Information on social media as a contribution to disaster supply chain responsiveness

(1)

1

Information on social media as a contribution to

disaster supply chain responsiveness

Master Thesis

Supply Chain Management

Julia Wattimury S3190862

j.wattimury@student.rug.nl

University of Groningen Faculty of Economics and Business

June 22, 2018

Supervisor and first assessor: Dr. K. Scholten Second assessor: Dr. O.A. Kilic

(2)

2

Abstract

Purpose – For an effective disaster response, communication is required to set up a supply chain.

Communication could be difficult due to damaged or overburdened communication lines. Information on social media could support relief organizations in setting up a supply chain for disaster response. The purpose of this research is to explore how information on social media can contribute to supply chain responsiveness after the occurrence of a disaster.

Design/methodology/approach – A case study consisting of four disasters was conducted. 12.588

tweets were collected for all disasters based on different search terms. From these 12.588 tweets, 3196 tweets were considered relevant and were analysed in order to answer the research question.

Findings – The findings found that the size of the disaster, timestamp of the message, origin of the

tweet, the duration of the disaster, the expression of demand or supply and validation of information are mechanisms that explain how social media can help organizations to set up a supply chain to respond. These findings are presented in a set of seven propositions.

Originality/value – This research is one of the first studies that explores how information on social

media can contribute to supply chain responsiveness after the occurrence of a disaster. This research adds valuable insights to current knowledge by describing the mechanisms that explain how information on social media can help organizations to set up a supply chain to respond after the occurrence of a disaster.

Keywords – Social media, supply chain responsiveness, disaster response, humanitarian supply

chain

Paper type – Case study

(3)

3

1. Introduction

Fast diffusion of critical information about demand and the size of the affected area after a disaster are essential for providing efficient and effective disaster response (Yoo, Rand, Eftekhar, & Rabinovich, 2016). The disaster response is regarded as chaotic because it is hard to determine which resources are needed and where these resources are required (Chakravarty, 2014). This is often the result of overburdened or damaged communication lines (Chakravarty, 2011). Social media gives organizations the possibility to backup these overburdened or damaged communications lines in order to gather information on social media about the location, time, content and validity of the disaster (de Albuquerque, Herfort, Brenning, & Zipf, 2015; Imran, Castillo, Diaz, & Vieweg, 2014; Wang & Ye, 2018). This information on social media can help to set up a supply chain and coordinate a disaster response (Alexander, 2014).

(5)

5 Previous research has been done on the use of social media after a disaster and research has been done on supply chain responsiveness after a disaster has been studied. However, to the best of the authors’ knowledge, no research has been done about how information on social media could help to respond after a disaster and what type of information on social media could aid the supply chain responsiveness. Therefore, the aim of this paper is to answer the following research question: How can information on social media contribute to supply chain responsiveness after a disaster? In order to answer the research question, a case study is performed where information on social media during four disasters are analysed. This research is a contribution to previous studies because it addresses the gap between the use of social media after a disaster and supply chain responsiveness after a disaster by giving details about what mechanisms explain how information on social media can contribute to supply chain responsiveness. Furthermore, the results of this research can contribute as a guideline for organizations how to use social media when there are problems with acquiring information for disaster response due to communication problems.

(6)

6

2. Theoretical background

2.1 Disasters Responsiveness

(7)

7 Volume supply chain responsiveness

The first dimension of supply chain responsiveness is volume and determines the scope of the response activities. This dimension is linked to responsiveness in the terms of variability of demand, customer expectations, lead-time and product variety (Holweg, 2005). An example of volume responsiveness is that when a disaster occurs, the volume is clear in terms of how many victims are in need of help so that relief organizations can provide sufficient relief to all victims. Often this information is not clear after a disaster. Volume supply chain responsiveness in the disaster context experiences higher uncertainty in terms of varying sizes of demand in comparison to regular supply chains because regular transportation systems could be damaged after the occurrence of a disaster (Holguín-Veras, Jaller, Van Wassenhove, Pérez, & Wachtendorf, 2012). Another characteristic of response in a disaster setting is that volume supply chain responsiveness requires high flexibility in order to respond to extreme demands at a specific location, whereas for regular supply chains the demand is more diffused over a country and less volatile (Kumar & Havey, 2013).

Product supply chain responsiveness

(8)

8 Process supply chain responsiveness

The final supply chain responsiveness dimension is process. Process supply chain responsiveness covers the operations of getting the product to the end customer (Holweg, 2005). An example of process supply chain responsiveness in a disaster situation is the degree of how fast a specific product can be delivered to the disaster location (Danese et al., 2013). The time to supply relief may take longer because the regular supply chains are not operating for weeks because of the disaster (Holguín-Veras et al., 2012). In order to be more responsive in the process dimension, the supply chain needs to respond faster, which can be achieved by introducing improvement strategies such as Lean, Agile and Leagile (Belvedere, Grando, & Papadimitriou, 2010). These improvement strategies help relief organizations to reduce lead time by improving the process, with the result that the process is more efficient and resources are delivered faster. Another method to improve process supply chain responsiveness is to make an organization more flexible (Belvedere et al., 2010). In the context of a disaster this flexibility could be achieved by decentralizing stocks intro regional warehouses (Holguín-Veras et al., 2012).

Relationships supply chain responsiveness dimensions

(9)

9 2.2 Information on social media

“Social media employ mobile and web-based technologies to create highly interactive platforms via which individuals and communities share, co-create, discuss, and modify user-generated content” (Kietzmann, Hermkens, McCarthy, & Silvestre, 2011: p. 54). The purpose for the use of social media as a method for communication during disasters focuses on bringing awareness to the disaster situation (Imran et al., 2014) and improving the humanitarian response (Steiger, de Albuquerque, & Zipf, 2015). As such, information on social media has the potential to help with the volume, product and process supply chain responsiveness.

