Applying social network analysis to evaluate preparedness through coordination and trust in emergency management

The purpose of this paper is to examine the status of the preparedness within emergency response team (ERT) in a refinery. Preparedness was investigated through trust and coordination relationships. Social network analysis as a quantitative approach was utilized in this research. To do this, social network analysis (SNA) indicators including density, degree, reciprocity and transitivity were utilized as a whole network. These indicators were calculated at the levels of first-line, supportive and whole teams. The required data were collected through structured interviews and were analyzed using UCINET 6.0 social network analysis program. The results of this study indicate that first-line teams can play a critical role in ERT, which is related to a higher level of SNA indicators and consequently the preparedness between team members can be easily achieved. In addition, the findings for the supportive teams revealed that they had relatively a low level of cohesion. However, the results of whole networks among all of teams had low level of cohesion that is a key challenge for performance of ERT. According to statistical results, there is a high correlation (82%) between trust and coordination networks. The finding of SNA provides a main opportunity for managers and planners to detect preparedness challenges based on coordination and trust ties among response teams of emergency management. This research suggests that fundamental efforts along with evaluation of the effectiveness of programs are needed to improve the presented situations and in order to optimize preparedness between response teams.


Introduction
Because of emergency and its negative consequences, emergency management (EM) is considered as a necessity in industries. Lowering the negative consequences requires effective management of emergency through preparedness of many different responding teams (Ford & Schmidt, 2000). This helps responding teams to react to emergencies as effectively and reasonably as possible. Effective coordination is an important issue in emergency preparedness of teams (Kapucu, Augustin, & Garayev, 2009). In fact, effective preparedness required close coordination among all the responding teams of emergency response teams. The purpose of coordination is that the teams that work together support and reinforce each other. Therefore, coordination results in the directing of energies and efforts so that all teams would come together to reach a common goal. In order to achieve an efficient coordination, team members should be thoroughly aware of each other's roles and responsibilities among different responding teams before an emergency occurs (Prizzia, 2008). This awareness is achieved through coordination of activities and services (Ford & Schmidt, 2000). It helps to assure the proper management of activities and entail schedule tasks interdependencies (Chen, Sharman, Rao, & Upadhyaya, 2005). It is also essential for the minimization of the duplication of services (Keeney, 2004). On the other hand, lack of coordination makes members compete with each other for resources, decision-making will take more time, and there will be incompatibility and misunderstanding in responders regarding responsibilities and services (2000), which was identified as a main problem in Katrina Hurricane (Moynihan, 2009).
According to studies (Comfort & Haase, 2006;Comfort, Ko, & Zagorecki, 2008) lack of coordination between responding teams has been accepted as one of the most important challenges that damaged the teamwork of response teams in EM, in situations like large California wildfires (William, Waugh, & Streib, 2006). In addition, the efficiency and performance of the emergency response team (ERT) was considerably affected by the achievement of effective coordination among responding teams. Moreover, coordination is strongly dependent on the trust perception among teams, and quality of the interactions for coordination is dependent on the trust level perception ( Figure 1) (Bodin, Crona, & Ernstson, 2006). As illustrated in figure, preparedness in EM can be described as a flow of trust within responders of ERT, which leads to the coordination of activities and services and then the improvement of the readiness of responding teams.
Researcher suggests that one of the most influential factors for the success of management process is the existence of trustful ties between various groups (Bodin & Crona, 2009) and it is an essential factor for effective emergency preparedness of different responding teams (Sommer & Pearson, 2007).
. Trust in Oxford Dictionary (Oxford dictionaries, 2011) defined as to have 'Belief in the reliability or ability of somebody or something.' So it can be defined among team members in EM as below: . To have a confidence toward each other's reliability or ability to do well ones duties and responsibilities those were assigned to them in the structure of the EM.
Trust promotes sharing and facilitates the flow of information and perceptions and creates a supportive climate for coordination (Varda, Forgette, Banks, & Contractor, 2009). It plays a vital role in establishing the conditions for effective inter-organizational coordination in emergency (Stephenson, 2005) and in sharing or retrieving information (Ley, Pipek, Reuter, & Wiedenhoefer, 2012). Investigations indicate that trust across teams allows more effective and have a good understanding of each other's abilities (Turoff, Chumer, Walle, & Yao, 2004). It is one of the most important aspects for good decision-making (Sommer & Pearson, 2007). Also, trust is considered as an essential attribute for more effective cross-organization coordination among groups and organizations (Stephenson, 2005). In fact, obtaining the desired goals during an emergency needs more structured coordination and trust among response teams. Thus, before the response teams enter an emergency and try to collaborate with each other, it is important to have an understanding of the preparedness status based on the coordination and trust ties among teams and to support it if necessary in order to have an effective response. The purpose of this research as a case study is to conduct a quantitative assessment of coordination and trust ties among team members of ERT in a refinery. To this aim, the principle of social network analysis (SNA) was applied.
In the remaining sections of the paper, first SNA and indicators are explained. Then, the methodology including ERT, data collection and analysis will be presented and, at the end, results and discussions will be dealt with.
2. Social network analysis 2.1. SNA indicators SNA expressed relational concepts for studying structures of a network that were referred to as structural analysis of relationships. It studies and analyzes the social relationships of large number of actors between different groups of organizations and provides a mathematical approach for measuring the strength of ties (Furht, 2010;Magsino, 2009). It comprises actors that are tied to one another and can be analyzed for structural patterns to examine their relations structure into the overall network (Scott, 2000;Wasserman & Faust, 1994). SNA has an important role in determining the degree of team's success in reaching their goals and in evaluating the performance of the whole network (Abbasi & Altmann, 2010). There is also a vital element in the determination of the relationship between different parts of a network which is known as cohesion (Carrington, Scott, & Wasserman, 2005;Scott, 2000;Wasserman & Faust, 1994) and it is regarded as an important feature of whole networks. The important concepts of this feature include density of ties, degree of centralization of the network, reciprocity and transitivity in the network, and if each of these indicators increases the cohesive level of network would enhance and coordination would improve (Carrington et al., 2005;Hirschi, 2010;Kilduff & Tsai, 2003;Scott, 2000). In the following, these indicators are explained.

