The evaluation of community disaster resilience is of great practical importance for building low-risk, sustainable, and disaster-resistant cities. With 12 communities in Luoyang as the objects, this paper adopts entropy weight TOPSIS and obstacle diagnosis to study the community disaster resilience of Luoyang from seven dimensions, such as demographic characteristics, economic development, and infrastructure. The results of the study are as follows: (1) the community disaster resilience of Luoyang (0.48) is at a medium level. Community capital is the main influencing factor of community disaster resilience. Government governance, community capacity, and community intelligence are the components that need attention in the construction of Luoyang’s community disaster resilience. (2) The community disaster resilience of Luoyang presents a decreasing trend from rural to urban areas. Moreover, communities with high disaster resilience are less than communities with low disaster resilience. (3) The obstacle to community disaster resilience of Luoyang focuses on population, economic development, and infrastructure. In addition, community trust, community dependence, popularization and intellectualization of disaster prevention information, and disaster information sharing also significantly restrict the construction of Luoyang community disaster resilience. (4) According to the results of sensitivity analysis, the entropy weight TOPSIS evaluation results are less sensitive. Moreover, changing the weight value, weight method, and evaluation method will not lead to major changes in the rankings.

1. Introduction

As the basic unit of social organization, community is the main body of disaster risk and the foundation of disaster prevention and mitigation work [13]. In 2015, the United Nations Sustainable Development Summit adopted Transforming our world: the 2030 Agenda for Sustainable Development, causing the building of inclusive, safe, resilient, and sustainable cities and human settlements in the 2030 Sustainable Development Goal [4]. In the same year, the United Nations Office for Disaster Risk Reduction (UNISDR) adopted the Sendai Frame, in which the construction of community disaster resilience is considered to be a priority for global disaster reduction efforts in the next 15 years [5].

At present, community-based disaster prevention and reduction (CBDRR) is a world-recognized important way to improve disaster response capacity and reduce disaster risks and losses [6, 7]. In response to severe disaster risk challenge, the United States Federal Emergency Management Agency released “A Whole Community Approach to Emergency Management: Principles, Themes, and Pathways for Action” in 2011, which emphasizes the establishment of partnerships involving various stakeholders and encourages community to reduce the chance of disasters and the loss of life and property [8]. The Great Hanshin Earthquake in 1995 brought huge losses to Japan. This painful lesson promoted the transformation of Japan’s disaster management model. The community-based disaster risk management model (CBDRM) replaced the top-down government-based disaster risk management model, which implemented the “disaster prevention and welfare community business plan,” and advocated the cooperation between citizens, plan promoters, and the municipal government [9]. In 2019, Canada promulgated its first emergency management strategy “Canadian Emergency Management Strategy: Towards a More Resilient 2030” under the basic principles of the EM Framework and the United Nations Sendai Framework [10].

Compared with Western developed countries, it is more urgent to build community disasters resilience in China. China is one of the countries most affected by various disasters. According to relevant statistics, more than 70% of cities and 50% of population in China are located in areas with high incidence of meteorological, earthquake, geological, and marine disasters. Over the past three decades, disasters have caused an average of five deaths per million people per year. Moreover, the direct economic losses caused by disasters account for 2.25% of GDP [11]. In general, disasters have had a serious impact on China’s social stability and sustainable development. The work of community disaster prevention and mitigation has always been highly valued by the Party and the government. In 2020, the General Secretary Xi Jinping pointed out when inspecting the prevention and control of the COVID-19 epidemic in Hubei: “This epidemic prevention and control work not only shows the important role of urban and rural communities but also exposes the shortcomings and deficiencies of grassroots social governance.” In addition, he proposed to “establish the awareness of full-cycle management,’ to explore new methods of modernization governance for megacities, and vigorously improve the urban and rural grassroots governance system” [12].

Since 2007, the National Disaster Reduction Committee of China has organized the establishment of comprehensive disaster reduction demonstration communities across the country to improve the comprehensive disaster resilience of urban and rural grassroots communities [13]. Although this work has produced a series of results over the past decade, the relevant assessment and construction standards are still dominated by postdisaster response. In addition, it relies on government-led, top-down disaster prevention and relief, but rarely on predisaster prevention and disaster prevention and control. The community’s own defense capabilities and the awareness and ability of community residents and external stakeholders to participate in governance independently need to be strengthened [14].