Information that is available on social media could have different types of subjects related to a disaster. Several studies focused on classifying these subjects into dimensions to execute an analysis of large amounts of information (Simon et al., 2015). Wang & Ye (2018) divided the information on social media into four dimensions; space, time, content and network information. Imran et al. (2014) classified social media information based on: information provided, information source, credibility, time, location and factual, subjective or emotional content. De Albuquerque et al. (2015) executed the classification of information based on if the information was on- or off-topic, after classifying the information as on-topic the following dimensions were made: volunteer action, media reports, traffic conditions, first-hand observation, official actions, infrastructure damage and other.

(10)

10 social media message has. Imran et al. (2014) described the validity of the information based on the credibility and the information of the source. De Albuquerque et al. (2015) adressed validity of the information on social media based on whether the information was on- or off-topic. The classification of the dimensions are illustrated below in table 2.1.

Wang & Ye (2018) de Albuquerque et al.

(2015) Imran et al. (2014)

Dimension: Definition:

Location Classification of social media information based on the location of the sender of the social media message

Space Location

Time Classification of social media information based on the time the social media message was posted

Time Time

Content Classification of social media information based on the content of the social media message

Content Volunteer action

Media report First-hand observation Official actions Infrastructure damage Traffic conditions Information provided Factual content Subject content Emotional content

Validity Classification of social media information based on the validity of the social media message

Network On- or off-topic Information source

Credibility

(11)

11 Location

This dimension is important for the response to a disaster because it helps to localize the point where relief is required (Wang & Ye, 2018). The information of the location of the social media message can be shared by mentioning the name of the location (toponym) or using a geotag, which gives the exact coordinates of the location the social media message was posted from (de Albuquerque et al., 2015; Wang & Ye, 2018). It is important to note that toponym location information does not have to be directly associated with the actual information from the mentioned location since the message could be posted from elsewhere in the world (Imran et al., 2014). The information on social media for the location of the disaster could be helpful for the process supply chain responsiveness in a disaster situation. This is because it makes it possible locate the people in need and to analyse the transportation networks available in that area to assess how the products could be supplied (Holguín-Veras et al., 2012).

Time

The time dimension classifies the information based on the timestamp the social media channel gives to the post (Wang & Ye, 2018). The time dimension is important for the process supply chain responsiveness because analysing the information based on the time could help to understand the stage of the disaster and to determine the response phase of the disaster (Imran et al., 2014). 72 hours after the occurrence of the disaster is marked as the most important for resource allocation during the response phase (Huang et al., 2015), and therefore this phase deals with time pressure (Schryen, Rauchecker, & Comes, 2015). Therefore, the messages posted within this timeframe are most useful to aid the supply chain responsiveness.

Content

(12)

12 Validity

This dimension classifies and selects the data based on whether the information on social media is on- or off topic (de Albuquerque et al., 2015). The classification of information based on the validity is beneficial for both product- and volume supply chain responsiveness. This is because it helps to assure that reliable information about the size and required demand is acquired. An example of a message on social media where validity is addressed, are messages where information is confirmed or corrected (Simon et al., 2015). It is important to mention that the quality of the content of a social media message could bring challenges such as the quality in terms of objectivity and the language the information contained (Imran et al., 2014).

2.3 Conceptual model

Figure 2.1 illustrates the conceptual model where the variables described in the section above are illustrated. In order for organizations to organize an effective disaster response, communication is required to gather information for setting up a supply chain. Acquiring this information could be difficult because of damaged or overburdened communication lines. Information on social media about the location, time, content and validity can be used to help set up a supply chain to provide disaster response (Alexander, 2014; J. Yin et al., 2015). However, it is not clear how information on social media could help to respond to the disaster situation and what type of information on social media could aid the volume-, product- and process supply chain responsiveness. For this, further research needs to be done on how information on social media can be used by organizations after the occurrence of a disaster (Kumar & Havey, 2013). Therefore, the research question of this paper is: How can information on social media contribute to supply chain responsiveness after a disaster?

(13)

13

3. Methodology

3.1 Research design

The research method for this paper is a case study. A case study is the appropriate method if a contemporary phenomenon is studied in its real-life context (Yin, 1981). The contemporary phenomenon of this study is the increasing use of social media after the occurrence of a disaster. The real-life context of which the contemporary phenomenon is studied is the disaster context. This study is necessary because little knowledge is available on the increasing use of social media after the occurrence of a disaster. Based on this, more in-depth knowledge is required on how the information on social could aid supply chain responsiveness in the disaster context is not clear (Kumar & Havey, 2013). Therefore, a case study is the appropriate research method for this study. The data for this research is collected by gathering information on social media messages during a disaster, therefore the unit of analysis is disasters.

3.2 Context

(14)

14 3.3 Case Selection

The cases selected aim for theoretical replication based on the different sizes of each disaster. The choice for theoretical replication is based on the fact that it is expected that the varying cases that include different locations, duration and size of the disasters will result in different outcomes in terms of how information on social media can contribute to supply chain responsiveness. Based on the aim of this case selection, four disasters with different volumes are selected. These disasters are hurricane Irma, hurricane Harvey, the Joplin tornado and the landslides in Sierra Leone. Information about these disasters is described below in table 3.1.