Density
The density is an indicator for the general level of relations of a network, which is the total number of direct relations between actors of a network divided by the number of possible relations (Furht, 2010;Wasserman & Faust, 1994). This measure has values between 0 and 1. In denser or more cohesive networks, the score is 1% or 100%, which implies that all actors in the network are connected together and trust each other. The score of 0 shows the actors are entirely disconnected and do not trust each other (Bodin et al., 2006;Kilduff & Tsai, 2003;Wasserman & Faust, 1994). For example, the density 0.67 indicates that 67% of all possible ties are present within the network.

Degree centrality
Degree centrality implies the whole number of direct relations that an actor has with the other network's actors (Carrington et al., 2005;Wasserman & Faust, 1994). In a directed network, where the direction of the relations is significant, the degree is divided into in-degree and out-degree centrality (Furht, 2010;Song & Miskel, 2005;Wasi, Usman, & Shoab, 2014) that indicated power and influence members in the network respectively (Abdel-Ghany, 2012;Ortiz-Arroyo, 2010). In-degree implies the number of coming-in connections or selections that a given actor receives. In other words, it is the number of times that an actor is selected by other actors. Out degree implies the number of going-out connections from a given actor in a network. In other words, it is the number of times that an actor selects other actors (Chai, Liu, Zhang, & Baber, 2011;Furht, 2010;Pryke, 2012). Both indexes are measures that vary between the values of 0 and 1. The high values of in and out degree in the whole network imply that large numbers of central actors have great reputation and influence on other members at the coordination and trust networks (Furht, 2010;Pryke, 2012;Scott, 2000;Wasserman & Faust, 1994).

Reciprocity
Reciprocity index describes the degree to which a member has mutual ties to another member (Kilduff & Tsai, 2003;Wasserman & Faust, 1994). In direct relations between A and B, reciprocity indicator implies if A has a tie with B, it is expected of B to have a tie with A that implies ties in both directions. The score of networks with high mutual ties is 1% or 100%, showing that all members in the network have tendency to be reciprocally connected together and trust each other, and the score of 0 shows the inverse results (Carrington et al., 2005;Kilduff & Tsai, 2003;Wasserman & Faust, 1994).

Transitivity
Transitivity index refers to the tendency of two members in the network to be connected if they have a common neighborhood to complete the triple relations (Carrington et al., 2005;Kilduff & Tsai, 2003;Wasserman & Faust, 1994). In direct relations among triples of actors including A, B and C, transitivity implies if A is coordinated with B or trusts it, and B is coordinated with C or trusts it, then A is likely to have coordination with C or trust it. This index has values between 0 and 1. The score of networks with high triple ties is 1% or 100%, which shows that all members in the network have tendency to have triple ties together and the score of 0 shows the inverse results (Carrington et al., 2005;Kilduff & Tsai, 2003;Wasserman & Faust, 1994).

