Abstract
An emergency management tool developed for training police, fire and ambulance teams in Norway and Sweden provided the data for this paper. The teams communicated with representatives of the local power companies and county and municipality officials in responding to various emergency scenarios in a web-based training tool. The project generated rich textual data of the content of the communications as well as a range of quantitative data on who communicated with whom, how often and with what type of information. The author analyzed the qualitative data using the NVivo software package and the quantitative analysis used the R statistics package and the social network analysis (SNA) module. The textual analysis shows distinct patterns of concepts and terms used by the various emergency response agencies. The quantitative analysis illustrates the flow of communication among the participants of the emergency management training (EMT). Visual representation of both the qualitative and quantitative data from the project provides a thorough insight into processes of communication among emergency response personnel in role-playing training sessions. The data visualization enhances the debriefing session following emergency response training. The research group at Mid Sweden University and NORD University recently received funding for a three-year continuation of the project. The new project will emphasize the demand side (community stakeholders) in addition to the supply side (the emergency personnel).
Keywords
Introduction
This paper reports from a research project funded by the EU through the Interreg Sweden-Norway funding agency of the EU Regional Development Fund (EU-RDF) from 2010–2013. The project developed a web-based tool for training emergency personnel in Norway and Sweden in handling critical situations in the border region of Jämtland and Trøndelag regions.
Recent research by Ha et al. [10] focus on the status of emergency management training (EMT) in Korea. The authors employ content analysis techniques and find that government instigated programs tend to emphasize the supply side with a focus on the training of personnel. The authors argue for a renewed focus on the demand side and stakeholders needs by including local businesses and the communities.
Training environmental agencies in handling emergencies by focusing on decision-making and communication within and between teams of participants is quite important, according to research by Stolk et al. [5]. McGuire [11] provides an analysis of 400 county-level emergency management agencies and concludes that the more severe the problem is the greater is the need for external collaboration. The Police College of Hong Kong has successfully used a Scenario-based Interactive Multi-player Simulation (SIMS) as an alternative to traditional face-to-face training, according to Leung [2]. The latter carries a higher cost compared with the computerized simulation. Ley et al. at the University of Siegen, Germany [1] have studied the role of improvisation in handling emergencies and found that this often requires the coordinated responses police, fire departments, public administration, electricity infrastructure operators and citizens. Sinclair [8] investigates several emergency operations centers in New Zealand, Canada and USA and finds that anecdotal data often becomes the base of practices related to preparedness –and consequently these practices are lacking systematic study and validation. Post-disaster research often uses the “lessons learned” approach of retracing decisions during the actual emergency. Birkland [14] identifies threats to the validity of such approaches describing the “lessons learned documents as “fantasy documents”. The approach utilizing a computerized emergency training tool will allow a focus on the ongoing processes during the training session rather than an after-the-fact approach.
The use of content analysis techniques in knowledge development is well documented. Savoia et al. [6] use content analysis of reports following emergencies to allow organizational and systems learning with respect to emergency preparedness. This paper suggests using content analysis techniques in analyzing open-ended data from logs of communications by participants in an emergency management training session in 2013 in the central part of Scandinavia.
Guy Walker et al. [7] at Brunel University analyzed communication between military and civilian organizations in the response to massive disasters. Content analysis of voice communications transcribed and parsed by a software package. The verbal transcripts allowed the establishment of networks of concepts (Propositional Networks) that aided in the understanding of problems relating to coordination of the disaster response agencies. The authors were also able to outline solutions to improve communications during multi-agency emergency response activities.
Several major emergencies occurred in the US from California forest fires in 1993 through the hurricane Katrina in 2005. D.P. Moynihan at the University of Wisconsin analyzed over 60000 text units with rich narrative descriptions of these major events by using the QSR N6 software package. The analysis shed light on two competing theories of how to manage communications during large natural and manmade disasters. The Incident Command System (ICS) is the mode of operation mandated by the Department of Homeland Security in the US, and the analysis sheds light on whether a hierarchical or network version of the ICS is the best approach. The network governance of response to crisis allows integration of “the need for inter-organizational collaboration and the need for rapid coordinated response” [4].