In recent years, the concepts of community resilience and resilient community have been proposed against the institutional and theoretical background of the downward focus of national governance [15]. The utilization of community resources to absorb and mitigate disasters has become a frontier field and a research hotspot in the global academic community [1618]. Holling first introduced resilience into the field of ecology and conducted pioneering research on ecological resilience [19]. During this period, a large number of viewpoints and insights on ecological resilience emerged. Timmerman first introduced the concept of resilience into the field of disasters and linked it to vulnerability. Furthermore, it was believed that resilience was the ability of a system to resist, adapt, and recover from the stress of environmental change [20]. The United Nations International Strategy for Disaster Reduction (UNISDR) promoted the concept of resilience as a measure of the ability of a system or community to resist or modify natural disasters [21]. Regarding disaster resilience, different scholars have proposed different definitions from different perspectives. However, the common point is to understand disaster resilience as the ability of a system to cope with changes and disturbances [22, 23]. It is this core connotation that promotes the expansion of the concept of resilience from the early field of ecology to societies, communities, families, and individuals [24, 25].

Resilient communities are less prone to disasters than less resilient places. In order to test this hypothesis, it is critical to understand how to determine, measure, and enhance resilience [16]. There are many evaluation models for disaster resilience in the world. Bruneau et al. proposed the “4R” characteristic model from the perspective of engineering theory [26]. Based on the research proposed by Bruneau, Renschler et al. combined the “4R” characteristics of disaster resilience with the dimensions of resilience (technology, organization, society, and economy), and proposed the “PEOPLES” community disaster resilience evaluation model. Unlike the disaster resilience framework from an engineering perspective, this model highlights the importance of socioeconomic factors [27]. Norris et al. proposed a conceptual model of disaster resilience networking [17]. Sherried et al. conducted an empirical analysis on the basis of this conceptual model [28]. Cutter et al. developed a geographic-based framework for disaster resilience (DROP), which predicts changes in disaster resilience by measuring the baseline level [16]. Cutter et al. constructed the BRIC disaster resilience framework based on the inherent disaster resilience of the DROP model and evaluated the level of disaster resilience in the southern United States from six dimensions of society, economy, institutions, infrastructure, community capital, and ecological capital. Due to the large scale of the study area, different ecological environment characteristics, and inconsistent ecological data, the evaluation of the ecological resilience of the study area was discarded [29]. In terms of measurement methods, the evaluation model of community disaster resilience can be divided into two categories, namely, measurement method based on objective index and measurement method based on subjective perception. For the measurement method based on objective index, the starting point is that the community disaster resilience comes from the inherent characteristics of the community. Couplings and interactions between community characteristics constitute the community disaster resilience. Currently, most studies use this approach in small-scale areas such as communities. Nevertheless, it is difficult to obtain some public statistics, and there is a lack of in-depth research on the motivation of community members.

Based on the measurement method of subjective perception of actors, it is committed to the development of community disaster resilience scales. The CCRAM scale of Leykin et al. and the CART scale of Pfefferbaum et al. are more representative. CCRAM contains 28 items, which are measured from six aspects of leadership, collective action ability, preparedness, community dependence, social trust, and social relations [30]. CART is measured from four aspects, which are connection and care, resources, change potential, and disaster management [31]. Community disaster resilience is a multidimensional concept involving social, economic, institutional, infrastructure, ecological, and other elements. In recent research on community disaster resilience, one of the important outcomes is the challenge recognition of evaluating community disaster resilience due to the complex interactions among individuals, communities, and local societies. When destructive events occur, the community capital and collective action ability of individuals and families are the key to achieving disaster resilience [2933].

At present, domestic research on community disaster resilience is still in its infancy [34], mainly focusing on conceptual analysis [35, 36], constituent elements [37], and the introduction of foreign disaster resilience evaluation scales [38, 39]. There is a lack of attention and research on the evaluation model and empirical analysis. On the basis of clarifying the connotation and elements of community disaster resilience, this paper attempts to design a set of scientific, reasonable, and systematic evaluation index system of community disaster resilience in line with China’s national conditions. On the basis of constructing the index system, the measure method based on the subjective perception of actors is used to evaluate the disaster resilience level of the community. In order to obtain the data required for the study, 12 communities in Luoyang City were selected as the research objects, and the disaster resilience questionnaire of community residents was designed. Through questionnaires and interviews with neighborhood committee members, the subjective perception data of community residents on the level of community disaster resilience were obtained.

Luoyang is a city in Henan Province and a municipal administrative unit in China. The community is the most grassroot level organization in China at the administrative level. A residents committee shall be set up to administer the community. The community population is between 1500 and 5000, among which there are 3126 communities (administrative villages) in Luoyang City and 956 communities (administrative villages) in the main urban area. After obtaining the research data, the methods of entropy weight TOPSIS and obstacle diagnosis were used to evaluate the disaster resilience of Luoyang communities. Based on the influencing factors of community disaster resilience, the corresponding improvement strategies are put forward to improve community disaster resilience and provide strong support for the sustainable development of social economy.