Case 1:

Hurricane Irma

(Halverson, 2018)

Case 2:

Hurricane Harvey

(Blake & Zelinsky, 2018; Upton, Kirsch, Harvey, & Hanfling, 2017)

Case 3:

Sierra Leone landslides

(Sierra Leone: rapid damage and loss assessment of August 14th, 2017 landslides and floods, 2017) Case 4: Joplin tornado (Kuligowski, Lombardo, Phan, Levitan, & Jorgensen, 2013)

Date 9/9/2017 – 11/9/2017 25/8/2017 – 27/8/2017 14/8/2017 22/5/2011

Location Florida, U.S.A Texas, U.S.A Freetown, Sierra Leone Joplin, Missouri

Location span

70,000 square miles 23,000 square miles 0,013 square miles 16.5 square miles

Budget

humanitarian aid

$89 million $493.3 million $4.8 million $7 million

Damage in $ (volume)

$65 billion $125 billion $31.6 million $3 billion

(15)

15 3.4 Data collection

(16)

16

Hurricane Irma Hurricane Harvey Sierra Leone landslides Joplin tornado Dates

tweet collection

8/9/2017 – 14/9/2017 24/8/2017 – 30/8/2017 13/8/2017 – 17/8/2017 21/5/2011 – 25/5/2011

Location Florida, U.S.A Texas, U.S.A Freetown, Sierra Leone Joplin, Missouri

Search terms #hurricaneirma, help #hurricaneirma, rescue #hurricaneharvey, help #hurricaneharvey, rescue #freetownfloods #sierraleonemudslide #Joplin, help #Joplin, rescue Collected tweets

4487 tweets 5706 tweets 600 tweets 1795 tweets

Exclusion

criteria - Messages posted after September 14

- Posts that are irrelevant to the disaster or posts containing the following words: pray, virgin, worried, Cuba, Caribbean, Harvey, Texas, Barbuda, Anguilla, marten, Haiti, god, maarten, prayers, lord, floridastrong - Messages posted after August 30 - Posts that are

irrelevant to the disaster or posts containing the following words: pray, worried, god, prayers, lord, praying, text

- Messages posted after August 17 - Posts that are

irrelevant to the disaster or posts containing the following words: pray, grief, shock, mourning,

prayers, praying, god, RIP, lord, heart,

condolences

- Messages posted after May 25

- Posts that are irrelevant to the disaster or posts containing the following words: pray, prayers, god, lord, worried, text, txt, telethon Relevant tweets

992 tweets 1361 tweets 240 tweets 603 tweets

(17)

17 3.5 Data analysis

After the data collection, a categorization of the raw data was done in Excel in order to use this data for an analysis. First, the relevant tweets are classified to the independent and dependent variable of this research (social media information and supply chain responsiveness). After reading the tweet, the type of information on social media was assigned to the tweet. If the tweet contained relevant information that could aid the supply chain responsiveness, the content category was assigned to the tweet. If the tweet consisted information about the location of the disaster or the need for help, the location category was assigned to the tweet. If the information of the tweet was relevant in terms of time the message was posted or mentioned in the tweet, the tweet was assigned to the time category. Finally, if the message consisted information that was relevant to the validity of the tweet, the tweet was assigned to the validity category. Besides labelling the tweet to a type of information on social media, the tweet was also assigned to one of the three supply chain responsiveness types. When the tweet contained information about the volume of the product required in terms of how many victims were affected by the disaster or how many products were required, the tweet was assigned to the volume supply chain responsiveness type. If the tweet consisted information about what product was required or supplied, the tweet was categorized in the product supply chain responsiveness category. Finally, for tweets containing information that is relevant for the process supply chain responsiveness. For example, information about how to deliver the product or how fast the products could be delivered, the tweet is assigned to the process supply chain responsiveness type. An overview of the codes assigned to the tweets can be found in appendix A.

(18)

18 requested. This descriptive code was used to establish an overview for all cases of what type of relief or resources the tweets described.

(19)

19 3.6 Research quality

In order to guarantee the quality of the research, validity is addressed in terms of construct, internal and external validity. The measures this research has taken to ensure the validity, are described in table 3.3.

Validity and reliability Description (Yin, 2009) Measures taken for this research

Internal validity To ensure that the results are not caused by other variables which are not included in the research

Achieved by formulating search terms based on previous research that has used Twitter analysis for disasters. The search terms used for this research are: #hurricaneirma & help, hurricaneirma & rescue, #hurricaneharvey & help, #hurricaneharvey & rescue, #freetownfloods. #sierraleonemudslide, #Joplin & help and #Joplin & rescue.

Construct validity To ensure that the research measure what it claims to be measuring

Addressed by using multiple sources of evidence, this is done by collecting tweets of different disasters from different Twitter users.

External validity The degree to which the outcomes of this research are applicable to other settings

Applying theoretical replication to the selected cases. This researched addressed external validity to select disasters with different sizes.

Reliability To ensure that if the study is executed again, that the same results are derived

Steps of the data analysis are described in detail, exemplary tweets are shown to explain the mechanisms and Excel screenshots of the data is put into appendix B.

(20)

20

4. Findings

After the data analysis, it was found that information on social media contributes to supply chain responsiveness. The mechanisms that explain how information on social media contributes to supply chain are the size of the disaster, time stamp of the message, origin of the tweet, duration of the disaster, the expression of demand or supply and validation of information. An overview of the mechanisms are summarized in table 4.1 below. The remainder of this section explains the mechanisms based on the dimensions of the dependent variable (supply chain responsiveness) for each combination with the independent variable dimensions (information on social media).