Correlation between two networks
Quadratic assignment procedure from UCINET can be used to obtain the correlation coefficient of two matrices (Borgatti, Eveertt, & Freeman, 2002). There are a lot of measures such as, Pearson's correlation coefficient, matching coefficient and Jacquard coefficient, which show correlation between corresponding cells of the two data matrices based on the type of relations. Since both matrices have binary relations, such as coordination and trust, the Jacquard coefficient would be a reasonable measure to obtain the correlation (Borgatti et al., 2002).
2.2. SNA and preparedness SNA has been applied to help decision-makers and planers to identify different relations between organization in different teams and organizations during emergencies and disasters. Inter-organizational collaboration research revealed that SNA is critical for designing interventions to promote collaboration among medical and social services (Kapucu, 2005;Magsino, 2009). SNA has been used to analyze the collaboration of users and agents in a recommender system (Palau, Montaner, L'Opez, & Rosa, 2004). Some researchers applied SNA to measure the coordination of project team members (Hossain, Wu, & Chung, 2006) and organizational coordination (Hossain & Wub, 2009). Responding to catastrophic disasters, SNA also assesses the relationship between responding organizations and their emergency coordination operations (Kapucu et al., 2009). In multi-organizational networks following the Katrina Hurricane and terrorist attacks, SNA was employed to find out the key actors and major organizations that coordinate in the response system (Kapucu, 2005). In trust researches, SNA has been used to assess the strength of social ties within social relationship of individuals and trust (Bapna, Gupta, Sundararajan, & Rice, 2011) to evaluate trust relationships between individuals through the degree of centralization (Ziegler & Lausen, 2005) and in informal economy (Lomnitz & Shenbaum, 2004). The research of inter-organizational coordination in emergency response management through the centralization index of SNA found that the effective response requires well-coordinated networks and trust between groups at all levels (Kapucu, 2005). In fact, it is important to have an understanding of the preparedness status based on the coordination and trust in EM by SNA. This paper tries to reach this purpose.
Note: HSE, health, safety and the environment. and rescue teams were considered as the first line of response, and then the medical team arrived on the scene. The remaining teams such as health, safety and the environment (HSE) acted as supportive of the first-line teams. It was expected from them to respond to the emergency as effectively and reasonably as possible. Two important questions are raised here: do the response teams of ERT have effective coordination together and trust in each other, and are they highly prepared for ensuring effective response. In order to respond to these questions, the principle of SNA was utilized.

Data collection and analysis
Open questions were employed in this study. The required data were collected through structured interviews with each team member of the ERT. Based on the coordination criteria, the following question was asked from each member of the ERT: . Which member did you coordinate with the most to understand each other's roles, responsibilities and activities before an emergency occurs?
To measure the level of trust, each member was asked: . Does each ERT member have enough competence and ability to do their duties and responsibilities so that he is trusted by other members?
In order to respond to these questions, the formal identification list including name, responsibility and affiliation of individual team members was given to the responders.
The key issues in these questions were whether the team members trust each other and were coordinated together. All selections were then recorded, archived and analyzed as the whole network. This study uses binary data (absent, i.e. 0.0, and present, i.e., 1.0) and directional relations. The value is present if there is a selection within the team members. If each pair of team members does not choose each other, the value of 0.0 will be allocated. The tie is directed from one member to another in a pair; that is, it has an origin and a destination (Wasserman & Faust, 1994). The connections for directed data are asymmetrical, since a directed line from officer of firefighting to rescuer will not necessarily involve a reciprocated line directed from members of rescuer to officer. The analysis and visualization of the survey are performed using UCINET (Version 6.0) SNA program (Borgatti et al., 2002).

Findings of density indicator for ERT
Based on the structure of ERT, density index of coordination and truth networks were measured in first-line teams (two teams and three teams), supportive teams and the whole teams separately.

Two teams
To determine the cohesiveness of two teams of firefighting and rescue (as the emergency first-line teams) in both networks, density indicator was calculated as a whole network ( Table 2). The finding indicated that both teams had dense networks. For instance, the density of trust between firefighting and rescue was 0.95, meaning that 95% of all the possible relationships between members of firefighting and rescue members were established. In addition, the possible relationship between members of rescue and firefighting in coordination was obtained to be 97%.

Three teams
The density of three teams, including firefighting, rescue and medical team based on the trust and coordination network, shows that it is nearly desirable ( Table 2).

Supportive teams
The results of these teams including HSE, logistic, public relations and security indicated that there is relatively low level of ties among them (Table 2).

Whole teams
To determine the cohesiveness of two networks, density indicator was calculated as a whole network within seven response teams (Table 2). For example, the density of the coordination network was obtained to be 0.23, implying that 23% of all possible ties among members of the response teams are present. It indicates that the actual tie among response teams is very limited compared to the probable ties that may occur if coordination among teams is conducted optimally. In addition, the density level of trust between team members is 0.18, meaning that 18% of all the possible relationships between members of ERT were established. The results showed that density level for two networks was small.