Emergency management training tool
The project GSS (Gaining Security Symbiosis) was developed jointly by Nord-Trøndelag University College and Mid Sweden University. With partners among emergency response agencies on both sides of the border, the project developed scenarios of emergencies including severe weather conditions, electrical power failure, traffic accidents and a family getting lost among steep cliffs on a cold, foggy night.
The computer based training tool consisted of standard web technology with LAMP (linux, apache webserver, MySQL database and PHP server-side programming language). A range of web browsers was used: Chrome, Firefox, Safari (iPad) and some versions of Internet Explorer.
The participants included personnel from all levels of the emergency teams –operational, tactical and strategic levels representing the supply side. In addition, the project also included participants representing the demand side –the municipality and county officials as well as the electric power companies in the region. In total 101 individuals have participated in training sessions over 3 years of simulations. This paper reports from the 2013 simulation in which more than 40 individuals participated.
Figure 1 shows the area of Scandinavia in which the simulated emergencies took place. In Sweden, the marked region covers part of the county of Jämtland while in Norway the region covers part of the counties of Nord- and Sør-Trøndelag.
The computerized training tool used text-based communications, and Fig. 2 shows the scenario and communications channel.
Figure 2 does not show the interface for initiating communications with other participants. This interface for participants allowed a choice of “requesting information” or “giving information” and a drop-down menu for choosing recipient of the message. The training tool allowed for open-ended messages with no limitation to the number of characters submitted. The training tool also allowed for relatively complex scenario messages that could include audio, video and multimedia content.
Textual analysis of open-ended communications
Content analysis
For more than fifty years computer technology has allowed efficient content analysis of open-ended text data. Such analysis utilized word count and presentation of selected words and expressions in the context of the overall text. Philip J. Stone and Earl B. Hunt at Harvard University and University of Sydney pioneered the use of computers in content analysis [12].
NVivo software for content analysis
NVivo (NUD*IST in the 1990’s and N6 in the 2000’s) is a software package developed by QSR International in Australia.
The basis of content analysis of open-ended text is the frequency with which terms and expressions are used. The text from the GSS project consists of texts in Swedish and Norwegian –and emergency teams on either side of the border are able to understand most of the neighboring language. The software will parse the sentences of any language and provide meaningful analysis of the frequency of use of these expressions in the local language.
Figure 3 shows the interface of the NVivo 10 software after having added the 13 word documents containing the communications of the various rescue teams in Norway (n) and Sweden (s).
Word cloud / word count
The analysis of the communications documents now proceeds to an exploratory phase –Query / Word Frequency / Select Items –and select one of the documents towards a comparison of the police in Norway with the police in Sweden –by first selecting n_Police_112 (Norway).
In Norway, three emergency numbers are used 110 (Fire), 112 (Police) and 113 (Ambulance) while in Sweden there is one central number 112 (SOS Alarm).
Figure 4 shows the selection of a single document of communications during the emergency training –that of the police in Norway (n).
Figure 5 shows the word cloud of communications by this group during the simulation.
Figure 6 shows the comparable word cloud of the Swedish police vs the Norwegian police in Fig. 5. By analyzing the relative frequency of the terms used in communications, Fig. 6 shows that the most frequently used term by the Swedish police was “Funäsdalen” –a geographical area west of Røros, Norway where a major bus accident took place in the scenario. In comparison, the most frequently used term by the Norwegian police was “orientert” –meaning being informed about the events in the scenario.
Cluster analysis: Exploringthe communications
The next step is to do an analysis and comparison of the communications among the emergency teams on either side of the border. The degree to which the various teams use similar or distinct terminology the cluster analysis will put teams that use similar terms near each other –and dissimilar teams further apart. Figure 7 shows the pattern of communications by the rescue teams, the power company and the county / municipal participants.