Volume Product Process

Location - Specificity on the scope and the location differs based on size of disaster (case 1,2,3,4)

- Specificity about the location differs based on the size of the disaster (case 1,2,3,4)

- Specificity about the location differs based on expression of demand or supply (case 1,2)

- Specificity about local operations differs based on expression of

demand or supply (case 1,2,3,4)

Time - No contribution because of time stamp of message (case 1,2,3,4)

- Information about relief differs based on duration of the disaster (case 1,2, 4)

- Information about operations differs based on duration of the disaster (case 1,2, 4)

Content - Specificity on the scope differs based on origin of tweet (case 1,2,3,4)

- Specificity about the type of relief differs based on origin of tweet (case 1,2,4)

- Specificity about operations differs based on size of disaster (case 1,2,3,4)

Validity - No contribution because no confirmation or correction messages are sent (case 1,2,3,4)

- No contribution because no confirmation or correction messages are sent (case 1,2,3,4)

(21)

21 4.1 Volume supply chain responsiveness

The data indicates that information on social media about location that is linked to volume responsiveness differs in specificity based on the size of the disaster and the information of the message is based on the origin of the sender. It was found that tweets giving information about how much help is required or supplied at a location are less specific when the disasters are big. Unspecific information make it more difficult for relief organizations to determine the scope of relief required at a location. For case 1 and 2, the size of the disaster was bigger in comparison to case 3 and 4. Because the affected area was bigger, the tweets were less specific about the location and the scope of the disaster. The following tweet is an example of low specificity for case 1 and 2 because only an estimate number of victims or volunteers is mentioned: “Hundreds of personnel from the Bay Area in Florida to help those impacted by #HurricaneIrma.”1_{(Case 1). The}

information about location that contribute volume supply chain responsiveness to determine the scope of the required relief becomes more specific for case 3 and 4. These messages contain information about a specific number and location as shown in the following example: “#SierraLeoneMudSlide 235 children registered by partners CODWeIA @ Regent epicentre food, clothing, blankets, medicine urgently needed.” 2_{(Case 3). As such, it seems like the size of the}

disaster influences the specificity of the tweets.

The results of the data analysis showed that information on social media about time is not linked to volume supply chain responsiveness because of the timestamp of the messages. For all cases analysed, users do not mention information about time because this information is already provided by the time stamp of the tweet. As such, it seems that the timestamp of the message influences the number of messages sent by users giving information about when what number of help is required.

1_{NBC Bay Area (@nbcbayarea). 10 September 2017, 14:29. Retrieved from:} https://twitter.com/nbcbayarea/status/906992941272625152

(22)

22 Information on social media about content that is linked to volume supply chain responsiveness differs in specificity based on the origin of the message. It was found for all cases that messages sent by the government or a company contain information that gives more details about the scope of the relief in comparison to message sent by a private individual. The following tweet is an example of an organization that gives detailed information about the supplied relief: “We’ve already given out 500+ emergency food packages including clean water; today 1,500 more #SierraLeoneMudSlide.”3_{(Case 3). The next example shows that private individuals are less} specific in giving information that could be useful for organizations to determine the scope of the relief required: “I was driving down I-95 today and saw SEVERAL convoys of utility trucks (y’all the real MVPs). Help is on the way, Florida! #HurricaneIrma.”4_{(Case 1). As such, it seems like} the origin of the message influences the specificity of the tweets.

No data was found that describe the mechanism between information on social media about validity and volume supply chain responsiveness. None of the cases analysed contained messages where validity was established by confirming or disproving information about the scope of demand that helps organizations to respond to the situation. Based on this, it seems the senders of the messages are more engaged in sharing information about what is needed at a location instead of correcting or confirming information that is available on social media.

4.2 Product supply chain responsiveness

The results of the data analysis indicate that information on social media about location that is linked to product supply chain responsiveness differs in specificity based on the size of the disaster and the expression of demand or supply. For case 1 and 2, the size of the disaster was bigger in comparison to case 3 and 4. Because the affected area was bigger, the tweets were more specific about the location. The following tweet is an example where specific information about the relief and location is shared: “Rescue needed. 6 families in “port Arthur” near 3056 25th St. Port Arthur,

3_{Street Child US (@streetchildus). 16 August 2017, 10:14. Retrieved from:} https://twitter.com/streetchildus/status/897869080316739585

(23)

23 TX send boats please #PortArthur #HurricaneHarvey #rescue.”5_{(Case 2). For case 3 and 4, tweets} were less specific about the location of where relief was required, because the location was already known due to the smaller size of the affected area. The following tweet is an example where less specific information was shared about the location where relief was required: “#FreetownFloods harrowed workers continue search @ Regent 2day #mentalhealth support needed now #SieraLeoneMudSlide.”6_{(Case 3). Furthermore, the data indicates that for case 1 and 2 the} specificity also differs based on tweets expressing supply or demand. It was found that tweets consisting of information about the location indicating demand were more specific about where relief was required in comparison to tweets consisting of information about the location expressing the supply of products. The following tweet is an example where specific information is given about the location of the demand of relief: “Elderly parents (whaltons pet store) not heard from since storm hit 29851 Fischer Ln Big Pine Key #bigpinekey #hurricaneirma please help!!.”7_(Case 1). The following example shows a message giving information about the supply of relief. This tweet shares less specific information about the location: “Neighors use their personal boats to rescue Friendswood, Texas residents stranded by flooding Sunday, #Houston #HurricaneHarvey”8 (Case 2). Based on this, it seems like the tweets expressing demand share more specific information because specificity about the location and type of relief is more important for executing rescue operations. Information about the supply of relief is less relevant because no relief is required at this location anymore. Based on this, organizations can use messages that express demand to find specific information about what resources are required at a specific location.