Degree of centrality
Degree was measured both in-and out-centralization at the whole network. According to the findings, coordination and trust networks had a relatively low in-and out-degree centralization index (Table 3) and there were limited central members with enough reputation and influence in both networks.

Reciprocity
The findings revealed that in both networks 87% of members in the ERT had reciprocated connections, which reflects high value of mutual relations (Table 3). In trust network, most of the team members had reciprocated trust. In addition, in the coordination network, responders were interested in making mutual coordination with each other and keeping up their coordination. In general, the findings showed that both network had stable ties.

Transitivity
Transitivity examines three members and the level of trust and coordination relationship between them. The results indicated that 50% and 35.42% of members in the ERT had triple connections based on trust and coordination, which reflects moderate and low values of triplet relations (Table 3). According to the findings, triple relations in both networks were reduced compared to the mutual connections, which was particularly more obvious in the coordination network (Table 3). This implies low stability of the network, which is clearer in the coordination.

Correlation between two networks
The results of the index showed that there was a high correlation of 82% between the two ties of trust and coordination. The statistical significance of the correlation of the two ties was zero (P < .001), implying that the correlation of the two ties was highly significant. Therefore, it can be concluded that if there is trust between team members of ERT, it could be said with 82% of probability that there will be coordination in the network.

Discussion
It is notable that preparedness for EM includes various dimensions, such as training, allocation of resources and facilities. In this study, this issue was dealt with just in two dimensions of coordination and trust, which were examined according to the nature of the network and its relationships. EM is a multi-team activity and each team is different in terms of goals and responsibilities. To reach a successful EM, if each team member focuses on his own responsibilities without considering other members' activities, the whole system's function will be impaired, no result may be obtained and all the energy of the system will be wasted on the friction between teams. Therefore, a successful management of emergency depends on effective and reasonable coordination between members and teams. In addition, the presence of trust between team members leads to optimum relation and coordination between members, and finally the preparedness of team members will manage the emergency. On the other hand, preparedness will take more time and cost that is inconsistent with the objective of EM. Therefore, it was expected from teams of EM, which were seven teams in this work, to be at a suitable level of preparedness. In the present study, social network was utilized as a quantitative approach in EM. It is able to help decision-makers to identify the challenges of the preparedness. It can analyze the nature of relations of trust and the coordination network among the members of the response teams.
In this study, SNA indicators, including density, degree, reciprocity and transitivity in the whole network of ERT with different interpretation, were utilized. To determine the level of cohesion in the ERT network, density index was calculated in both networks in four levels. Dense networks are suitable for coordination of activity among team members. It caused members become familiar with each other's responsibilities (Furht, 2010) and will facilitate the coordination and trust among response teams, which play an important role in the preparedness. At the inside level of two teams of emergency first line, including firefighting and rescue, the results showed an acceptable high value of cohesion. The findings suggested that both teams had a high level of trust and coordination, and consequently the preparedness between team members was easily achieved. The result of three teams, when the medical team was added to the two previous teams, revealed a significant reduction at the cohesion level. The findings showed that the medical team with fewer interactions affected the level of index. In addition, the findings for the supportive teams revealed that they had relatively low cohesions. The results for the whole teams showed that with the increase in the response teams, the index heavily decreased. A high percentage of this value is allocated to supportive teams that operate as independent units and do not have active roles in both networks. According to the findings the first-line teams, including firefighting and rescue teams, were denser and had cohesive ties that played critical roles in creating sustainable ties among response teams and increases the chance for preparedness of teams during normal operations and emergencies. In this issue, essential actions are required. Finally, the results of the density in trust and coordination ties showed that cohesion of both whole networks was low, meaning that team members of the response team were weakly connected to prepare together. This is a challenge for EM and needs efforts to reach dense ties among response teams.
According to the results of the centralization index, both networks had a relatively low centralization. The finding of in-degree centrality means that the percentage of ties that were coordinated or trusted on key members was relatively low. It reflects the absence of key members around team members and poor relationship with key members. The out degree also replies that the influence percentage of the key members in terms of creating coordination and trust was relatively low. This finding implies that coordination and trust among team members are more evenly done. This finding was supported by previous studies (Hirschi, 2010).
