EDMG611Wk5

An Investigation of Interaction Patterns in Emergency Management: A Case Study of The Crash of Continental Flight 3407

Rohit Valecha1

# Springer Science+Business Media, LLC, part of Springer Nature 2019

Abstract The nature of emergency response operations varies based on agency, county, population, leadership, etc., which leads to serious challenges to interactions between emergency responders during routine emergencies. There have been very few studies that have provided direct insights into the emergency interactions from an information processing view based on real experience from the field. With the help of a case study, this paper aims to explore inter-agency interactions by addressing two research questions: (a) how do emergency agencies interact? (b) what characterizes interactions in the different stages of an emergency? This paper develops a framework for identifying emergency interactions that illustrate the templatic and dynamic styles as well as the active and passive patterns of interactions by examining data from the emergency reports. To summarize, this paper contributes to research in areas of inter-agency cooperation and emergency management.

Keywords Emergency interactions . Situation awareness . Emergency response

1 Introduction

Continental Flight 3407, from Newark Liberty International Airport in New Jersey to Western New York Niagara International Airport in New York State departed late from Newark on February 12, 2009, at 9:20 p.m. EST. On the landing approach to the airport, the plane stalled about 9.3 km, short of the runway and crashed into a house in the northeast Western New York suburb of Clarence Center at 10:17 p.m. The conditions were freezing. Fifty lives were lost – this included two pilots, two flight attendants, 45 passengers (including one off-duty pilot), and one person in the house into which the plane crashed.

It is within this environment that the governmental agencies such as local, state and federal have to cooperate in order to respond to the emergency situation. The different agencies are required to work together to effectively and efficiently respond to an emergency incident. However, due to differences in gov- ernance structure, training, applicable polices, command and control structures, legal requirements and culture, the nature of operations vary based on agency, municipality or township, county, population, leadership, etc. (Valecha et al. 2010,

2013). This leads to serious challenges to cooperation during multi-agency response to emergencies (Cigler 1988).

Inter-agency emergency cooperation is defined as interac- tion between two or more parties working together towards a common goal or purpose (Malone and Crowston 1994). Inter- agency cooperation has attracted a great deal of attention in the context of emergency response (Chen et al. 2005). McGuire and Silvia (2010) describe inter-agency cooperation as vital both before and after an emergency. Waugh Jr and Streib (2006) argue that effective response is unlikely to hap- pen without cooperation. Despite the growing interest in inter- agency cooperation, there are several issues that require more research (Janssen et al. 2010).

First, cooperation is based on interactions which have re- ceived little attention within this field (Janssen et al. 2010). Second, traditional models of cooperation have been designed based on probable scenarios, which are inadequate for volatile and dynamic situations (Faraj and Xiao 2006). Third, there has been limited work that utilizes information processing view and provides direct insights into the emergency interac- tions. In order to address these issues, it is important to devel- op a framework that identifies emergency interactions under varying situations, and that is concrete enough to be useful while also sufficiently abstract to be reusable.

This paper investigates techniques used by teams to describe interactions i.e. how responders cooperate in different situations. In particular, this paper performs a situational analysis based on

* Rohit Valecha [email protected]

1 Department of Information Systems and Cyber Security, The University of Texas at San Antonio, San Antonio, TX, USA

https://doi.org/10.1007/s10796-019-09896-z Information Systems Frontiers (2020) 22:897–909

Published online: anuary 2019J30

real-world data to explore emergency interactions from an in- formation processing view that integrates information and com- munication dimensions. This paper addresses two research questions: (a) in what way (i.e. styles) do emergency agencies interact? (b) what interactions (i.e. patterns) characterize coop- eration in the different emergency situations?

The paper contributes to research in the areas of emergency management by discussing the role of information, communica- tion, systems and related human efforts (Valecha 2015). Using the Design Science research methodology (Hevner et al. 2008; Peffers et al. 2007), this paper develops a framework for charac- terizing emergency interactions. By utilizing interviews with four experts (two fire chiefs and two dispatchers) and raw incident communication reports, this framework (a) illustrates the templatic and dynamic styles of interactions in a complex emer- gency incident using a real life case study, and (b) explicates the active and passive patterns of interactive messages (following Rao et al. 1995) over emergency timeline by examining data from the emergency reports.

From a methodological perspective, this work draws inspi- ration from the work of Benbasat et al. (1987). They argue that a case study examines a phenomenon in its natural setting, employing multiple methods of data collection to gather in- formation from one or more people, groups, or organizations. Further, they suggest that single cases are useful in specific instances where (1) situation is inaccessible to scientific inves- tigation and (2) situation is extreme or unique case. The plane crash was the first of its kind of emergency in Western New York. So, the situation is both inaccessible and unique.

This research is important because it recognizes require- ments for effective emergency cooperation. It is also needed to generate recommendations and principles for emergency system design using a case study. Thus, it can guide the de- velopment of interactive systems that are utilized in complex real life emergencies. The paper is organized as follows. In the next section, we provide a background of emergency interac- tions. Then, we highlight the research methodology. Subsequently, we illustrate the case study of Western New York Plane Crash followed by an analysis of interaction using Situation Awareness framework. Finally, we conclude with limitations and future work for this paper.

2 Literature Review

In this section, we elaborate on the literature in the area of emergency interactions. Then we introduce the theoretical lens of situation awareness.

2.1 Emergency Cooperation

Woratschek and Roth (2005) use the term cooperation as a form of inter-organizational interaction rooted in common

intention, and characterized by common objective. Other terms generally used to describe cooperation include, but are not limited to, partnership, collaboration and coordination (Rautenstrauch et al. 2003; Schulz 2009). In this paper, we consider cooperation as synonymous to collaboration and co- ordination, and define it as interaction between two or more parties working together towards a common goal or purpose.