The results of the data analysis indicate that information on social media about time that is linked to product supply chain responsiveness differs based on the duration of the disaster. The

5_{Phil Keil (@Phil_Keil). 20 August 2017, 6:29. Retrieved from:} https://twitter.com/Phil_Keil/status/902885981438824448

6_{S4CCC-SL (@scccsierraleone). 16 August 2017, 12:27. Retrieved from:} https://twitter.com/scccsierraleone/status/897902610249895937

7_{Cathryn Jones (@OneofMarysgirls). 10 September 2017, 11:30. Retrieved from:} https://twitter.com/OneofMarysgirls/status/906948051977359360

(24)

24 information that was shared about time for case 1, 2 and 4 was a description of moments when rescue and relief efforts could not be provided because the disaster did not end yet. An example of a tweet giving information about time is the following message: “2000 buildings destroyed in #Joplin tornado, 400 ppl still missing, search and rescue likely ending at 7pm due to weather risk.”9_{(Case 4). For case 3, the disaster ended before the response activities started. As such, the} relief efforts were not hindered. Because of this, no tweets analysed gave information about time.

The data indicates that information on social media about content that is linked to product supply chain responsiveness differs in specificity based on the origin of the sender. It was found for case 1, 2 and 4 that tweets that were sent from a private individual were more specific about what resources were required or supplied. As such, organizations should use information on social media sent from private individuals to determine what resources are required for response. The following tweet show a message where a private individual requests a specific product: “airboats needed in #rockport, texas for search and rescue #harvey #hurricaneharvey. “10_{(Case 2). Tweets that were} sent from companies or the government contained more general information about relief without mentioning specific products. An example of a tweet with less specific information is: “We’re pledging $1 million to help with relief efforts in #Joplin to ensure the essential needs of the community are being met.”11_{(Case 4). However, for case 3, tweets sent by companies were just as} specific as tweets that originated from private individuals. This is because companies involved during this disaster were charity organizations depended on those specific resources. An example of an organization that sent a message expressing the supply of a specific resource is the following tweet: “Thank you #UnitedArabEmirates for helpin @WFP provide life-saving food to families affected by #SierraLeoneMudSlide.”12_{(Case 3).}

9_{theNewsWorthy with Erica Mandy (@theNWpodcast). 24 May 2011, 6:54. Retrieved from:} https://twitter.com/theNWpodcast/status/73024070173732864

10_{Emma (@bymyelf). 26 August 2017, 8:57. Retrieved from:} https://twitter.com/bymyelf/status/901473688352174080

11_{Walmart Today (@WalmartToday). 23 May 2011, 16:45. Retrieved from:} https://twitter.com/WalmartToday/status/72810327477190657

(25)

25 The data analysed for all cases did not include tweets giving information on social media about validity that contributes to product supply chain responsiveness. For relief organizations, information on social media is only useful to gather information about what specific products or resources are needed, no information is available about users correcting or confirming other messages.

4.3 Process supply chain responsiveness

The data indicates that information on social media about location that is linked to process supply chain responsiveness differs in specificity based on the expression of demand or supply in the tweet. It was found for all cases that tweets giving information about the supply of relief to a location gave specific information about how relief was delivered. This can help relief organizations in setting up operations for response to this location because it is known how relief can be delivered to that location. The following tweet is an example of a message expressing specific information about the delivery of relief to a location: “And very delayed response. Day 2 after #SierraLeoneMudSlide burial team cars pass on Wilkinson Rd every 10 mins. None of that on Mon or Tues.”13_{(Case 3). The following example shows that tweets expressing demand} contains less specific information that is useful for organizations on how to provide relief to an affected area: “Flooding in #Lithia.. These people are going to need help #hurricaneirma.”14_(Case 1). Based on this, it seems that the expression of demand or supply influences the specificity of the tweets with regard to the location.

The results of the data analysis showed that information on social media about time that is linked to process supply chain responsiveness differs based on the duration of the disaster. For case 1, 2 and 4, information was shared about when rescue and relief efforts could not be provided because the disaster did not end when response phase started. An example of a message that gives information that is useful for organizations setting up their processes for response is the following tweet: “City of Miami Beach: Rescue teams no longer able to respond due to extreme sustained

13_{Anne Karing (@AnneKaring). 16 August 2017, 8:57. Retrieved from:} https://twitter.com/AnneKaring/status/897849755065364481

(26)

26 winds #HurricaneIrma #HurricaneIrma2017 #Florida.”15_{. (Case 1). For case 3, the disaster ended} before relief efforts began. Because of this, no tweets were collected with information about time that was relevant to relief organizations in setting up their processes for response.

The results of the data analysis indicate that information on social media about the content that is linked to process supply chain responsiveness differs in specificity based on the size of the disaster. The size of the disaster for case 1 and 2 was bigger in comparison to case 3 and 4. Based on this, the specificity of information between these cases differ as well. For case 1 and 2, the disaster affected a big area and damaged the main supply routes. These tweets were more specific and contained information about how these areas could be reached to provide relief efforts. An example of a message that gave specific information that helps organizations to set up response operations is the following tweet: “It’s so amazing to listen as the coast guard helicopter rescue people in my neighbourhood!! #godsgrace #hurricaneHarvey #houstonflood.”16_{(Case 2). For case 3 and 4, the} size of the affected area was smaller and no main supply routes were damaged. Because of this, the information on social media was less specific about how the relief was supplied. The following tweet gives an example of the information that is less specific in comparison to the previous example: “Clean up and rescue efforts hasten as crew members rush to help before storm hits. #Joplin #Tornado #SGF.”17_{(Case 4).}

Based on the analysed tweets, it was found that information on social media about validity does not contribute to process supply chain responsiveness. The users on Twitter did not sent messages where information was confirmed or disproved. Based on this, it seems like organizations can only use this information to determine how resources or help can reach the affected area. This information cannot be used to determine the validity of the information available on social media.

