection was that the data is about emergency response after hurricane Harvey strikes in Houston, Texas. This criterion refers to messages from both the citizens of Houston and governmental authorities on a local level responsible for emergency response. The social media platforms from which data has been collected are: Websites, Twitter, Facebook and Instagram. Over a thousand social media reports have been collected and analyzed of both the citizens of Houston and local governmental authorities.

3.4.2 Data collection citizens of Houston

In order to collect data from the citizens of Houston, the different actors to collect data from were identified. Through desktop research about hurricane Harvey, individuals and

24

organizations regarding ‘citizen response’ after hurricane Harvey hit Houston were identified. These organizations are: the Cajun Navy Relief2_{, the Red Cross Houston, the Houston Food}

Bank.

After these actors were identified, information regarding emergency response of hurricane Harvey in Houston was collected from the Twitter-page, Facebook-page and if applicable the website. Data was collected by searching the ‘Twitter name’ (e.g. @RedCrossHouston) and by limiting the time scope of the messages posted to the 24th _{– 30}th

of August 2017 (the immediate aftermath). Data from the Facebook pages of the identified actors were again collected by selecting the month ‘August’ and gathering posts from the 24th

until the 30th _{of August 2017. Messages regarding emergency response of hurricane Harvey in}

Houston were selected for document analysis.

Data collection from individual citizens of Houston has been gathered from Twitter, Facebook and Instagram. Messages from Twitter and Instagram were gathered by firstly identifying the main ‘hashtags’3 _{used regarding hurricane Harvey and emergency response in}

Houston, secondly by searching ‘Houston’ as specific location and thirdly selecting the 24th _of

August – 30th _{of August 2017 as specific time period. Collecting individual messages through}

Facebook was not possible due to the ‘private status’ many individuals have got on their accounts. Therefore firstly Facebook pages from the Facebook group ‘Hurricane Harvey 2017’4

were collected. Secondly messages posted by individuals on social media platforms of local authorities (e.g. the Houston Police Department). Finally desktop research was used to gather data which otherwise could not be found on social media platforms, such as the individual rescue missions set up by ‘Keri Henry’ through Facebook (see analysis paragraph 5.1.2).

3.4.3 Data collection local governmental authorities of Houston

In order to collect data from the local governmental authorities of Houston, firstly the specific local authorities were identified. Through desktop research about hurricane Harvey, the local governmental authorities involved in the emergency response after hurricane Harvey hit Houston were identified. These local actors were: the FEMA (Federal Emergency Management Agency), the Senator of Texas ‘Ted Cruz’, the Governor of Texas ‘Greg Abbott’, the Texas National Guard, the U.S. Coast Guard, the Houston-OEM (Office of Emergency Management),

2_{The Cajun Navy consist of informal groups whose members are private owners of boats. Those members actively participate}

in search and rescue activities in the state Louisiana. In the aftermath of hurricane Harvey the Cajun Navy was active in Houston as well.

3_{Identified hashtags: #Harvey, #Houston, #HarveyRelief, #HoustonStrong, #Looting, #Help.} 4_{A Facebook group founded for individuals with questions and information about hurricane Harvey.}

25

the Mayor of Houston Sylvester Turner, the Houston Police Department, the Houston Fire Department and the U.S. Department of Justice.

After these actors were identified, information regarding emergency response of hurricane Harvey in Houston was collected from the Twitter-page, Facebook-page and if applicable the website. Data from the Twitter-pages of each actor was collected by searching the ‘Twitter name’ (e.g. @Houston-OEM) together a limited time scope of the messages posted to the 24th _{– 30}th _{of August 2017 (the immediate aftermath). Data from Facebook-pages of each}

actor was selected by searching the Facebook page and selecting the month ‘August’. Posts on the Facebook page between the 24th _{and the 30}th _{of August 2017 were selected. Messages}

regarding emergency response of hurricane Harvey in Houston were selected for document analysis. Data from websites was selected for document analysis when updates regarding hurricane Harvey were posted between the 25th _{and the 30}th _{of August 2017. Instagram was not}

a social media tool used by local authorities in Houston responsible for the emergency response of hurricane Harvey.

3.5 Data analysis

After the data collection, qualitative document analysis was performed (Bowen 2009). Qualitative document analysis is a method in which documents (e.g. books, journals, social media) are systematically reviewed (Bowen 2009: 27). Through qualitative document analysis, the data is researched and interpreted to extract knowledge, increase understanding and gain empirical knowledge about a specific topic or event (Bowen 2009: 27). The documents used for qualitative document analysis could consist of words and/or images and are not adjusted by the researcher (Bowen 2009: 27).

Qualitative document analysis is the best method to analyze the data for this research for several reasons: 1. ‘Social media messages’ as documents consisting of words are the data to extract knowledge about disaster myths during and directly after hurricane Harvey hit Houston. 2. Through systematically analyzing the meaning of the social media messages, the understanding about the possible influence of social media use by both citizens and governmental authorities is increased. 3. By analyzing social media messages, empirical knowledge about disaster management in Houston is gained.