The entropy weight TOPSIS is a comprehensive evaluation method, which uses the entropy weight method to determine the weight of each index and uses the TOPSIS method to quantify and rank the comprehensive disaster resilience level of the community. This method has been widely used in urban research [40], sustainable development research [41], tourism research [42], supply chain research [43], food security [44], industrial design [45], and other fields [46, 47]. There are also many applications in the field of disaster science. Pan and Li used entropy weight TOPSIS to study the disaster vulnerability of 11 coastal provinces in China, obtained the development characteristics of vulnerability in coastal areas, and explored the factors affecting vulnerability [48]. Han and Zhang studied the coupling coordination relationship between urbanization and geological disasters based on the entropy weight TOPSIS and the coupling coordination degree formula [49]. In addition, Liu et al. used the entropy weight TOPSIS-PCA method to evaluate the risk of urban waterlogging disaster in China [50]. After comprehensive evaluation, the obstacle degree model is used to explore the weak index of the disaster resilience of each community, which is the key influencing factor to reduce the ranking of community disaster resilience.

Finally, the sensitivity analysis is carried out. The comprehensive evaluation results will be affected by the weight coefficients and evaluation methods. Moreover, different weight coefficients and evaluation methods will lead to different disaster resilience ranking. Sensitivity analysis is the last step of comprehensive evaluation. In the research on sustainable supplier selection, Puška et al. used the PIPRECIA method to determine the weight of each indicator. Then, the MABAC method was used to determine the ranking. Finally, a sensitivity analysis was carried out. By changing the evaluation method, the ranking order was analyzed. In this part of the study, the evaluation methods of MARCOS, SAW, ARAS, and TOPISI were used. Then, the weight value of each indicator was changed. The weight value of one indicator was increased by eight times, while the other weights remained the same. 19 scenarios were formed and the ranking changes of the 19 scenarios were compared [51]. In the same year, a similar approach was adopted by Puška in his study on the potential of sustainable rural tourism [52]. Kizielewicz et al. analyzed the sensitivity of the results in the supplier selection study through the WS similarity coefficient and the weighted Spearman correlation coefficient. Equal weight, entropy, and standard deviation methods were used to determine the weight of the criteria, and COMET, TOPSIS, and SPOTIS were used to determine the program ranking [53]. Yazdani et al. conducted a similar study. This paper is divided into two parts for sensitivity analysis [54]. The first is the weight sensitivity analysis. By changing the weight determination method and weight value, the ranking change of disaster resilience evaluation results is analyzed. The weight method uses three methods, namely, CRITIC, equal weight, and principal component analysis. The disaster resilience rankings were obtained and compared with the results of the entropy weight method. The weight value is changed by changing the weight of each index, and the calculated results are compared with the results under the same weight. In the second part of sensitivity analysis, the evaluation method was changed. VIKOR, MABAC, ARAS, gray correlation, and SAW were used to calculate the ranking of disaster resilience, and the results were compared with that of TOPSIS method. In order to further analyze its sensitivity, this paper uses the Spearman rank correlation coefficient to calculate the ranking correlation of the results. If the correlation is significant, it indicates that the results have good stability.

2. Community Disaster Resilience Evaluation Index System

2.1. Community Disaster Resilience Model

At present, a considerable part of the research results of community disaster resilience in the United States come from the derivatives of the groundbreaking achievements made by Norris et al. [17], Cutter et al. [16], Pfefferbaum et al. [30], and Cutter et al.[55]. It is not surprising that many concepts and indicators are combined in a certain way. Therefore, this paper adapts the community comprehensive disaster resilience model (Figure 1) based on the DROP model proposed by Professor Cutter of the University of South Carolina in 2008 and the disaster resilience resource network model of Norris et al. [17]. The related concepts of the model are derived from the DROP model and the disaster resilience resource network model. There are four key concepts, namely, community antecedent conditions, disaster absorption ability, adaptive disaster resilience, and community resource. The definition of the initial state adopts the viewpoints of DROP on inherent disaster resilience and inherent vulnerability. As a pair of intersecting concepts, inherent disaster resilience and inherent vulnerability are the products of the interaction of social, natural, and built systems in a community. Interactions between preconditions and hazard characteristics, including the frequency, duration, and risk of hazards, have a direct impact on communities. Disaster absorption ability can reflect the community’s ability to absorb disaster impact using community resources (not just community emergency response measures), which is similar to the discussion on crisis and resilience in the disaster resilience resource network model. When the disaster impact does not exceed the disaster absorption capacity, disaster resilience is generated and communities achieve a high level of recovery. Adaptive disaster resilience refers to the adaptive network of a community in the face of disturbances. When the disaster impact exceeds the disaster absorption capacity, the community adaptive disaster resilience plays a role. Collective efficacy, social learning, and improvisation based on community resources have an impact on community initial states and recovery levels. Community recovery levels are on a low-to-high continuum. The recovery level of adaptive disaster resilience is not fixed but depends on the resource level of community. The definition of community resources differs from the disaster recovery resource network model. In the disaster resilience resource network model, population health is viewed as a result of postdisaster recovery, which is not incorporated into community resource. This paper uses a broader definition of resource. Community resources are the sum of resources that facilitate the community disaster resilience and recovery process.