15_{Newspoint (@NewspointRadio). 10 September 2017, 8:16. Retrieved from:} https://twitter.com/NewspointRadio/status/906899173496291328

16_{Clarissa Fletcher (@ClarissaColors). 28 August 2017, 15:51. Retrieved from:} https://twitter.com/ClarissaColors/status/902302732576534528

(27)

27

5. Discussion

Because previous literature did not describe how information on social media can contribute to supply chain responsiveness, more in-depth knowledge was required. This research analysed social media to find mechanisms that explain the relationship between information on social media and supply chain responsiveness. The results found that the size of the disaster, timestamp of the message, origin of the tweet, the duration of the disaster, the expression of demand or supply and validation of information are mechanisms that explain how social media can help organizations to set up a supply chain for disaster response. This section links these findings to theory. For all mechanisms, the findings are going to be discussed. Based on this, a set of propositions is presented.

5.1 The size of the disaster

The results have shown that information available on social media about location contributes to volume- and product supply chain responsiveness and has higher specificity when the disaster is small because the area that is affected is smaller. Previous literature stated that information on social media helps to localize where the relief is required (Wang & Ye, 2018). However, previous literature did not mention that information on social media about the location also gives information about how much and what type of relief is required. Based on this, the results of this research add to previous theory that information on social media about the location can help organizations to determine the type and scope of relief required at the affected area. Additionally, the results add to theory that the information on social media about the location become more specific when the disasters are small. Based on this, the following proposition is proposed:

(28)

28 Previous research has found that information on social media about the content contributes to process supply chain responsiveness (Qu et al., 2016). This research adds to this theory that information on social media about content that contributes to process supply chain responsiveness, is more specific during big disasters. This is because during big disasters, more infrastructure could be damaged that is important for response operations. When little to no damage has been done to the infrastructure, less information will be shared about this on social media. For organizations, the information available on social media can help in setting up their supply chain for relief because information is shared how the disaster area can be reached. Therefore, the following proposition is suggested:

P2: The size of the disaster influences the specificity of information on social media about the content that contributes to process supply chain responsiveness.

5.2 The timestamp of the message

Previous studies found that information on social media about time only contributes to process supply chain responsiveness and not to volume supply chain responsiveness because information is given about the phase of the disaster to determine response activities (Imran et al., 2014). The findings of this research support the theory that information on social media about time does not contribute to volume supply chain responsiveness. This research adds to this theory that messages do not contain information about the time and scope of the disaster because the information about time is already provided by the timestamp of the message. From this, the following proposition is stated:

(29)

29 5.3 The origin of the tweet

The results have shown that the information on social media about content that contribute to volume- and product supply chain responsiveness differs in specificity based on the origin of the tweet. Organizations that use social media to find information about the scope of the disaster will attain more specific information from messages sent by the government or companies. When the aim of using information on social media is about acquiring knowledge of what relief is required, messages sent from private individuals will be more specific. Previous research has found that information on social media about the content helps activity for response (Wang & Ye, 2018), this is in line with the findings of this study. However, the findings of this research add to this knowledge that the origin of the tweet on social media influences the specificity of information that helps to set up response activities. Considering this, the following proposition is proposed:

P4: The origin of the tweet influence the specificity of information on social media about content that contributes to volume- and product supply chain responsiveness.

5.4 The duration of the disaster

Previous research has found that information on social media about time contributes to process supply chain responsiveness because this information helps organizations to understand the stage of the disaster to determine when to start the response activities (Imran et al., 2014). This theory is in line with the results of this research. It is important to note that this research has found that not only information on social media about time contributes to process supply chain responsiveness, but also information about time contributes to product supply chain responsiveness. Furthermore, an addition to previous research is that this study found that the duration of the disaster influences the information available about the response activities. When the disaster has not ended when the response phase has started, organizations can use information on social media to determine how and what resources can be delivered to the affected area. Therefore, the following proposition is stated:

(30)

30 5.5 The expression of demand or supply

Previous literature described that information on social media about the location contributes to process supply chain responsiveness because this information enables organizations to locate the demand for relief so that processes can be set in motion (Holguín-Veras et al., 2012). The results of this study does not support this theory. This research found that information on social media about the location expressing demand shares more specific information about what relief is required at a location and therefore contributes to product supply chain responsiveness. In contrast to previous literature, this research found that information on social media that expresses the supply of relief at a location provides more specific information that can help organizations to set up the processes for relief. Therefore, the following proposition is suggested:

P6: The expression of demand or supply influence the specificity of information on social media about location that contributes to product and process supply chain responsiveness.

5.6 Validation of information

Previous literature found that information about validity helps organizations to acquire reliable information about the size and type of demand by users confirming or contradicting other messages (Simon et al., 2015). The results of this study does not support this theory because from the analysed tweets, no message confirmed or disproved the information available on social media. This means that organizations cannot use information on social media to verify the validity of the information available. Considering this, the following proposition is proposed:

(31)

31

6. Conclusion

Sharing information after the occurrence of a disaster is essential for organizations to set up an effective disaster response. Communicating information may be problematic because of damage or overburdened communication lines, social media could help to acquire information for setting up a supply chain for disaster response. The main objective of this research was to investigate how information on social media can contribute to the supply chain responsiveness after a disaster. The research question: “How can information on social media contribute to supply chain responsiveness after a disaster?” can be answered by the six mechanisms that were identified that clearly explain how the variables are linked. It can be concluded that the size of the disaster, timestamp of the message, origin of the tweet, the duration of the disaster, the expression of demand or supply and validity of information are the mechanisms that explain how social media can help organizations to set up a supply chain to respond. To the best of the authors’ knowledge, this study is the first to present the mechanisms on how information on social media could aid the supply chain responsiveness. Therefore, this research adds valuable new insights to theory by proposing seven propositions that explain how information on social media can help organizations to gather information so that a supply chain can be set up to provide disaster relief.