The area of cooperation has been extensively studied in fields of systems science and computer science. One stream of litera- ture has focused on defining cooperation (Rautenstrauch et al. 2003), determining factors that affect cooperation (Thomson et al., 2009), identifying factors that analyze cooperative pro- cesses (Kuenzel and Welscher 2009). Besides that, another stream of literature is focused in identifying issues with inter- agency cooperation (Samba 2010), and exploring factors that lead to diminishing cooperative ability (Curra et al. 2009).

The expanding literature in emergency response has also focused on cooperation as a facilitator of efficient response. Jarvenpaa and Ives (1994) and Hollingshead et al. (1993) contribute to a better understanding of responder groups’ co- operation in improving emergency response. Majchrzak et al. (2007) make the case for better understanding cooperation among responder groups during an emergency situation. Turoff (2002) discuss systems such as PREMIS that facilitate effective cooperation during emergencies. Diniz et al. (2005) argue that the current emergency response systems should provide cooperative knowledge systems for exchanging pro- fessional information between responders.

2.2 Emergency Interactions

Although increasing research in emergency cooperation has been conducted recently, there are few issues that need to be addressed: First, emergency interactions have received little attention within the emergency context (Janssen et al. 2010). The importance of emergency response systems lies in facili- tating interactions rather than cooperative processes. This view might prevent situations in which the sharing of infor- mation fails due to lack of clarity about information needs and capabilities (Valecha et al. 2012a, b).

Second, the majority of information systems for achieving emergency interactions have been designed based on probable scenarios, which are inadequate for volatile and dynamic sit- uations (Faraj and Xiao 2006). The interaction mechanisms may vary with changes in the emergency situation. Therefore, there is a need to develop flexible interaction mechanisms that can be easily customized for the specific situation and provide better supports for improvised responses (Chen et al. 2008; Mendonça 2007). Such mechanism will ensure that necessary information is shared.

Third, the research in the area of emergency cooperation is still confined to institutional silos (Elliott 2006), which lack insights into the emergency cooperative processes. While

Inf Syst Front (2020) 22:897–909898

there have been a number of studies that focus on facilitation of cooperation during emergencies, there have been fewer studies that deal with providing direct insights into the emer- gency interactions based on real experience from the field (Chen et al. 2005). To address these issues, this paper per- forms a situational analysis based on real-world data to iden- tify interaction patterns from an information processing view that integrates information and communication dimensions.

2.3 Emergency Interaction Patterns

Patterns are a way of describing situations that are encoun- tered repeatedly. They define stable solutions to recurring sit- uations. They are concrete enough to be useful, while also sufficiently abstract to be reusable (De Moor 2009). Interaction patterns are repeatable techniques used by teams to help them cooperate. They describe, in a broad way, reus- able interactions (or lessons learnt). They make emergency situations actionable by describing how responders could co- operate in a specific situation. Thus, the application of inter- action patterns helps to improve the process of information sharing during emergency situations.

Interaction patterns can be created by assessing how ac- tions are performed in various situations and how mutual un- derstanding and commitments within situations evolve into cooperation strategies. They form reusable programmable scripts that can be executed when responders engage in certain cooperative actions within various situations. This paper pro- vides foundations for the design of an emergency platform that is based on interaction patterns to effectively respond to unpredictable and complex situations.

2.4 Need for Situation Awareness

In order to conduct an investigation of emergency interac- tions, it is important to define temporal boundaries that can characterize the different stages of emergency response. There are some existing frameworks, such as emergency manage- ment cycle, which define the different phases (response, mit- igation, recovery and preparedness) of emergency manage- ment. But such frameworks are more general in nature and do not necessarily take into account the information process- ing that is inherent within the interactions – the integral part of emergency cooperation process.

On the other hand, situation awareness framework (Endsley 1995) has been utilized to characterize emergency stages in accordance with the nature of the information available. SA

framework is ideal for defining the information boundaries that inevitably exist in an emergency event (Bachura et al. 2017a, b; Valecha et al. 2016). This enables the establishment of temporal boundaries that can be used to define the stages of emergency interactions. So prior to investigating emergency interactions, it is necessary to leverage SA framework to categorize the emer- gency event temporally.

2.5 Situation Awareness

During an emergency, the Incident Commander plays a vital role in coordinating the efforts of responders from various on- scene and off-scene agencies and organizations working in constantly changing environment. In this paper, we use the Situation Awareness (SA) framework to characterize the dif- ferent emergency phases as well as analyze the requirements of the situation for the development of valuable artifacts (Endsley 1995). SA is an important concept developed in the military domain that provides an understanding of the environment as a basis for efficient decision-making.

SA is a descriptive framework that characterizes informa- tion processing into three stages, namely perception, compre- hension and projection. In the first stage, the responder strives to perceive relevant elements from the dynamic environment at the crash site. In the second stage, the perceived information is comprehended into meaningful understanding about the current state of resources. Finally, in the third stage, the un- derstanding of the environment is projected into interaction patterns. Since the incident commander cooperates between agencies with the help of emergency messages, we consider projection of these messages for future actions such as time of response, etc. This is depicted in Fig. 1. For the purposes of this paper, we deal with only the Situation Awareness block.

3 Methodology

This research is based on qualitative design in the form of interviews with responders from various agencies, derivations from previous literature on emergency interactions and inci- dent reports generated during the February 2009Western New York plane crash. This research also draws from other emer- gency case study articles such as (Samba 2010; Curra et al. 2009) to develop support for better interaction mechanisms. This case study provides details that are very unique to this incident, however, it also helps to serve as a strong base in identifying weaknesses of multi-agency cooperative response.