Reciprocity and transitivity indicators showed the stability of a network and the cohesion of studied ties among team members (Hirschi, 2010;Kilduff & Tsai, 2003). The interval of two indicators is between 0 and 1. High performance can be achieved when the percentage of mutual and triple relations is high, and vice versa. Factually, these indicators are key factors in cohesion of trust and coordination networks and it will consequently facilitate the preparedness among response teams. According to the findings of this study, team members had a high level of mutual relations, which promotes more stable and cohesive ties within response team and is beneficial for strengthening the preparedness process of the ERT. When one member builds coordination and trust with two other members in response team, it means they are likely to be prepared together, provide coherent structure to each other and create denser ties in the ERT. Some research conducted in the field of management confirmed this finding (Bodin & Crona, 2009;Kilduff & Tsai, 2003). The results of triple relationships of the coordination ties revealed a decreased value in comparison to the mutual connections. This implies low stability of ties and shows that there was difficulty in coordinating the activities in the EM. In addition, the triple connections of trust revealed a moderate value. This result implies that members had cohesive relationships compared to the coordination network. The results of local collaboration networks and sustainable development study support this finding (Hirschi, 2010). According to the results, there was a high correlation between trust and coordination. The existence of correlations in coordination ties and trust ties between team members reflects a close relationship between the two networks under study. It can be said that the presence of trust is an important factor in coordination activities, the point that is supported by other studies as well (Kapucu, 2005).
The findings suggest that there are high levels of reciprocal ties with low to moderate transitivity in coordination and trust networks. Previous studies have shown that high reciprocity and in-degree for relationships between team members in the networks with high density are indicators of high team performance (Zenk, Stadtfeld, Windhager, & Allmendiger, 2013). In addition, some researchers also found that high transitivity in networks with high density reflects more cliques (a subset of members fully connected to each other) than those with low transitivity, and members have tendency toward standardizing performance and actions in the structure of the teams to which they belong (Carrington et al., 2005;Kilduff & Tsai, 2003). These findings are almost inconsistent with the results of this study. In fact, with regards to the low to moderate number of transitivity and high level of reciprocal relations in networks with low density, it can be concluded that there are low levels of preparedness in the ERT. Finally, SNA help planners and managers to have a clear understanding of the challenges and to make effective efforts to enhance and strengthen the situation.
In general, trust is an important factor in human and social relations and interactions among team members and groups. It plays a significant role in building coordination. Therefore, increasing trust in the emergency response management system can enhance preparedness. The finding of this study showed that lack of trust and coordination between the first line and supportive teams is a key challenge for EM. The results determined that due to regular holding of maneuver and training courses there were more trust and coordination between first-line teams compared to supportive teams. The findings of trainings using a standard questionnaire (Cronbach α = 0.8) showed that the average amount of training in first-line teams was 0.61 and in supportive teams was 0.26. According to the results, these teams played a more vital role in the management team, and being more effective, they can motivate other teams to enhance their preparedness. The main reason for insignificant density between first-line teams and supportive teams was lack of common programs. That was because the refinery under study had been newly established, it had a poor cooperation history, and consistent and cooperative training and maneuver courses were absent among teams. Therefore, in planning for the management of emergency, it is necessary to think of ways to build trust and coordination between teams and for this purpose, supportive teams should not be ignored. Participatory training, drills and maneuver are creative approaches to build coordination and trust among response teams in order to ensure that all the members of response team have opportunity to become familiar with each other's responsibilities and activities. Enhanced interaction among team members and improved relationships increase trust between members of EM system. Thus, the formation of informal groups, doing group activities and so on will result in the exchange of more coordination between members. Also, social similarity of team members in EM influences interpersonal trust, for example, drivers of different units, members responsible for the control of muster points, etc. Thus, these similarities are suggested to be taken into account in the division of labor, so that it provides the ground for more trust, and finally improved coordination and preparedness. Factually, these could be beneficial in improving the level of coordination and trust, and consequently in strengthening emergency preparedness.

Conclusion
This study was a case study that assessed preparedness based on the ties of trust and coordination in ERT in a refinery. For this purpose, the SNA was utilized, which enables planners to view the actual status of response teams. The findings demonstrated that SNA provides a logical and quantitative approach to examine the status of preparedness among response teams in the ERT. The most obvious finding that emerged from this study was that the results of whole response teams have low percentage of preparedness based on trust and coordination, although, the findings of first-line and supportive teams revealed that they had relatively high and low level of preparedness, respectively. The research concludes that EM has been revised toward the existing training programs and has used creative programs to improve the presented situations that require fundamental efforts along with evaluation of the effectiveness of programs. Considering the time-consuming nature of the process of building trust between team members and the importance of trust in building coordination, it is necessary to exactly examine the causes and consequences of lack of trust and conduct studies to find solutions to increase trust and coordination in the population under study. These issues should be dealt with in future research and enquiries.