6.1 Managerial implications

(32)

32 6.2 Limitations and suggestions for further research

(33)

33

References

Acimovic, J., & Goentzel, J. (2016). Models and metrics to assess humanitarian response capacity. Journal of Operations Management, 45, 11–29.

https://doi.org/10.1016/j.jom.2016.05.003

Alexander, D. E. (2014). Social Media in Disaster Risk Reduction and Crisis Management. Science and Engineering Ethics, 20(3), 717–733. https://doi.org/10.1007/s11948-013-9502-z Barclay, I., Poolton, J., & Dann, Z. (1996). Improving competitive responsiveness via the virtual

environment. IEMC 96 Proceedings. International Conference on Engineering and Technology Management. Managing Virtual Enterprises: A Convergence of

Communications, Computing, and Energy Technologies, 52–62. https://doi.org/10.1109/IEMC.1996.547789

Belvedere, V., Grando, A., & Papadimitriou, T. (2010). The responsiveness of Italian small-to-medium sized plants: Dimensions and determinants. International Journal of Production Research, 48(21), 6481–6498. https://doi.org/10.1080/00207540903234751

Blake, E. S., & Zelinsky, D. A. (2018). National Hurricane Center Tropical Cyclone Report: Hurricane Harvey.

Chakravarty, A. K. (2011). A contingent plan for disaster response. International Journal of Production Economics, 134(1), 3–15. https://doi.org/10.1016/j.ijpe.2011.01.017

Chakravarty, A. K. (2014). Humanitarian relief chain: Rapid response under uncertainty. International Journal of Production Economics, 151, 146–157.

https://doi.org/10.1016/j.ijpe.2013.10.007

Crooks, A., Croitoru, A., Stefanidis, A., & Radzikowski, J. (2013). #Earthquake: Twitter as a Distributed Sensor System. Transactions in GIS, 17(1), 124–147.

https://doi.org/10.1111/j.1467-9671.2012.01359.x

Danese, P., Romano, P., & Formentini, M. (2013). The impact of supply chain integration on responsiveness : The moderating effect of using an international supplier network. TRANSPORTATION RESEARCH PART E, 49(1), 125–140.

https://doi.org/10.1016/j.tre.2012.08.002

(34)

34 de Albuquerque, J. P., Herfort, B., Brenning, A., & Zipf, A. (2015). A geographic approach for

combining social media and authoritative data towards identifying useful information for disaster management. International Journal of Geographical Information Science, 29(4), 667–689. https://doi.org/10.1080/13658816.2014.996567

Halverson, J. B. (2018). The Costliest Hurricane Season in U.S. History. Weatherwise, 71(2), 20– 27. https://doi.org/10.1080/00431672.2018.1416862

Holguín-Veras, J., Jaller, M., Van Wassenhove, L. N., Pérez, N., & Wachtendorf, T. (2012). On the unique features of post-disaster humanitarian logistics. Journal of Operations

Management, 30(7–8), 494–506. https://doi.org/10.1016/j.jom.2012.08.003

Holweg, M. (2005). The three dimensions of responsiveness. International Journal of Operations & Production Management, 25(7), 603–622. https://doi.org/10.1108/01443570510605063 Huang, K., Jiang, Y., Yuan, Y., & Zhao, L. (2015). Modeling multiple humanitarian objectives in

emergency response to large-scale disasters. Transportation Research Part E: Logistics and Transportation Review, 75, 1–17. https://doi.org/10.1016/j.tre.2014.11.007

Imran, M., Castillo, C., Diaz, F., & Vieweg, S. (2014). Processing Social Media Messages in Mass Emergency: A Survey (Vol. 47). https://doi.org/10.1145/2771588

Jahre, M. (2017). Humanitarian supply chain strategies – a review of how actors mitigate supply chain risks. Journal of Humanitarian Logistics and Supply Chain Management, 7(2), 82– 101.

Kietzmann, J. H., Hermkens, K., McCarthy, I. P., & Silvestre, B. S. (2011). Social media? Get serious! Understanding the functional building blocks of social media. Business Horizons, 54(3), 241–251. https://doi.org/10.1016/j.bushor.2011.01.005

Kim, J., & Hastak, M. (2018). Social network analysis: Characteristics of online social networks after a disaster. International Journal of Information Management, 38(1), 86–96.

https://doi.org/10.1016/j.ijinfomgt.2017.08.003

Koubi, V., Stoll, S., & Spilker, G. (2016). Perceptions of environmental change and migration decisions. Climatic Change, 138(3–4), 439–451. https://doi.org/10.1007/s10584-016-1767-1 Kuligowski, E. D., Lombardo, F. T., Phan, L. T., Levitan, M. L., & Jorgensen, D. P. (2013).

Technical investigation of the May 22, 2011 tornado in Joplin, Missouri. National Institute of Standards and Technology. https://doi.org/10.6028/NIST.NCSTAR.3

(35)

35 support framework. International Journal of Production Economics, 145(2), 613–629. https://doi.org/10.1016/j.ijpe.2013.05.016

Moradlou, H., Backhouse, C., & Ranganathan, R. (2017). Responsiveness, the primary reason behind re-shoring manufacturing activities to the UK An Indian industry perspective. International Journal of Physical Distribution & Logistics Management, 47(2/3), 222–236. Murthy, D., & Gross, A. J. (2017). Social media processes in disasters: Implications of emergent

technology use. Social Science Research, 63, 356–370. https://doi.org/10.1016/j.ssresearch.2016.09.015

Ngwenya, N. K., & Naude, M. J. A. (2016). Supply chain management best practices : a case of humanitarian aid in southern Africa : original research. Journal of Transport and Supply Chain Management, 10(1), 1–9.