In recent years, domestic and foreign research fields have reached a certain consensus on the connotation, components, and characteristics of community disaster resilience. For example, community disaster resilience includes social, economic, institutional, infrastructure, ecological, and other elements [56, 57]. Based on the research of domestic and foreign scholars such as the DROP model and the disaster recovery resource network model, there are seven types of community resources, which are demographic characteristics, economic development, infrastructure, community governance, community capacity, community capital, and community intelligence [16, 17, 29, 5863]. The BRIC model proposed by Cutter et al. excludes ecological capital from community resources because of its large spatial variability, which makes it inconvenient to compare the disaster recovery capabilities of communities between regions [29]. Furthermore, information communication is called community intelligence, as the use of technologies such as the Internet, big data, and the Internet of Things in recent years has brought about major changes in the way of community dissemination. Relevant studies have shown that community intelligence has a significant positive effect on community communication ability [64, 65]. The connotation and constituent elements of these seven types of resources will be explained in detail below.

2.2. Construction of Community Disaster Resilience Index System

On the basis of evaluating the community disaster resilience model, community disaster resilience index system is constructed. The secondary indicators consist of seven types of community resources. The selection of variables is based on the understanding of the connotation of community resources, following scientificity, representativeness, feasibility, and data availability. Based on literatures [16, 17, 37, 59, 60, 63, 6578], an evaluation system of community resilience in China is constructed, comprising seven dimensions and 36 variable indicators (Table 1).

For the convenience of analysis, this paper abbreviated the acronyms of secondary indicators as D (demographic characteristics), E (economic development), I (infrastructure), G (government governance), C (community capacity), CC (community capital), and CI (Community Intelligence). Variables are named after secondary indicators plus numbers. For instance, education level is named D1, and key infrastructure is named I3. Each variable corresponds to one or more questionnaire items. If it corresponds to multiple questionnaire items, it is named with multiple letters. For example, vulnerable group proportion is named D3-D4.

(1) Demographic characteristics (D) include population vulnerability and human capital. The measurement of population vulnerability with negative indicators involves two indicators of migrant worker proportion (D5) and vulnerable group proportion (D3-D4). The human capital with positive indicators is measured by three indicators of education level (D1), youth proportion (D2), and disaster risk awareness (D6-D7). (2) Economic development (E) describes the professional structure and income of community members. The measurement of community economic development involves indicators such as per capita annual income (E1), income stability (E2), income source diversity (E3), disaster insurance (E4), social security (E5), and cash borrowing capacity (E6). (3) Infrastructure (I) represents the physical environment in which people live, including structures, shelters, key infrastructure, and lifeline systems (water, electricity, gas, and integrated pipe gallery). Community infrastructure is measured in terms of shelters (I1), medical service capabilities (I2), key infrastructure (I3), emergency sign indication capabilities (I4), and emergency passages (I5). (4) Government governance (G) reflects the government’s emphasis on community disaster prevention and mitigation. The measurement of community government governance involves indicators such as emergency material reserve (G1), disaster publicity and education (G2-G3), capital investment (G4), emergency plan (G5), risk assessment (G6-G7), and postdisaster psychological support (G8). (5) Community capacity (C) represents the collective action capacity of community, namely, the community’s ability to generate collective effectiveness and improve postdisaster recovery through economic resources, community intelligence, community capital, and other resources. The measurement of community capacity involves indicators such as community public participation (C1-C2), problem assessment (C3), learning ability (C4), community plan management (C5), resource mobilization ability (C6), and volunteer spirit (C7). (6) Community capital (CC) is an indicator of actual or potential resources in community social relations that can promote collective action and achieve common goals. The measurement of community capital involves indicators such as social network (CC1–CC6), community dependence (CC7-CC8), community trust (CC1-CC5), social network (CC1–CC6), and community trust (CC1–CC6). The same name is because the corresponding questionnaire items are the same. (7) Community intelligence (CI) is the embodiment of community communication capabilities in the context of smart cities and smart communities and highlights the role of information technology in community disaster resilience. The measurement of community intelligence involves grid management (CI1), disaster information collection (CI2), disaster information sharing (CI3), popularization and intellectualization of disaster prevention information (CI4), community governance intelligence (CI5), and other indicators.