Fig. 1 Situation Awareness Framework (Strater et al. 2001)

Inf Syst Front (2020) 22:897–909 899

3.1 The Case of Western New York Plane Crash1

The Western New York plane crash that happened on Feb 12, 2009 was a first of its type in Western New York. The plane crashed into a house in the suburbs of Western New York just a few miles short of the airport en route from Newark, New Jersey. Significant ice buildup on the wings and windshield of aircraft as it descended through light snow and mist, were deemed as major contributing factors that led to the crash. In addition to the passengers on board, the plane was loaded with 5800 lbs. of fuel. Everyone aboard the plane – 44 passengers, 4 member crew, and an off-duty airline employee lost their lives. In addition the person living in the house where the plane crashed also did not survive the crash.

About 12 nearby houses were evacuated after the crash and a limited state of emergency was declared. The incident lasted for about 48 h. There were 41 responding units that reported on-scene during the course of the incident. More than 200 employees were involved in mitigating the intensity of the incident. Based on Federal Emergency Management Agency’s (FEMA), Incident Command Structure (ICS), the response to the incident was divided into five roles: Command, Operations, Planning, Logistics and Finance.

The Command was responsible for all activities including developing and implementing the strategic plan, attention to organizing and managing the scene, cooperation for setting pri- orities for work accomplishment, coordination with other public officials and agencies, and other executive. The Operations was responsible for carrying out the directions of the command in- cluding managing all operations of implementing strategic plan on the scene, and maintaining discipline and accountability of materials, resources and responders.

The Planning was responsible for collecting and dissemi- nating information including gathering and analyzing situa- tional data along with providing appropriate displays for situ- ational status. The Logistics was responsible for coordinating the information technology and information system needs on the scene, medical care for the incident as well as communi- cation between departments. The Finance dealt with account keeping at various departmental levels.

The emergency support functions combined the capabili- ties of various agencies. Table 1 highlights the responsibilities of various agencies throughout the incident. During the plane crash incident, the response was segmented into two phases: First, the operations phase beginning with the instant of the crash, continuing till the fire was completely put out. Second, the recovery phase starting immediately after the fire was out. The response i.e. the first phase lasted for about 20 h, and consisted of events between the incident commander, fire agencies and the dispatch (that acted as mediators). The

recovery i.e. the second phase lasted for over 28 h, and consisted of events between the onsite agencies, not including dispatch.

For this paper, we deal with emergency interaction in the first phase. In addition, we restrict our analysis to the Bgolden hour^ of the incident. While incidents may last for more than one hour, Curra et al. (2009) and Samba (2010) suggest that the response in the first 60 min of the incident is the most critical. Besides that, Valecha et al. (2013) also focus on the golden hour of the emergency response in their analysis of the emergency messages. Rao et al. (1995) also focus on mes- sages while discussing about team interactions.

3.2 Western New York Plane Crash Data

Figure 2 shows an excerpt from crash incident report. This report is a log of messages that are exchanged between the responders and agencies, and is extremely useful in strategic planning and incident management. The data for the crash was mainly obtained from the BVehicle Summary^ and the BDispatch Comments^ section of the report. The former pro- vides details on the responding agency, and the latter provides messages as exchanged between the dispatch and incident commander. We interpret the vehicle codes with the help of setup values, and the messages with the help of responder interviews. The messages were classified into two main cate- gories based on their objectives, namely notification (that in- form about the incident) and update (that update information about the incident). The message classification is not detailed here, since it is not the focus of this paper. From here on, we refer to agencies, resources and responders (ARR) as BARR units^ in order to maintain standard with the incident reports.

From the report, it was observed that the Incident Commander was managing the interactions between 19 re- sponders, from 13 agencies, utilizing 22 different resource types after the first hour of incident. Table 2 provides terminology for ARR units as identified from the incident report. Tables 3 and 4 depicts the additional interacting ARR units along with its fre- quency, which were reporting every 15 min to the scene of the incident, with the help of terminology. The alphabetical charac- ters refer to agencies, while the numeric characters refer to re- sources and responders. For example, CC2 is the pumper from Clarence Center agency, CC9 means Clarence Center’s fire chief, while SW6 refers to ladder resource from Swormville agency. Figure 3 depicts the interacting ARR units as cumula- tive counts at the end of each time period.

4 Situation Awareness for Analyzing Interactions

Situation Awareness framework provides essential insights for analyzing Incident Commander’s decision-making process

1 Preliminary details of the case were presented at the American Conference on Information Systems (AMCIS) 2012 in Seattle, Washington.

Inf Syst Front (2020) 22:897–909900

during the Western New York Plane Crash incident. The pro- cess for the analysis is explained in this section as follows: First, we provide an illustration of emergency interactions. Second, we discuss the styles of interactions that were identi- fied in the time-series analysis. Finally, we identify the pat- terns of emergency interactions.

4.1 Illustration of Emergency Interactions

The illustration of on-scene interaction is shown in Fig. 4a, b. The first figure depicts the pattern of interacting units, while the second figure depicts the pattern of interacting messages. The arrival times of the interacting ARR units were used in a time series analysis, along with frequency of interacting mes- sages. The horizontal axis depicts the timeline with an interval of 15 min. The vertical axis depicts the frequency, in percent- age of total count. The different colored graphs indicate the different interacting units (ARR units) and interacting mes- sages (notifications and updates).

In the first 15 min of the crash, the incident commander perceives the significant intensity of crash/damage and calls for additional help that is observed in higher levels of notifica- tions that leads to higher number of ARR units reporting on- scene. Since the plane was carrying 49 passengers aboard, the incident command perceives support of additional rescue trucks for victims that may be alive. This perception of crash intensity equates to level one (perception) of situation awareness.