The cluster analysis of the Norwegian teams show that the police and the ambulance teams are quite central in the communication process presenting two “pillars” of communications. The municipal fire teams and the Fire_110 operators are closely related to the ambulance team in terms of the content of the communications. Similarly, the teams representing the power company and the county / municipal leaders express themselves more like the police team.
Looking at the communications pattern among teams on the Swedish side of the border displays a quite different “map” of communications (Fig. 7).
Among the Swedish participants in the emergency training sessions (Fig. 7) it is clear that the police and the central fire department switchboard in Östersund use a terminology that is different from the other teams. There were no representatives of municipality teams in Sweden, unlike among the Norwegian participants. The county officials and the fire department in Järpen (the municipality of Åre) communicate using similar terminology –and the SOS alarm emergency operators (112) and the power company show similar use of terms during the emergency training session.
Norway does not have a single emergency number like Sweden (112) –and in Norway the police have a special coordinating function in all emergencies, unlike in Sweden. Consequently, Fig. 7 shows that the 113 and 110 emergency numbers communicate using similar terminology while the police in Norway are more aligned with the power company, the county and municipality administration.
Query: Communications in context using word trees
The next step in the analysis is to look at specific terms used by various teams and the context that the terms are used. The analysis begins with the Norwegian power company and references to the term “electric power*” (strøm*).
Figure 8 presents the context of the use of the term electric power. The text refers to “.. that we have problems with the power supply..”, “In Lierne municipality we have reestablished the power supply..”.This provides a rich source of the context that the term is used.
In Fig. 9 we present the comparable word tree with the Swedish power company. The term used here is oil* (olje*) since the particular problem in the scenario on the Swedish side of the border was related to oil leaking from the main transformer in Åre.
Figure 9 shows the context of the use of the term oil leakage with expressions like “.. from the press with questions about the oil leakage from the main transformer in Åre..”. The Swedish power company handled primarily problems relating to transformers while the Norwegian power company faced an extensive power outage in one of the municipalities. The word tree of the Norwegian power company is consequently a more complex structure.
Figure 10 shows the context of the use of the term crisis (kris*) by the Norwegian police representatives. The context of the term shows “Police chief and crisis management team are informed”.
Figure 11 presents words in context from the Swedish police. The term focused on is ordered* (“larm*), and the context is “Åre police and civilian avalanche dog teams are ordered to the landslide area”. The terms used by the Swedish police represent more direct action of civilian dog teams and a police patrol (Fig. 11), while the Norwegian police use terminology more aligned with the command and control chain and passing on information to other agencies.
Query: English translationsof communication text
The comparison of rescue teams across the border is difficult in that the data consists of distinct Swedish and Norwegian texts. The native texts of the police participants in both countries needed translation in order to do a cross border comparison.
Figure 12 shows the word cloud representing the translated communication of the Norwegian police department. Often used terms “regarding”, “location”, “oriented” and “respect” represent terms relating to the emergencies.
Figure 13 presents the corresponding terms used by the Swedish police. The dominant terms here are “landslide”, “rescue”, “mountain” and “patrol”, and the latter three terms show that the Swedish police use more direct descriptions of the emergencies.
Figures 12 and 13 show similar patterns of words used by the Norwegian and the Swedish police as in the word tree Figs. 10 and 11.
Quantitative analysis of data from communications
The GSS project used the Social Network Analysis (SNA) component of the R statistics package to analyze the rich quantitative data from the emergency training sessions [3]. The project integrated the open source software R into the software solutions developed for the project. SNA and R statistics allow for performing the statistical analysis in a “batch” fashion with command based control of the analysis. This allowed presenting results from the training sessions within a short time following a three-hour simulation of an emergency simulation.
Table 1 presents the R and SNA batch command sequences required for the social network analysis provided in Fig. 14.
