Disaster response network analysis in rural Temerloh, Pahang communities during the Malaysia 2020-2021 flood

Disaster risk reduction practices can be viewed as a collaborative environment managed by a diverse group of stakeholders including governments, private sectors and non-governmental organizations and research institutes as well as local communities. Insufficient collaboration and failure to coordinate across groups can lead to unsuccessful disaster recovery efforts. This study investigates the organizational roles and collaboration network among governmental and community organizations participating in Malaysia 2020-2021 flood response in rural Temerloh, Pahang. Social network analysis was conducted using Gephi open-source software to examine the general patterns of structures and the characteristics of the networks of stakeholders. News reports and organizational situation reports about the inter-organizational interaction and collaboration of stakeholders were identified using the manual coding analysis and analysed using Gephi, a social network analysis open-source software. The analysed results were ranked based on the categories of the centrality parameter, which highlights the extent of collaboration of key stakeholders in the network. The findings of this study indicate Malaysian Civil Defence (APM) and local government have high degree and betweenness centralities in the network. The number of private sectors active in disaster response was minimal, as were their centralities within the network, where they ranked last in every network measure. Rural communities and victims had lower betweenness centrality scores showed they had low network influence. NGOs are less involved in disaster response but are more involved in relief efforts such as cleaning muddy houses, recruiting medical and non-medical volunteers to help flood victims, distributing cleaning and healthcare supplies, and giving meals.


Introduction
Over the last two decades, Malaysia has seen several catastrophic floods with increased intensity and frequency. In late 2020 and early 2021, in Peninsular Malaysia, torrential rains caused severe flooding in many east coast states. The heavy rains in the South China Sea were caused by the Northeastern Monsoon winds in the region. Thousands of people were

Setting
This study involved political, academic, NGOs, community and disaster management agencies and stakeholders in Temerloh, Pahang. Temerloh is a town in Central Pahang, Malaysia. Excessive soil erosion and generation of sediment load from the upstream of Sungai Pahang Basin lead to an increased tendency of flooding. Most residential areas are located at the lowland and the flood plain region and amidst the bad irrigation system especially in big residential areas. Temerloh faces a larger magnitude of floods because it is located at the confluence of main tributaries (Sg. Jelai and Sg. Tembeling) in the mid-stream area.
The Temerloh district activates its Control Post on Scene (PKTK) and Disaster Operation Controlling Centre (PKOB) during disaster emergencies, as it did in response to the 2020-2021 Malaysia flood [9]. Emergency service providers and community stakeholders play essential roles in helping communities recover from disasters; however, their function and mechanism in collaborating and sharing resources to meet community needs are underresearched. This study contributes to the resilience planning efforts of rural communities. To evaluate organizational roles and collaborations, we conducted a manual content analysis.

Content analysis process
This study draws on data from the content analysis of English and Malays news sources and archives in Astro Awani, Berita Harianin, social media and and website using search query ("mission flood relief Malaysia Januari 2021") and (Temerloh and Pahang), situation reports from by Pahang State government departments, Malaysia Red Crescent (MRC) and Tzu Chi between January 8 and February 15, 2021.
Node and tie are two fundamental concepts of social network theory. A node represents various 'actors', i.e., people, organizations, or countries, acting within the context of an event or relationship. A tie is a term that refers to social connections that exist between any two nodes [10]. The types of ties could include similarities, social relations, interactions, and flows. The content analysis identifies nodes and ties in each article's disaster response actions, communications, interactions, information exchange, or resource flow. The people and major organizations that participated in the response operations were identified as node and the interactions between those organizations were identified as ties through content analyses. In this way, organizations collaborating with another organization in Malaysia 2020-21 flood response were identified [11]. The Gephi (Version 0.9.2) social network analysis programme was used to analyse the data acquired from the content analysis. The program contains several network analytic routines (e.g., centrality measures, dyadic and network-level measures) [12].

Organizational roles
Among the 46 organizations on the actor list, the majority of participating organizations were federal/state/local government (59.57%), followed by non-profit (27.66%), and others stakeholders such as private sector and educational institutions (12.77%). Table 1 shows the stakeholders reported roles and response activities based on news and situation reports.   Figure 1 presents collaboration networks of the 46 organizations in emergency response and recovery, co-sponsoring relief missions, and supporting victims in Temerloh. The node size reflects their degree centrality within the network. The bigger the node, the more links it has.