Qu, Y., Huang, C., & Zhang, P. (2016). Microblogging after a Major Disaster in China : A Case Study of the 2010 Yushu Earthquake, (January 2011).

https://doi.org/10.1145/1958824.1958830

Schryen, G., Rauchecker, G., & Comes, M. (2015). Resource Planning in Disaster Response – Decision Support Models and Methodologies. Business and Information Systems

Engineering, 57(4), 243–259.

Sierra Leone: rapid damage and loss assessment of August 14th, 2017 landslides and floods. (2017).

Simon, T., Goldberg, A., & Adini, B. (2015). Socializing in emergencies - A review of the use of social media in emergency situations. International Journal of Information Management, 35(5), 609–619. https://doi.org/10.1016/j.ijinfomgt.2015.07.001

Sklaver, B. (2003). Humanitarian Daily Rations: Evaluation and Guidelines Need. Disasters, 27(3), 259–271.

Steiger, E., de Albuquerque, J. P., & Zipf, A. (2015). An Advanced Systematic Literature Review on Spatiotemporal Analyses of Twitter Data. Transactions in GIS, 19(6), 809–834.

https://doi.org/10.1111/tgis.12132

Upton, L., Kirsch, T. D., Harvey, M., & Hanfling, D. (2017). Health Care Coalitions as Response Organizations: Houston after Hurricane Harvey. Disaster Medicine and Public Health Preparedness, 11(6), 637–639. https://doi.org/10.1017/dmp.2017.141

(36)

36 exploring a new supply strategy. International Journal of Operations & Production

Management, 27(12), 1362–1379. https://doi.org/http://dx.doi.org/10.1108/JEIM-07-2014-0077

Wang, Z., & Ye, X. (2018). Social media analytics for natural disaster management. International Journal of Geographical Information Science, 32(1), 49–72. https://doi.org/10.1080/13658816.2017.1367003

Yin. (1981). The Case Study as Serious Research Strategy. Knowledge, 3(1), 97–114.

Yin, J., Karimi, S., Lampert, A., Cameron, M., Robinson, B., & Power, R. (2015). Using social media to enhance emergency situation awareness: IJCAI International Joint Conference on Artificial Intelligence, 2015–Janua, 4234–4239. https://doi.org/10.1109/MIS.2012.6 Yin, R. K. (2009). Case Study Research (4th ed.). Thousand Oaks: Sage Publications Inc. Yoo, E., Rand, W., Eftekhar, M., & Rabinovich, E. (2016). Evaluating information diffusion

(37)

37

Appendices

Appendix A: Code tree

Information on

social media Example:

location

"Help still needed at 9100 Fondren Rd, #HTX. #HoustonFlood #HoustonFlooding

#HurricaneHarvey#Harvey2017 #HarveyStorm #Harvey"(case 2)

time

"BREAKING- Rescue efforts in #Joplin will end at 7 p.m. due to the risk of severe weather. Rising death toll now at 117." (case 4)

content

"Sierra Leone appeals for urgent help after deadly floods #SierraLeone #SierraLeoneMudSlide #flood #disaster http://thepeninsulaqatar.com/article/15/08/2017/Sierra-

Leone-appeals-for-urgent-help-after-deadly-floods …"(Case 3) validity

"FAKE NEWS I75 FL GA does not look like this. Be safe & help each other. Don't spread FAKE NEWS panic

#HurricaneIrma #IrmaShelter" (Case 1)

Supply chain

responsiveness Example:

volume

"#SierraLeoneMudSlide: WFP is distributing food to over 7,500 people affected by the devastation in Sierra Leone … https://twitter.com/i/web/status/897837325824208897 …"(C ase 3)

product

"#SanAntonio ice companies can you help? We need ice! #PadreIsland no electric or water. We survived

#HurricaneHarvey only request we have🙏 "(Case 2) process

(38)

38

Descriptive code 1 Example:

supply

"We've dispatched crews to help restore power in areas of the southeast U.S. damaged by #HurricaneIrma.

💪http://ow.ly/68do30f35hF" (Case 1) demand

"BOAT RESCUE NEEDED in Port Arthur #Texas #rescue

#helpneeded #harvey #HurricaneHarvey #portarthur Phone: 409-344-3319" (Case 2)

tip

"Want to volunteer to help those in Joplin in the future? Call 211 from your landline or 800-427-4626 from your cell #Joplin" (Case 4)

Descriptive code 2 Example:

private

"300 graves prepared to bury #SierraLeoneMudSlides victims. #RIPP #FreetownFloods#Environmentaldisaster #women #Health #Children" (Case 3)

government

"TF3 has set up base ops @FordPark in Beaumont, TX. Have started rescue ops for #HurricaneHarvey@HillsFireRescue @StPeteFR" (Case 2)

company

"Florida, we're headed your way next with 33,000 lbs of product to help fuel clean-up from #HurricaneIrma. Thanks @ConvoyofHope 💪" (Case 1)

descriptive code 3 Example:

product

"Twitter friends I need help. Where can I get gas in Miami? #posthurricaneirma #gasstations #miami#hurricaneirma #postirma" (Case 1)

shelter

"#Patriots RT @OzarksRedCross As shelters are opened, locations can be found here.http://rdcrss.org/9JDVYO #Joplin #tornado #twisters #Help" (Case 4)

rescue

"Police appear strongly focused on getting help to injured/trapped. Debris, lack of manpower current prob #Joplin #tornado" (Case 3)

volunteer

(39)

(40)

(41)

(42)

Information on social media as a contribution to disaster supply chain responsiveness