In the 15–60 min time period, the incident commander com- prehends the availability of ARR units at hand. In case of scar- city, the incident commander requests additional resources that is observed in higher levels of units arriving at scene with lower levels of messages. Before getting into suppression of fire, the incident commander comprehends the chances of injury, and thus requests additional ambulances to the scene. This compre- hension of availability of supporting units equates to level two (comprehension) of situation awareness.

Finally, after the 60 min of crash, the incident commander projects his/her learning of the incident to further decision- making actions that is observed in higher levels of messages

Fig. 2 Western New York Plane Crash Incident Report Excerpt

Table 1 Response Agencies for the Western New York Plane Crash

Agency Types Responsibility

Response Teams EMS, Fire, Police, Dispatch, Incident Commander Provide support for critical tasks: EMS – updates to local hospital, support medical activities Fire – respond to fire prevention and control Police – support scene security and perimeter safety Dispatch – mediate operations between chiefs and agencies

Health Operations Medical Examiners, Erie County Health Operation Center, Critical Incident Stress Debriefing Unit, Twin City Ambulance Unit

Provide on-site support to EMS Units Provide physicians for rehab centers Provide on-site treatment and evaluation of responders Minimize stress related injury through education and intervention

Special Team Incident Management Team, Hazmat Team, SMART, FBI Response Team

Setup Incident Command Structure Providing action against hazardous material Provide Recovery control and investigation

Media Relation Public Information Officer, Western NY 211 Organize daily press briefings, communicate with local media Handle calls related to victims, crash zones, etc.

Other Facilities Cheektowaga Senior Center, Clarence Library, Clarence Center Fire Auxiliary

Provide in-house support for responders including lunches, rescue operations, shelter areas, etc.

Inf Syst Front (2020) 22:897–909 901

and higher levels of units reporting to the scene. The incident commander projects the support of pumpers for suppressing fire and thus makes requests accordingly. This projection of actions for efficient decision-making equates to level three (projection) of situation awareness. These results are summa- rized in Table 5:

4.2 Styles of Emergency Interactions

In general, the frequency of interacting ARR units depicts an alternating high and low periods (see Fig. 5). This implies that there was an active run of high number of interacting units reporting in the first period, followed by a dry run of very few interacting units reporting in the next period, continuing up to a total of five periods.

In considering the response to the plane crash emergency, the interaction in the first 15 min relied on the preset template determined the course of action, and the interaction in the subsequent time periods was dynamic in nature i.e. the effect of existing action stimulated the performance of next action. We explicate these two types of interactions (see Table 6), namely templatic and dynamic interactions, next.

4.2.1 Templatic Interaction

In the early stage of the response, the incident commander perceives the severity of the crash, and determines the extent of response required to curb the emergency. Based on this analysis, the incident commander raises alarms of various levels. These alarms are specified as a part of a mutual aid agreement2 that identifies the quantity and type of resources and responders that should be arriving at the scene as a result. The frequency of the interacting ARR units arriving at the scene depicts a decreasing trend. There are no additional calls for help or requests for resources in the early stage. This im- plies that the resources and responders specified in the mutual aid agreements are sufficient to manage the operations of the emergency. We characterize this type of interaction during the

early stage of the response that presets the resources’ and responders’ involvement in the emergency as templatic inter- action, since most of the communication and resource alloca- tion falls into a template. This translates into the following proposition:

Proposition 1: Templatic interactions will be dominant (greater) in the perception stage.

4.2.2 Dynamic Interaction

In the later stages of the response to plane crash, the incident commander perceives the severity of the damage and raises additional alarms. Similar to early stages, the alarms specify the quantity and type of resources and responders that should be arriving at the scene. Due to the spread, magnitude, expan- sion of the fire, the cooperative units are not sufficient to control the emergency. As a result, responders update the sta- tus of the emergency.

The incident commander uses the information traces from the environment (perceive) to request additional resources (project). If the fire still does not come under control, the incident commander initiates an additional cycle of Bperceive-project^. There is an alternating pattern that denotes an active run of interacting ARR units reporting in the first period, followed by a dry run of very few interacting units reporting in the next period, and so on. This type of interaction in which the resources and responders arrive at the scene in a repetitive perceive-project cycle wherein the effects of an ac- tion stimulate the next action is termed as dynamic interaction. The templatic and dynamic interactions in the various stages of the plane crash emergency are summarized in Table 6. This translates into the following proposition:

Proposition 2: Dynamic interactions will be dominant (greater) in the comprehension and projection stages.

4.3 Patterns of Emergency Interactions

Figure 6 depicts the pattern of interacting messages. The dif- ferent colored graphs indicate the messages (notifications and

2 Mutual aid agreement is a request template among responding agencies to lend assistance across jurisdictional boundaries.

Table 2 ARR Units Terminology

Agencies Resource Responders

CC: Clarence Center N: Newstead 1–4: Pumper 9: Chief

C: Clarence MT: Main Transit 5: Heavy Rescue 91: Chief first assist

EA: East Amherst NA: North Amherst 6: Ladder 92: Chief second assist

SW: Swormville A: Akron 7: Light Rescue 93: Chief third assist

R: Rapids MIL: Milgrove 8: Ambulance 94: Chief fourth assist

HH: Harris Hill BOW: Bowmansville The alphabetical characters refer to agency. The alphabetical characters refer to agency.

G: Getzville Example: SW8 – Ambulance from Swormville agency Example: R9 – Chief of Rapids agency

Inf Syst Front (2020) 22:897–909902

updates). The frequency of messages shared between on- scene and off-scene agencies followed a U-curve. This implies that the number of messages was relatively higher in the initial and final periods and lower in the middle periods.

In response to the plane crash emergency, interaction in the first 15 min relied on notifications i.e. active messages that request resources to the scene, and interaction in the subse- quent time periods was dominated by update messages that provide passive status information about the resources. We explicate these two processes of interaction, namely active and passive interaction, next.