Network graph and structure
A line between two organizations shows collaboration in related task toward achieving a common goal. Thus a more significant number of lines within the network indicates a denser collaboration pattern. Some of the descriptive network statistics for the network were obtained from Gephi as shown in Table 2. In the network, 46 organizations that participated in disaster response were identified, of which 1 (0.02% percent) was an isolated node, i.e., an independent flood victim that did not collaborate with any other actors during the disaster response (Fig. 2). The indegree is the number of incoming links and out-degree is the number of outgoing edges as shown in Table 3.
The degree to which a network is connected to the broader structure is referred to as its connectedness. Network density measures the proportion of potential linkages in a network that are connected. The calculation of network density is equal to known connections divided by maximum possible connections. (an ideal, fully connected network would have a density of 1.00) The result shows a network density of 0.032 and an average network degree of 1.679. To get the average degree for a graph, is the number of edges divided it by the total number of nodes in the graph. The average path length of the network is 2.52 based on the statistics, which means that to meet another organisation, a particular organisation must navigate roughly two organisational linkages. The average path length is the sum of all shortest paths between all nodes and divide number of all possible paths. On the other hand, the network diameter is 6, which means that the longest of all computed shortest pathways connecting all pairs of nodes in this network is 6. The longer the length, the weaker the connection which implies overall network structure is relatively constrained, preventing a specific organisation from effectively reaching other organisations via a shorter path.
The connectivity of a network is a measure of how well-connected the overall network structure is. The network density of a network indicates the proportion of potential connections that are connected. The results show a network density of only 0.032 (3.2%). These findings imply that 96.8% of the network's potential connections are not realised, and node connections are relatively limited on average. This finding is consistent with one study on social network analysis of disaster risk reduction in Asia and the Pacific which produced a similar result of 3% network density [13]. The low density can be explained by the lack of collaboration ties among the organizations participating in the disaster response network, which is typically lacking in most emergency management networks [14,15].

Degree centrality
Centrality measures are used to examine the most central actors. The top 10 most-central actors in the network were ranked: these were the actors that had the most connection in the network and an immediate influence on many other actors participating in disaster response.
As it can be seen in Table 4, APM and Temerloh District Council scored the top 3 in terms of the number of interactions it had during the flood disaster response. One non-profit organisation made it to the top 10, i.e., the Malaysian Red Crescent. As illustrated in Table 5, using the betweenness centrality measure, the top 7 gatekeeping government agencies and departments were positioned to broker connections between groups who could influence the flow of information among communities or organizations. Table 5 shows that APM has the highest betweenness centrality means it play as an important bridge among organizations in the network. JAKOA ranked third highest in the highest betweenness centrality, making JAKOA staff the most important government agencies as gatekeepers who could work well with Orang Asli families. Gatekeepers benefit significantly from pre-existing connections, indigenous knowledge and trusting relationship, all of which will help to engage the Orang Asli communities [15].
Malaysian Red Crescent (MRC) scored the 5 th highest in its role as broker. The Malaysian is the only non-profit organization that is formally assigned disaster emergency functions and roles in Malaysia National Security Council Directive 2.0. MRC hands out cash assistance to the most vulnerable groups and coordinates with the disaster management of the national headquarters, the International Federation Red Crescent (IFRC) project coordinator and members of the regional and local disaster response team [9].
The betweenness centrality of the RELA, MPKK, PDRM, JKR, and KIR, on the other hand, is zero, implying that neither of these parties has the authority to connect the other organizations. Hence, they are unable to broker opinions or influence information flow. All of these parties should collaborate with other parties to share data and information to carry out effective emergency response.

Conclusion and Recommendation
A social network analysis of stakeholders involved in disaster response in Temerloh, Pahang, showed that governmental, non-governmental organizations are active in disaster response operations. The Malaysian Civil Defence (APM) played a crucial role in disaster response. APM and local government agencies occupy central positions in the network, as indicated by their high degree and betweenness centralities. Private sector participation in disaster response was limited in terms of the number of private sectors involved in disaster response and their centralities within the network, where they rank last in every network metric. As indicated by their lower betweenness centrality scores, rural communities and victims had less power to influence the network. Such limited influence is consistent with research on the role of local communities in disaster management, where communities are excluded in the top-down rather than bottom-up decision-making processes. Rural communities in flood-prone areas can reduce their financial damage by adopting communitybased disaster management approaches and self-protective behaviour, minimising the need for government assistance, supporting self-recovery, and building back safer through disaster risk reduction programmes and training.
NGOs groups have less collaboration ties in the areas of disaster response but have the highest involvement in relief mission by co-sponsoring relief activities, cleaning muddy homes, mobilizing medical and non-medical volunteers to aid flood victims, deploying hygiene, cleaning and healthcare kits and providing food in supporting the victims (Figs. 3  and 4). Further qualitative research need to be conducted to understand NGO roles and barriers and challenges in forming partnership with other NGOs, international NGOs and government institutions in the context of disaster response and recovery.
The flood relief efforts were short-lived. According to situation reports and newspaper publication, many organizations collaborated in flood response operations in the first few weeks with a subsequent drop in reporting. The areas with less collaboration were longerterm disaster recovery and development stages that represent areas to strengthen within disaster management practice such as limited access to resources to rebuild homes and replace damaged house contents.
Findings from one specific rural district may not generalise to another part of the country. Therefore, we did not use statistical techniques in this study for comparing two or more groups. Analysis of social networks on disaster response should include more