3. Methodology and Data

3.1. Introduction to Research Area

Luoyang is located at east longitude of 111.8′ to 112.59′ and north latitude of 33.35′ to 35.05′, spanning both sides of the middle reaches of Yellow River, in the north-south climate transition zone. Its terrain is diverse and complex, especially mountains and hills. The overall terrain is high in the west and low in the east. Affected by the monsoon, the seasonal distribution of precipitation is uneven and the interannual variation is large. As one of the geological disaster-prone areas in Henan Province, there have been many serious disasters such as floods and droughts in the past. In 2021, Henan Province encounters rare heavy rainstorms and heavy floods. The number of dead and missing people in the province will reach 398, including 380 in Zhengzhou, 10 in Xinxiang, and 2 in Pingdingshan, Zhumadian, and Luoyang each, 1 from Hebi City and one from Luohe City. The severe disaster situation in Henan shows us the vulnerability of cities in the face of catastrophic disasters. How to fully absorb the lessons of this disaster and study the ways and methods for cities to reduce disaster losses in the face of disasters is an important issue facing Henan Province.

It is an important birthplace of Chinese civilization, a regional central city, and a subcentral city of the Central Plains urban agglomeration. It has jurisdiction over seven counties, seven districts, two national-level development zones, two provincial-level development zones, and 18 provincial-level industrial clusters. The total area is 15,200 square kilometers, of which the urban area is 2,274 square kilometers. The city’s resident population is 7.069 million, of which the urban population is 4.657 million, and the urbanization rate is 65.88%. The city map of Luoyang is shown in Figure 2. The stable economic development of Luoyang is of great significance to the sustainable development and rise of Henan. Based on this, this paper studies the disaster resilience of Luoyang communities, aiming to provide reference for disaster resilience construction in Henan Province as well as the central region of China.

3.2. Data Sources

This investigation was conducted in collaboration with the Luoyang Emergency Management Agency, and the emergency management agencies of each district and county in Luoyang. The purpose of this study is to obtain the data of community residents’ perception of community disaster resilience. The data were collected by means of questionnaire survey and field survey among neighborhood committees and community residents. The design of questionnaire was based on the China’s disaster resilience index system, with reference to mature questionnaires at home and abroad; see Annex 1. Please refer to Table 2 for the setting and distribution of questionnaire items.

The specific investigation process is as follows. A preliminary survey was conducted in September 2020, namely, a small-scale online questionnaire in Luoyang. A total of 143 questionnaires were returned. Through the analysis of the presurvey data, the contents and indicators of the questionnaire were modified to increase the scientific validity and rationality. In October 2021, the formal investigation was carried out. Twelve communities in Luoyang were randomly selected as the research sample sites. These communities are scattered throughout Luoyang (Figure 3). Because Yanshi District has just been included in Luoyang, it was not included in this survey. Mengjin District was formed by the merger of Mengjin County and Geely District. Because this survey is mainly concentrated in the urban area, it is only one community in Mengjin District. A total of 649 questionnaires were returned. After eliminating 70 invalid questionnaires, a total of 579 valid questionnaires were retained, with an effective rate of 89%. The questionnaire survey covered a wide range of communities in Luoyang (Table 3). More specifically, 151 questionnaires came from the old city, 119 from Luolong District, 84 from Chanhe District, 50 from Xigong District, 58 from Western District, 58 from High-tech Zone, 59 from Mengjin District, and 20 from Yibin District (among them, Yibin District and High-tech Zone were administratively divided into Luolong District). Table 3 shows the sample distribution. In terms of gender composition, the ratio of male to female is 41.6% and 58.4%. In the composition of education, high school and above account for the majority (87%). In terms of the age structure, young people are the main group, in which the proportion of population aged 16–59 reached 88%. Therefore, it can be concluded that the sample is highly representative.

Based on the questionnaire data, the collinearity and reliability of the index system were tested by SPSS 26.0 statistical analysis software. AMOS22.0 was used for confirmatory factor analysis of questionnaire validity. In the collinearity test, the linear regression equation with one indicator as the dependent variable and the remaining indicators as the independent variables was built. The larger the variance inflation factor VIF, the more linearly this indicator can be represented. The criterion is VIF < 10 [79]. According to the collinearity test results, all indicators VIF are less than 10 (Appendix 2). Therefore, there is no need to review the indicators. The reliability test of 28 scale items in the questionnaire showed that the Cronbach α coefficient of the scale as a whole and each dimension was greater than 0.9, indicating that the overall scale had a high reliability (Appendix 3). The validity test adopted the confirmatory factor for CFA analysis. Table 4 shows the results of model fitness. The absolute fitting index is slightly larger than the recommended value. The value-added fitting index and the parsimonious fit index are within the recommended standard range. The model fitness is acceptable. The AVE of each dimension of aggregate validity of the scale was greater than 0.7. CR was all greater than 0.9, which was within the acceptable range (Appendix 4). As a result, the validity of the model is verified.