4.3.1 Active and Passive Patterns

In the first phase, there are numerous reports of the plane crash from the neighborhood where the crash took place. Since the responders are just getting started with the investigation process, there are higher number of notifications about the crash, and lower number of updates about the crash. In the second phase, the incident commander comprehends the availability of units at hand. This results in lower number of notifications and updates. In the third phase, the incident commander seeks status of all resources, responders and agencies at the scene and off-site, and projects the learning about the incident to request further decision-making actions. This results in higher notifications and updates. The active and passive interactions in the various stages of the plane crash emergency are summarized in Table 7. This translates into the following propositions 3a and 3b:

Proposition 3a: Notifications will be dominant (greater) in the perception and projection stages.

Proposition 3b: Updates will be dominant (greater) in the projection stage.

The research propositions 1, 2 and 3 were derived taking into consideration the stages of Situation Awareness as they translate to system design. In particular, the research proposi- tions investigate interaction patterns that are dominant within the different situation awareness stages of perception, compre- hension and projection.

5 Towards A Framework of Emergency Interactions

In order to make the above discussion useful for effective inter- action, we interviewed four responders who were asked ques- tions related to interactions including dispatch, fire and police. The responders interviewed were experts from dispatch and fire agencies (two from dispatch agency and two from fire agency). The incident commander was not interviewed because of avail- ability of his time. However, the other interviewed responders were a part of incident commander’s team and thus could pro- vide valuable insights. These emergency responders were an active part of the operations. The interviews took place in mul- tiple rounds with each round lasting for about 90 min. Their responses helped identify various dimensions of interactions during the various phases of the incident. The results of respond- er interviews have helped in identification of key dimensions that aid in the design of interactive emergency systems.

5.1 Information Dimension

Lack of information has always limited the efficiency of the response. Comfort et al. (2004) show that the access to the information also plays a vital role in improving efficiency. This leads to an important argument on how to manage the information (Rossel et al. 2016). The interviewed responders provided responses for identification of who is in charge, what

Table 4 Frequency of ARR units interacting at the scene <15 min 15–30 min 30–45 min 45–60 min >60 min

Agencies 4 1 5 No additional 3

Resources 4 1 9 No additional 8

Responders 8 1 3 No additional 6

Table 3 ARR Units Reporting at the Scene

<15 min 15–30 min 30–45 min 45–60 min >60 min

Agencies CC, C, EA, SW R HH, G, N, MT, NA No additional A, MIL, BOW

Resources EA5, SW5, C5, C1 SW8 C8, HH8, G7, R8, R81, N8, N81, MT8, NA8

No additional R41, A4, MIL5, N5, SW2, HH5, BOW5, CC2

Responders CC91, C9, CC92, CC93, EA9, SW9, SW92, C91

R9 HH9, HH91, NA92 No additional SW91, N94, HH9, HH91, SW9, N9

Inf Syst Front (2020) 22:897–909 903

is the responsibility and capability of each agency (in term of resource availability) and responder (in terms of training), and what information is accessible to them (see quotes below).

Not all information was routed through us [dispatch]. Due to lack of centralized information, it was easy to get out of loop during peak times The command structure showing ‘who is responsible for what’ was hard to maintain during the initial operations phase

The who-, whom-type structure recognizes agencies or responders assigned to each task, the when-type structure identifies the time instance, the where-type structure re- fers to the location, and the what-type structure identifies the resources assigned to each task (Chen et al. 2005). Such a structuring helps incorporate the entire command structure that provides the following: task-prioritization for decision-making, summary of agencies assignments, and allocation of resources to appropriate tasks. This makes it easy for responders to find the appropriate infor- mation (Bharosa et al. 2010).

Fig. 4 a: Illustration of Interacting Units (ARR Units). b: Illustration of Interacting Messages (Notifications and Updates).

Fig. 3 Interacting ARR Units

Inf Syst Front (2020) 22:897–909904

5.2 Communication Dimension

Center for American Progress’ National Security issue of 2005 states that, currently, the US has very few systems in place to allow emergency response personnel communicate reliably and effectively in a crisis. The current systems in place include radio communicators, cues on paper, and mental notes (Chen et al. 2008). The interviews with experts clearly iden- tified the importance of communication, as elaborated in the below quotations.

There was large number of responding units. There was large number of incoming calls from the scene. Recognition of the communicator was very important in deciding prioritization of that call. Incident Commander was topmost priority Personal information like Chief’s phone number could not be transmitted on air. Thus people needing the infor- mation had to request us [dispatchers]. This information was used by media for latest news. This led to influx of calls from media during crunch periods

During an emergency event, the lack of standardization in a messaging structure renders these messages incomprehensible for the receiver (Chen et al. 2012). Thus, the development of a messaging format not only helps in standardizing the messag- ing structure, but also helps in its reusability and

reproducibility. The messaging format that identifies the co- operator and the cooperative message objectives also provides for its semantic interpretation. Consequently, the standardized format can lead to appropriate interpretation of data by other agencies that are a part of the response (Valecha et al. 2013).

5.3 Systems Dimension

Emergency dispatch systems provide essential support to emer- gency responders that enable dispatch responders to answer more calls, prioritize responses to critical calls, and deal with complexity and pressure in an emergency context (Kim et al. 2006). These systems mainly depend on radios for communica- tion between responders and agencies. However, there are sev- eral problems associated with radios that call for development of an alternate mode of communication. Following insights, asso- ciated with radio issues, were provided by the dispatchers.