The data has been analyzed to provide knowledge about ‘disaster myths’ in the immediate aftermath of hurricane Harvey in Houston amongst local governmental authorities and citizens. The data has been analyzed through color coding of the documents. Three codes

26

have been used to analyze the data. The three codes are the central concepts (see paragraph 3.2) in this research to analyze the three ‘disaster myths’ amongst both the citizens of Houston and the local governmental authorities of Houston responsible for emergency response:

1. Yellow: Helplessness and Dependency 2. Orange: Panic

3. Pink: Looting

After the color coding of the data, the corresponding indicators of each code (and concept) have been analyzed in relation to the dataset. In paragraph 3.2 an example of each indicator looked for in the data set has been given. Furthermore an indication of the amount of messages posted about each indicator has been given in the analysis:

1. Little amount of messages about a single indicator: 0 > 10 2. Average amount of messages about a single indicator: 10 > 100 3. Large amount of messages about a single indicator: 100 ≥

The reported quantity of messages for each indicator, allowed the researcher to present to what extent each indicator and concept is visible in the data set. Based on this analysis of the data, the results have been further assessed in relation to the theoretical framework. Together this analysis has been used to answer the research question (see Chapter 6).

3.6 Limitations

In the following paragraph the limitations of the research is discussed. Firstly the reliability of the research is elaborated on. Secondly the validity is shortly addressed. Finally ethical concerns important to take into account in this research are discussed.

To increase the reliability of the research all the actors and stakeholders involved in the emergency response after Hurricane Harvey have been identified. Based on these actors social media messages of these actors in relation to emergency response have been collected. Furthermore little codes have been used, to make the measurements as consistent as possible (Neuman 2014: 218). Document analysis is however dependent on the interpretation of the researcher (Bowen 2009: 32). The bias of the researcher might influence the reliability of the researcher. To make sure this bias does not limit the research, the subjectivity of the data has

27

been taken into account (Bowen 2009: 38). This has been done by firstly emphasizing on consistency in the coding of social media messages (Neuman 2014: 218). Secondly through transparency about the data set. This transparency is established by incorporating the main results (social media messages) in the analysis and the appendices,

The external validity in this research is not optimal (Neuman 2014: 221). As this research is not experimental in nature, the results cannot completely be generalized (Neuman 2014: 221). Therefore the results from this particular research cannot be generalized to other natural disasters. Furthermore limitations appeared in the internal validity of this research, because the collected data for this research are social media messages. Although the objective of this research is to analyze social media in relation to disaster myths, social media as main data set might be one-sided. Social media messages are personal and potentially biased by the individuals who posted them. These limitations regarding the internal validity of this research have however been taken into account through systematic data collection and the coding of the data minimizes Furthermore knowledge and information gained from desktop research (e.g. newspaper articles) to support or counter the findings from the data, increases the internal validity (Bowen 2009: 38; Neuman 2014: 128).

Finally ethical concerns arose from this specific research (O’Leary 2014). A large part of data are social media messages of individuals. These social media messages are personal, but accessible to the general public. To take the privacy of each individual into account, the messages have been anonymized as much as possible. In the following chapter the time line of hurricane Harvey is given.

28

4. Case time line

This timeline is based on information from various American media platforms: ‘The New York Times’, ‘ABC News’, ‘CNN’, ‘US News’, ‘The Balance’, ‘The Washington Post’. The factual information about hurricane Harvey was the same on each media platform. Each platform focusses on different additional information about the event. The New York Times and US news focus more on the activities of the NHC. ABC news and CNN discuss the activities of actors on a national level (e.g. U.S. Department of Justice). Finally The Washington Post and The Balance focus more on activities on a local level of both the citizens as the local governmental actors.

Date Event

17th of August 2017

The National Hurricane Center in Miami warns islands in the Caribbean for a tropical storm. After 6 hours the tropical storm is named ‘Harvey’ 19th of August

2017

Harvey is downgraded from tropical depression to tropical wave as it moves along the northern coast of South-America

22th of August 2017

Harvey revives as a tropical depression in the South-East of Port O’Connor in Texas.

23th of August 2017

- Harvey develops into a Category 2 hurricane 254 km from Port O’Connor, Texas

- Texas Governor Gregg Abbott declares a state of emergency for 30 counties

24th of August 2017

- Harvey develops from a Category 3 hurricane 72 km away from Corpus Christi, Texas into a Category 4 hurricane.

- The National Hurricane Center realizes that the intensity of Harvey is more intense than anticipated.