3.3. Entropy Weight TOPSIS

The entropy weight TOPSIS was employed to measure community disaster resilience. Specifically, the entropy weight method was used to determine the weight of each index, and the Topsis method was applied to quantitatively rank the community disaster resilience. The entropy weight method is an objective weighting method, which was originally derived from the concept of thermodynamics in physics, and then introduced into information theory to reflect the chaos in the system. Currently, it is used in many fields [80, 81]. Compared with subjective weighting methods such as the analytic hierarchy process, the entropy weight method can reduce the interference of subjective human factors and reflect the indicator information more accurately [82]. The TOPISIS method evaluates the measurement index according to its distance from the optimal solution and the worst solution. If the evaluation object is nearest to the optimal solution and furthest from the worst solution, it is optimal. Otherwise, it is not optimal. Each index value of the optimal solution reaches the optimal value of each evaluation index, and each index value of the worst solution reaches the worst value of each evaluation index. The closer the measurement index is to the optimal solution, the higher the score [83]. The measurement process is as follows:(1)Standardization of indexes is as follows:The benefit indexes are The cost indexes are Supposing there are n community and m pieces of indexes in the index system, Xij is the jth index’s value in the ith community ((i = 1, 2, …, n), (j = 1, 2, …, m)).(2)Calculation of index’s entropy can be given as(3)Calculation of index’s entropy weight can be given as(4)Determination of the weighted decision matrix can be given asThe weighted decision matrix is determined by the normalized decision matrix multiplication with weights of indexes.(5)Determination of the ideal solution can be given as follows:The ideal solution isThe negative ideal solution isThe ideal solution is composed of the optimal value of every attribute from the weighted decision matrix, and the negative ideal solution is composed of the worst value of every attribute from the weighted decision matrix.(6)Calculation of the distance can be given as (7)Calculation of the relative degree of approximation can be given as

The evaluation object is ranked according to the value of the relative degree of approximation. The bigger the value is, the better the evaluation object is.

3.4. Obstacle Diagnosis

For the evaluation of community disaster resilience, it is necessary to establish a scientific community disaster resilience evaluation index system. It is of practical significance to identify the obstacle factors in the construction of community disaster resilience and put forward reasonable countermeasures for improving community disaster resilience. Therefore, this paper introduces obstacle diagnosis into the evaluation of community disaster resilience to evaluate and explore related indicators and explore the main constraints [84, 85]. The measurement process is as follows:where Qij is the jth index’s obstacle factors in the ith community ((i = 1,2, … n), (j = 1,2, …, m)).

4. Results and Discussion

4.1. Evaluation of Community Disaster Resilience

Table 5 presents the index scores for each dimension of disaster resilience of 12 Luoyang communities based on the entropy weight TOPSIS method. The average score of community disaster resilience index of 12 Luoyang communities was 0.48, which was generally at a medium level. The score of Chundu community was the lowest (0.17), and the score of Baimaqun community index was the highest (0.81). With the difference of 4.76 times, it indicates that there is a large difference in the overall disaster resilience of Luoyang communities.

The average score of each dimension of community disaster resilience ranged from 0.46 to 0.52. Among them, Luoyang communities are in the medium range in seven dimensions, such as demographic characteristics. By digging into the standard deviations of each dimension, the dimensions with larger standard deviations include community capacity, community capital, and community intelligence. The difference between the highest score and lowest score is more than eight times. According to the weights determined by entropy weight TOPSIS, community capital has the highest weight, which is the most important factor affecting disaster resilience. In addition, the weights of government governance, community capacity, and community intelligence are at a moderate level, but the difference is large. There is a need to concentrate on building the disaster resilience of Luoyang community.

Through SPSS 26.0, the results of entropy weight TOPSIS evaluation are systematically clustered (Table 6). 12 Luoyang communities are divided into three categories, which are high disaster resilience, medium disaster resilience, and low disaster resilience. The results are shown in Table 5. Among them, four communities have high disaster resilience, seven communities have low disaster resilience, and one community has medium disaster resilience. Therefore, the community disaster resilience generally presents a “pyramid” distribution. Figure 4 shows that of the four communities with high levels of disaster resilience, two are more than 10 km away from the city center and one is more than 5 km away. The medium-level disaster resilience communities are located within 5 kilometers to 10 kilometers away from the city center. Among the communities with low disaster resilience, all are within 5 km, except for those within 5–10 km and 5 km of the city center. Yao’ao community is a new rural community far from the city center. Sima community and Nanhua New Village community are located in the urban-rural integration area. White Horse Group community and Zhongjian community are unit communities located on the fringes of the city. These communities have high community disaster resilience, which is consistent with the evaluation results of Sichuan community disaster resilience by Zheng et al. The community disaster resilience is decreasing from rural areas to urban areas [38].