Radios are lifelines during emergencies. However, ev- eryone was talking on the radio at the same time. Co- working of radio channels for different operations or different emergencies was extremely challenging Plane crash incident was managed over four channels. Some people faced problems in channel switching ow- ing to ‘fat fingers’ due to gloves or other hand protec- tions. Some people didn’t remember to switch to right the channel

Templa�c Interac�on Dynamic

Interac�on

Fig. 5 Illustration of Interacting Units (ARR Units)

Table 5 Interactions at each level of situation awareness

Perception (<15 min)

Comprehension (15–60 min)

Projection (>60 min)

Interaction More emphasis on responders More emphasis on resources More emphasis on agencies

Messages More emphasis on notifications More emphasis on updates More emphasis on updates

Objective RESCUE TREATMENT SUPPRESSION

Resources More emphasis on rescue trucks More emphasis on ambulances More emphasis on rescue trucks

Inf Syst Front (2020) 22:897–909 905

Recently, a number of data standards have been developed for effective information exchange (Bharosa et al. 2010). However, emergency systems are still fragmented and disintegrated. The non-interoperable systems have made it difficult to communicate critical information between differ- ent departments in a timely manner. Interoperability is one of the important steps in responding to national emergencies (Chen et al. 2013). The dispatchers identified how information into and out of the systems was a problem during the incident as follows:

For gas shutoffs in the neighborhood, the Incident Commander had difficulty in identifying the account- able utility company due to lack of information from their systems. This resulted in back and forth between National Fuel and NYSEG (gas companies) arrivals to the scene There was no easy means for information to outside public. It was all routed though us [dispatch], which led to increased volume of calls pertaining to that information

A standardized system is very important for message encoding and decoding perspective. It allows for providing interoperable solutions for emergency systems. Universal Core (UCORE) provides such a framework that facilitates emergency interactions for incidents, such as forest fires, by providing a means for standardizing emergency messages. To address the problem of interaction across departments, National Information Exchange Model (NIEM) develops and supports standards for cooperating during an emergency

situation, using XML data model to standardize interactions between emergency agencies. Utilizing such systems will im- prove interactions between the emergency responders (Valecha et al. 2012a, b).

In this paper, we discuss the response to the incident focus- ing on interactions, and along the dimensions of information, communication and system use. The results from situation analysis of the interactions at each stage of emergency help in learning about the interactions in each phase (see Table 8): In the first phase, the incident commander perceives the crash intensity, and calls for additional help. This results in greater information sharing and variety of communication processes. In the second phase, the incident commander comprehends the availability of units at hand. This results in high informa- tion processing. In the third phase, the incident commander projects the learning of the incident to further decision-making actions, using various simulation tools and equipment. This results in higher levels of communication and system utiliza- tion. The learning from emergency interactions can used to derive generalized design requirements.

6 Discussion & Conclusion

This paper contributes to research in emergency management. On the basis of case study, this paper aims to explore emer- gency interactions, and develop a framework to identify the patterns of emergency interactions in a routine emergency context, and characterize emergency interactions in different stages of emergency response. With the help of interviews with four experts (two fire chiefs and two dispatchers) and

Ac�ve Interac�on

Passive Interac�on

Fig. 6 Illustration of Messages for Interaction (Notifications and Updates)

Table 6 Interaction types at each stage of situation awareness

Perception (<15 min)

Comprehension (15–60 min)

Projection (>60 min)

Templatic Interaction Request for resources based on mutual aid. N/A N/A

Dynamic Interaction N/A Multiple cycles of request for additional resources based on status checks.

Inf Syst Front (2020) 22:897–909906

raw incident communication report, key interactions are iden- tified in the categories of system use, communication and information. This paper also puts forth propositions for eval- uating the framework.

Such a framework spells out the actual characteristics of emergency responder interactions in extreme situations (Valecha et al. 2019). It would allow emergency organizations several benefits such as: studying responder interactions under the prism of information communication technologies (ICT), getting an evaluation of emergency systems supporting re- sponder interactions, training responders to control informa- tion based on their professional activities, tailoring system designs to respond to individual behavioral dynamics, and promoting interactive practices in time-sensitive situations.

In addition, this paper recommends the three dimensions, namely information, communication and system/technology, for emergency system design: First, focusing on the communi- cation dimension of interaction, we argue that emergency sys- tems should develop message formats. Second, referring to the information dimension, emergency responders need to identify key information structures. Third, emergency systems should implement standardized frameworks for effective interaction. These dimensions can be validated by replicating which dimen- sion in dominant in which emergency situations in future studies.

Moreover, the mapping of the dimensions to situation forms the theoretical foundations for the design of an emer- gency response system. In the perception stage of Situation Awareness, the system should provide observation capabili- ties – ‘what incident commander sees’ – including any con- flicts and uncertainty expressed at arbitrary levels of detail, which are a part of templatic interactions. Similarly, in the comprehension stage of Situation Awareness, the system should provide evaluation capabilities – ‘what incident com- mander thinks’ – presenting the incident commander with plans, justifications and group reasoning, which form dynamic interactions. Additionally, in the projection stage of Situation

Awareness, the system should provide capabilities – ‘what incident commander wants to execute’ – including requests, responses, referrals and recommendations of third parties, which are a part of active and passive interactions.

This research also has further implications for design sci- ence research: (a) It discusses design of a framework based on the findings of real world data. This branch of IS research methodology has also been shown to be effective in develop- ing various software (Luse et al. 2011). (b) This study offers a knowledge contribution that is a situated design of an artifact. Based to the guide proposed in Gregor and Hevner (2013), this research serves as a foundation to contribute to a design theory (Kuechler and Vaishnavi 2012). (c) This paper is one of the early works in investigating interaction patterns from real world data in the context of understanding emergency re- sponder behavior. This echoes that the outcome of the artifact is novel (Gregor and Hevner 2013; Peffers et al. 2007).