Figure 14 visualizes the pattern of communication among primarily operative level personnel in the police, fire and county representatives in Norway and Sweden during face-to-face emergency training sessions. Figure 14 shows the pattern of requests for more information regarding the simulated emergencies. The pattern of communications giving information to other emergency units is presented in another publication [9].
The police on both sides of the border (the letter “N” indicates the Norwegian side of the border and an “S” indicates the Swedish side). On the Norwegian side there are two counties where the police is labelled N-policeNT and N-policeST (for Nord-Trøndelag and Sør-Trøndelag). We see that the police on both sides of the border are key actors during the emergency training session. The S-112 represents the Swedish 112 emergency operator service that receives all incoming emergency calls and distributes the calls for assistance to the police, fire or ambulance services. In Norway, there is no similar single phone number for emergencies –the police, fire and ambulances have their own phone numbers.
Figure 15 presents participants providing information to other participants in the training session across the border. We notice that the Swedish 112 operators are centers of communication across the border while the police on both sides of the border play a lesser role in channeling information across the border. We also note that the county officials on both sides of the border exchange information –and so does the representatives of the electric power companies in Nord-Trøndelag and Jämtland –but these communication channels do not include the other actors in the training sessions.
In Table 2 based on data also used in Fig. 14 - we see that the proportion of messages requesting information across the border initiated in Sweden is about 30% (7 out of 23) –and the comparable number of messages initiated in Norway is about 15% (6 out of 41).
One of the advantages of the training tool developed for this project is the fact that it allowed operative personnel to take part in the training sessions. Previously, training sessions occurred during the annual meeting of the Border Rescue Council and the participants at these meetings were primarily tactical and strategic personnel. The simulated emergencies during these meetings were primarily paper-and-pencil table training sessions.
Conclusion
The challenge of rich sources of voice and/or written communications in connection with inter agency communications may be met by using computerized content analysis techniques. This paper has documented the feasibility of using a software package like NVivo in making sense of the text of rich communications submitted during an emergency training session of police, fire and ambulances in contact with the power company, county and municipality officials in the border region between Norway and Sweden. Multilingual texts may be used when comparing agencies within each country, but when comparing across borders a common language need to be used in order to make sense of the communications.
The project GSS also has a rich source of quantitative data on communication between rescue teams in Norway and Sweden during simulated emergencies. R statistics and the Social Network Analysis module (SNA) enables the visualization of patterns of communications that would be difficult to detect with the qualitative textual data presented in this paper. In this sense, the combination of qualitative and quantitative analysis of communication data from these emergency training sessions presents the data to its fullest extent.
The project team at Mid-Sweden University and Nord-Trøndelag University College has recently received funding in for a follow-up project (GSS 2)funded by the EU Regional Development Fund / Interreg Sweden-Norway. This project will include the use of social media as a communications channel between the emergency response agencies and the public, thus providing an even richer set of textual and network data for further analysis of communication between response agencies, the county and municipality officials, the power companies and the public on both sides of the border. Volunteer groups for search and rescue operations will also participate in the new project from 2015 until 2018.
Since 2013, both Sweden and Norway have implemented a new public safety network using radio terminals based on the TETRA (Terrestrial Trunked Radio) European standard. In November 2016 the project GSS 2 will analyze the cross-border communication between rescue teams in the border region between Norway and Sweden. Hollis [13] argues that international regional cooperation in disaster risk management (DRM) is required as huge disasters –often weather related –overwhelm the resources of the individual nations. Using the terminology suggested by Ha et al. [10], the supply side (emergency personnel) and the demand side (the community stakeholders) in both Norway and Sweden are looking forward to being able to use the new digital radio TETRA units as they cross the national borders in order to assist rescue teams in the neighboring country.
Footnotes
Acknowledgments
The project GSS (Gaining Security Symbiosis) received funding from the Sweden-Norway Interreg Fund / EU Regional Development Fund from 2010 to 2013.The same source of funding has financed the follow-up project GSS 2 which is starting in 2015 lasting until 2018.