4.2. Analysis Of Obstacle Degree

On the basis of community clustering, the obstacle diagnosis model is used to calculate the obstacle degree scores of communities with high, medium, and low disaster resilience. Among them, the top five scoring indicators are considered to be the key constraints to improve community disaster resilience. The results are shown in Table 7. For ease of understanding, the indicators corresponding to the codes in Table 7 are listed. D1 is education level, D2 is youth proportion, D5 is migrant worker proportion, D7 is disaster risk awareness, E1 is per capita annual income, E2 is income stability, E3 is income source diversity, E4 is disaster insurance, I2 is medical service capabilities, and G2 is disaster publicity and education.

The biggest constraint to building communities with high disaster resilience is medical service ability. Medical service is part of social public services. Studies have shown that community medical services are best provided within a 15-minute walk of residents. Through a survey of Luoyang communities, it can be found that the average time for residents to reach the nearest medical institution is 15–30 minutes. That is to say, Luoyang communities have weak emergency medical rescue capabilities. Community medical institutions are of great significance to the lives of community residents, which is mainly reflected in the guarantee of life, health, and quality of life in daily life and emergency rescue when disasters occur [86, 87]. In accordance with the characteristics of the population, employment density, and age structure, the community can set up emergency medical service centers or regional comprehensive support centers (integrating nursing care, housing, medical treatment, prevention, and rescue), community health construction centers, as well as health and epidemic prevention liaison stations. It is recommended that these institutions and facilities serve one or more 15-minute community living circles [88, 89].

The diversity and stability of income sources are the limiting factors for community with high disaster resilience. Disaster insurance and per capita annual income are the limiting factors for communities with moderate disaster resilience. Disaster insurance and income source diversity are the limiting factors for communities with low disaster resilience. Meanwhile, the economic level affects the demographic characteristics of community residents. When the income of residents is unstable, there is an outflow of community population, which is manifested in the decrease in the proportion of youth population and the increase in the number of migrant workers [90]. As for the constraints imposed by economic development on the construction of community disaster resilience, the employment of community residents should be guaranteed. For instance, vocational skills training for low-income and unemployed persons is carried out [91]. It is essential to improve the industrial structure and develop the vitality of enterprises, thereby enhancing the overall economic level of the community [92]. Community participation in residents disaster insurance is an effective means to reduce disaster economic losses. Through the study, it can be found that the participation rate of community disaster insurance is not high. Furthermore, the majority of insurance policies are individual health insurance and pension insurance. The effect of disaster insurance on property safety is ignored. Thus, government and private insurance companies need to work together to promote disaster insurance for communities and residents [93]. Another key constraint in building community resilience is disaster risk awareness, which reflects the level of community preparedness. The higher the level of preparedness, the greater the disaster resilience and absorption capacity. Therefore, community should pay attention to the cultivation of disaster risk awareness among residents [94].

The above analysis of the constraints on community disaster resilience mainly focuses on three dimensions: demographic characteristics, economic development, and infrastructure. However, this does not indicate that there are no constraints on dimensions, such as community capital. Through comparative analysis of disaster resilience of 12 communities, the main constraints of Chundu community, Oriental Jindian community, Hengda Oasis community, and Sunqi Village community are community trust, community dependence, emergency access, and emergency sign indication. The main constraints of Quanshun community, Tunnel Group community, and Longcheng Jiayuan community include community trust, community dependence, emergency access, emergency sign indication, popularization and intellectualization of disaster prevention information, and disaster information sharing.

5. Sensitivity Analysis

5.1. Weight Sensitivity Analysis

The results of the comprehensive evaluation will be affected by the weight coefficients and evaluation methods, and different weight coefficients and evaluation methods will lead to different rankings of disaster resilience levels. Commonly used weight determination methods include Criteria Importance Though Intercrieria Correlation (CRITIC), entropy weight method (EWM), coefficient of variation method (CV), and analytic hierarchy process (AHP). These weighting methods are widely used in different fields, and different weighting methods get different values. In order to analyze the sensitivity of entropy weight method to determine the weight, this paper first selects three methods to determine the weight Criteria Importance Though Intercrieria Correlation (CRITIC), coefficient of variation method (CV) and equal weight (EW). The calculation process can be found in the literature [53, 95, 96]. The TOPSIS evaluation method was used to determine the disaster resilience level, and the results were compared with the entropy weight method. Second, the weight value of each indicator was changed, and the changed weight value was set by itself in accordance with the practice of Puška et al. and the evaluation result determined by the equal weight method [51, 52]. In this paper, the value is set as 0.1 and 0.02 for other indicators. The indicators that change the weight value have a weight value five times higher than that of the other indicators. For example, the weight value of D1 is changed for the first time to 0.1, and the other indicators are valued to 0.02. For the second time, the weight value of D2 is changed to 0.1, and the remaining values are 0.02. This paper has 46 indicators. After calculating the ranking of 46 indicators with changed weights, the ranking results with equal weights are calculated again. In the case of equal weight, the weight value of each indicator is 0.022. After 47 operations, the results of the equal weight operations are used as reference to compare the changes in the rankings and observe the stability of the rankings. Finally, the Spearman rank correlation coefficient is used to test its weight sensitivity. Spearman rank correlation (R) can be used to determine the association measure between the ranks achieved in two different methods, which is usually used for multicriteria decision-making. If there is a large correlation between the rankings of two methods, it proves that there is little difference between the rankings determined by the two methods. The calculation formula can be found in reference [97].