To further extend this research, researchers can perform in- terviews with responders that are a part of the second phase i.e. recovery phase, in addition to performing interviews at each level of incident command structure. This would help generalize the framework to higher levels of hierarchy, and in broader sections of the incident command structure. Some other future extensions may implement a prototype system and perform ta- ble top exercises with the expert in order to measure improve- ment in the response time. The paper has certain limitations. First, it considers findings based on a single incident. Second, it considers qualitative data in the form of interviewswhich have to be generalized to derive the findings. Third, since the plane crash was the first of its type inWestern NewYork, the details of the findings are unique to only this incident.

Acknowledgements The authors would like to thank James Zymanek (Town of Amherst Disaster Coordinator, NY; past Chief of Williamsville Fire Department, NY), Dave Humbert (Chief of North Bailey Fire Department, Amherst, NY), Steve McGonagle (Captain/Chief of Amherst Police Department), Tom Maxim (CEO of Twin City Ambulance), Tiger Schmittendorf (NY State Office of Fire Prevention and Control), Dennis Carson (Police Emergency Services Coordinator, Town of Tonawanda, NY), and Tim Oliver (Dispatch Center, Audubon Parkway) for their great help in this research project. The author would like to thank the review team for their critical comments that have greatly improved the article.

Publisher’s Note Springer Nature remains neutral with regard to jurisdic- tional claims in published maps and institutional affiliations.

References

Bachura, E., Valecha, R., Chen, R., & Rao, R. H. (2017a). Data breaches and the individual: An exploratory study of the OPM hack. In In international conference on information systems (ICIS) 2017. Seoul: South Korea.

Bachura, E., Valecha, R., Chen, R., & Rao, H. R. (2017b). Modeling public response to data breaches. In In American conference on information systems (AMCIS). Boston: Massachusetts.

Table 7 Patterns of Interactions in Emergency Stages

Perception (<15 min)

Comprehension (15–60 min)

Projection (>60 min)

Notifications HIGH LOW HIGH

Updates LOW LOW HIGH

Table 8 Interaction Requirements at Each Stage of SituationAwareness

Perception (<15 min)

Comprehension (15–60 min)

Projection (>60 min)

Information High High Low

Communication High Low High

Systems Use Low Low High

Inf Syst Front (2020) 22:897–909 907

Benbasat, I., Goldstein, D. K., & Mead, M. (1987). The case research strategy in studies of information systems. MIS Quarterly, 11(3), 369–386.

Bharosa, N., Lee, J., & Janssen, M. (2010). Challenges and obstacles in information sharing and coordination during multi-agency disaster response: Propositions from field exercises. Information Systems Frontiers, 12(1), 49–65.

Chen, R., Sharman, R., Rao, H., and Upadhyaya, S. (2005). Design prin- ciples of coordinated multi-incident emergency response systems, in Intelligence and Security Informatics, P. Kantor, G. Muresan, F. Roberts, D. Zeng, F.-Y. Wang, H. Chen and R. Merkle (eds.). Springer Berlin / Heidelberg, pp. 177–202.

Chen, R., Sharman, R., Chakravarti, N., Rao, H.R., and Upadhyaya, S.J. (2008). Emergency response information system interoperability: Development of chemical incident response data model. Journal of the Association for Information Systems (9:3), p. 7, 200, 230.

Chen, R., Sharman, R., Rao, H.R., andUpadhyaya, S. (2012). Datamodel development for fire related extreme events: An activity theory ap- proach. MIS Quarterly. Forthcoming.

Chen, R., Sharman, R., Rao, H. R., & Upadhyaya, S. J. (2013). Data model development for fire related extreme events: An activity the- ory approach. Mis Quarterly, 37(1), 125–147.

Cigler, B. A. (1988). Emergency management and public administration. Crisis management: A casebook, pp., 10–12.

Comfort, L. K., Ko, K., & Zagorecki, A. (2004). Coordination in rapidly evolving disaster response systems. American Behavioral Scientist, 48(3), 295–313.

Curra, T., Bergin, R., Josefek, R. (2009). A new role for emergency management: Fostering trust to enhance cooperation in complex adaptive emergency response systems. Naval Postgraduate School Thesis from http://www.dtic.mil/dtic/tr/fulltext/u2/a514087.pdf. Accessed 21 Jan 2019.

De Moor, A. (2009). Collaboration patterns as building blocks for com- munity informatics. In Proc. of the 6th Prato Community Informatics Research Network Conference, Prato, Italy.

Diniz, V., Borges, M., Gomes, J., & Canós, J. (2005). Knowledge man- agement support for cooperative emergency response. IEEE, 1182, 1188–1193.

Elliott, M. (2006). Stigmergic cooperation: The evolution of group work. m/c journal, 9(2).

Endsley, M. R. (1995). Toward a theory of situation awareness in dynam- ic systems. Human Factors, 37(1), 32–64.

Faraj, S., & Xiao, Y. (2006). Coordination in fast-response organizations. Management Science, 52(8), 1155–1169.

Gregor, S., & Hevner, A. R. (2013). Positioning and presenting design science research for maximum impact. MIS Quarterly, 37(2), 337– 355.

Hevner, A. R., March, S. T., Park, J., & Ram, S. (2008). Design science in information systems research. Management Information Systems Quarterly, 28(1), 6.

Hollingshead, A. B., McGrath, J. E., & O’Connor, K. M. (1993). Group task performance and communication technology: A longitudinal study of computer-mediated versus fact-to-face work groups. Small Group Research, 24(3), 307–333.

Janssen, M., Lee, J., Bharosa, N., & Cresswell, A. (2010). Advances in multi-agency disaster management: Key elements in disaster re- search. Information Systems Frontiers, 12(1), 1–7.

Jarvenpaa, S., & Ives, B. (1994). The global network organization of the future: Information management opportunities and challenges. Journal of Management Information Systems, 10(4), 25–33.