- Harvey hits land between Port Aransas and Port O’Connor. - 250.000 people are left without power

- Cell phone towers remain functioning

- The hurricane liaison from the FEMA put information from

meteorologists into actions plans for emergency response on the ground. - FEMA works with the Coast Guard, Customs Border Protection and the Immigration and Customs Enforcement to prepare for the storm and the aftermaths.

- Corpus Christi’s mayor Joe McComb tells resident to evacuate voluntary and does not exclude mandatory evacuation as a result of the anticipated flooding.

- President Trump and Texas Governor Gregg Abbott discuss storm preparations.

- Headline of Harvey Bulletin5: ‘Preparations along the Texas coast should be rushed into completion today’

- Weather forecasters warn for ‘life-threatening and devastating flooding’ flooding in the next few days.

5 _{The NHC brought out bulletins about Harvey (Harvey bulletin) with storm surge warnings including warning of}

29

- End of the day: No more warnings in the Harvey Bulletin about preparation. Preparations which are not made yet are too late. 25th of August

2017

- Harvey is 24 km inland slows down and develops into a Category 1 hurricane together with large amounts of rain.

- The hurricane develops further as predicted by hurricane experts. - Residents strand on rooftops waiting to be rescued

- The first death of the hurricane is called in Rockport, Texas.

- FEMA creates a support base approximately 200km inland from where the storm is expected to hit land. Personnel is stocking up on utilities such as blankets and food.

26th of August

2017 - Harvey Bulletin describes that rainfall has become a bigger problem than wind. - Harvey moves on to Houston

- Two water reservoirs overflowed

- 2/3 of Houston’s Carris County becomes flooded. 4.5 million people live here

- Hard-hit areas become urban rivers as a result of storm surge from the Gulf of Mexico and extreme rainfall.

- Texas Governor Gregg Abbott declares a state of emergency for 20 more counties.

27th of August - Mandatory evacuations start to take place (e.g. Bay City)

28th of August - People are impersonating police officers in areas hit the most by the flooding. The impersonators tell residents that evacuations are

mandatory in order to pursue looting related activities.

- The Houston Police Department redirects its activities from search-and rescue activities to potential criminal activities.

- 14 looters have been arrested by the Police Department

- Governor Greg Abbott activates the whole Texas National Guard (12.000 deployed guardsmen). They are assigned to search and rescue activities for people in immediate danger.

- The US Army Corps of Engineers tries to manage flood levels in various areas

29th of August 2017

- The city of Beaumont has been without drinking water since the storm - The mayor of Houston Sylvester Turner enforces a mandatory curfew from 12 am to 5 am. Looting is named as the main reason for a curfew. 31th of August

2017 - A chemical plant in Crosby, Texas, ignited as result of disabled cooling equipment - President Trump requests for federal funding for the affected areas of hurricane Harvey.

1th of

September 2017

- One-third of Houston is under water.

- 39.000 people are homeless as a result of flooding - Dallas facilitates a shelter for 5000 evacuees - Heavy rainfall in Nashville, Tennessee

30

5. Analysis

This chapter provides an analysis of both citizen behavior and emergency response by local governmental authorities during and immediately after hurricane Harvey strikes Houston, Texas. In particular an analysis of citizen behavior and emergency response by local governmental authorities in relation to disaster myths (c.f. Helsloot & Ruitenberg 2004: 103) and social media use. In this analysis the results are elaborated on in relation to the theoretical framework. The objective of this chapter is to provide an analysis of both disaster myths and social media use during and directly after hurricane Harvey in Houston. Firstly, the results of ‘citizen behavior’ are assessed. Secondly, the results regarding ‘emergency response by local governmental authorities’ are discussed. In the analysis the propositions from the theoretical framework regarding ‘disaster myths’ and ‘social media use’ are confirmed or rejected. The table on p. 73 provides an overview of the analyzed propositions.

5.1 Citizen behavior

In the following paragraphs citizen behavior after hurricane Harvey hit Houston is elaborated on. Firstly, ‘panic’ as disaster myth is analyzed. Secondly, ‘helplessness and dependency’ is discussed and finally ‘looting’ as disaster myth is assessed (c.f. Helsloot & Ruitenberg 2004: 102: 103).

5.1.1 Panic

In the following paragraph ‘citizens panic after a disaster strikes’ as disaster myth (Helsloot & Ruitenberg 2004: 103) is analyzed. The four factors of Perry and Lindell (2003: 52)6 _{to indicate}

the chance of panicked behavior of citizens after a disaster strikes, are analyzed in relation to hurricane Harvey in Houston. The theory of Perry and Lindell (2003: 52) leads us to expect that when the four factors are visible in the findings below, the chance of panicked behavior amongst the citizens of Houston increases. The findings below discuss each factor in relation to citizen response of hurricane Harvey in Houston. Moreover is analyzed whether the propositions from the theoretical framework are strengthened or weakened in relation to the findings.

6 _{1. Perception of immediate and serious danger, 2. Perception of a few escape routes, 3. Perception that escape}