Kim, J., Sharman, R., Rao, H., Upadhyaya, S. (2006) Framework of analyzing Critical Incident Management Systems (CIMS). Proc. Of Thirty-Ninth Hawaii International Conference on System Sciences (HICSS-39), 2006, Hawaii.

Kuechler, B., & Vaishnavi, V. (2012). Characterizing design science the- ories by level of constraint on design decisions. In International

conference on design science research in information systems (pp. 345–353). Berlin, Heidelberg: Springer.

Kuenzel, E., & Welscher, H. (2009). Management Model for Successful Collaboration in the Public Sector. [Managing Collaboration] Federal Ministry for Economic Cooperation and Development.

Luse, A., Mennecke, B. E., Triplett, J. L., Karstens, N., & Jacobson, D. (2011). A design methodology and implementation for corporate network security visualization: A modular-based approach. AIS Transactions on Human-Computer Interaction, 3(2), 104–132.

Majchrzak, A., Jarvenpaa, S. L., & Hollingshead, A. B. (2007). Coordinating expertise among emergent groups responding to disas- ters. Organization Science, 18(1), 147–161.

Malone, T. W., & Crowston, K. (1994). The interdisciplinary study of coordination. ACM Computing Surveys (CSUR), 26(1), 87–119.

Mendonça, D. (2007). Decision support for improvisation in response to extreme events: Learning from the response to the 2001World Trade Center attack. Decision Support Systems, 43(3), 952–967.

McGuire, M., & Silvia, C. (2010). The effect of problem severity, man- agerial and organizational capacity, and agency structure on inter- governmental cooperation: Evidence from local emergency man- agement. Public Administration Review, 70(2), 279–288.

Peffers, K., Tuunanen, T., Rothenberger, M. A., & Chatterjee, S. (2007). A design science research methodology for information systems research. Journal of Management Information Systems, 24(3), 45– 77.

Rao, H. R., Chaudhury, A., & Chakka, M. (1995). Modeling team pro- cesses: Issues and a specific example. Information Systems Research, 6(3), 255–285.

Rautenstrauch, T., Generotzky, L., Bigalke, T. (2003). Kooperationen und Netzwerke - Grundlagen und empirische Ergebnisse. Josef Eul Verlag, Lohmar-Köln.

Rossel, P. O., Herskovic, V., & Ormeño, E. (2016). Creating a family of collaborative applications for emergency management in the firefighting sub-domain. Information Systems Frontiers, 18(1), 69– 84.

Samba, M.M. (2010). Intergovernmental Cooperation in Emergency Management: The Case of the September 2009 Georgia Flood, from http://digitalcommons.kennesaw.edu/cgi/viewcontent.cgi?article= 1429&context=etd. Accessed 21 Jan 2019.

Schulz, S. F. (2009). Disaster relief logistics: benefits of and impediments to cooperation between humanitarian organizations (Vol. 15). Haupt Verlag AG.

Strater, L.D., Endsley, M.R., Pleban, R.J., and Matthews, M.D. (2001). Measures of platoon leader situation awareness in virtual decision- making exercise. USArmyResearch Institute for the Behavioral and Social Sciences.

Turoff, M. (2002). Past and future emergency response information sys- tems. Comm. of the ACM, 45(4), 29–32.

Valecha, R. (2015). Information & communication in mediated crisis response. Doctoral dissertation, State University of New York at Buffalo.

Valecha, R., Sharman, R., Monteiro, J., Rao, H. R., Upadhyaya, S., Keerthana, B., Patel, M., Singh, A., & Sharma, K. (2010). A proto- type of a fire-related extreme EventZ system (FREEZ). St Louis: Proceedings of WITS.

Valecha, R., Sharman, R., Rao, R., & Upadhyaya, S. (2012a). Emergency response system design: An examination of emergency communi- cation messages. In International conference on design science re- search in information systems (pp. 139–146). Berlin, Heidelberg: Springer.

Valecha, R., Sharman, R., Rao, H. R., Upadhyaya, S. (2012b). Messaging Model for Emergency Communication. Proceedings of Midwest Association on Information Systems (MWAIS) 2012. Green Bay, Wisconsin.

Valecha, R., Sharman, R., Rao, H. R., & Upadhyaya, S. (2013). A dispatch-mediated communication model for emergency response

Inf Syst Front (2020) 22:897–909908

systems. ACM Transactions on Management Information Systems (TMIS), 4(1), 2.

Valecha, R., Bachura, E., Chen, R., & Rao, H. R. (2016). An Exploration of Public Reaction to the OPM Data Breach Notifications. In Workshop on E-Business (pp. 185–191). Springer, Cham.

Valecha, R., Sharman, R., Rao, H. R., Upadhyaya, S. (2019). An activity theory approach to modeling dispatch-mediated emergency re- sponse communications. Journal of the Association for Information Systems (JAIS), 20(1), 308-327.

Waugh, W. L., Jr., & Streib, G. (2006). Cooperation and leadership for effective emergency management. Public Administration Review, 66, 131–140.

Rohit Valecha is an assistant professor in the Department of Information Systems and Cyber Security at The University of Texas at San Antonio. He graduated from the University at Buffalo. He has research interests in crisis response, social media and system design. He also specializes in methodologies using Natural Language Processing, Text Mining, Social Network Analysis and Machine Learning. He has published in the ACM Transactions on Information Systems (TMIS), Journal of the Association for Information Systems (JAIS), Information Systems Frontiers (ISF) and the Networking and Information Technology Research and Development (NITRD) Program. He has also received NSF EAGER and RAPID grants.

Inf Syst Front (2020) 22:897–909 909

Reproduced with permission of copyright owner. Further reproduction prohibited without permission.