According to Figure 5, the ranking determined under the CRITIC method is the most different from that under the EWM method. Sima community ranks fourth under the entropy weight method and ninth under the CRITIC method. The ranking has changed greatly. Nanhua New Village Community ranks under both methods. Two rankings are changed. The community with the largest ranking difference between the CV method and the EWM method is the Longcheng Jiayuan community, whose ranking has changed by two places. Moreover, the ranking of the rest communities has not changed much. The community rankings under the EW method and the WEM method has not changed much. With the exception of four communities, such as Hengda Oasis Community, which have changed by one place, the rankings of other communities has not changed. Combined with previous clustering results of disaster resilience levels, the rankings of Sima Community and Longcheng Jiayuan Community have reversed under the CRITIC method. The Nanhua New Village Community has also changed from an intermediate disaster prevention community to a low-level disaster prevention community, and the ranking results of the CRITIC method and EWM method are inconsistent. Although the rankings under the CV, EW, and EWM methods have slightly changed, their disaster resilience classification has not changed, showing good stability. The calculation results of Spearman rank correlation coefficient show that the rankings under the three weighting methods are highly correlated with that under the EWM method. Thus, the EWM method has lower sensitivity, and the calculation results under other methods have higher consistency. (Tables 810)

Figures 69 show the results of the ranking of disaster resilience after changing the weight value of each indicator. As can be seen from Figure 6, the ranking of disaster resilience changes greatly, indicating that indicators such as population characteristics and economic development are highly sensitive to weight changes. Especially, after the weight changes of indicators such as D1, E1, and E3, the ranking changes the most, and the sensitivity is relatively high. As can be seen from Figures 79, the change of the weight of each index does not cause a great change in the ranking of community disaster resilience. When the weight is equal, it is basically consistent with the ranking of disaster resilience, showing good stability. The results of Spearman rank correlation coefficient showed that the ranking of changing the weight value of each index was highly correlated with the ranking under the equal weight method, indicating that the TOPSIS method was less sensitive to changing the weight value.

5.2. Evaluation Method Sensitivity Analysis

In this paper, the TOPSIS evaluation method was used to determine the comprehensive score. In order to carry out sensitivity analysis, five methods including VIKOR, MABAC, ARAS, GRA, and SAW were used to calculate the disaster resilience level, and the results were compared with that of TOPSIS. VIKOR, MABAC, ARAS, gray correlation degree, and SAW are common comprehensive evaluation methods, and the calculation process is shown in the literature, respectively [98101].

The calculation results are shown in Figure 10. By comparing the calculation results of the five methods with the TOPSIS results, it is found that the results of the five methods are different from the results of TOPSIS. Among them, the MABAC ranking is only exchanged between the White Horse Group and the Zhongjian community, and the ranking of the other communities does not change. The ARAS methodology had five community ranking changes but no reversal in its ranking among high, low, and medium resilience communities. In the SAW method, the rankings of 11 communities changed. In the VIKOR method, only the rankings of Yaoao and Zhongjian communities were exchanged. In the GRA method, only the rankings of White Horse Group and Zhongjian community were exchanged. The rankings of other communities do not change. MABAC, VIKOR, and GRA methods have the highest ranking similarity. The results of Spearman rank correlation coefficient showed that the ranking of changing the comprehensive evaluation method was highly correlated with the ranking of TOPSIS method, indicating that the sensitivity of TOPSIS method was low.

6. Conclusion

The evaluation of community disaster resilience has important practical significance for the construction of resilient city with low risk and great sustainability. Based on the concept and connotation of community disaster resilience in domestic and foreign literature, this paper establishes the index system from seven dimensions such as demographic characteristics, infrastructure, and economic development. In this paper, 12 communities in Luoyang were selected as the research objects. The method of questionnaire survey and entropy weight TOPSIS method were used to measure the community disaster resilience, and the obstacle diagnosis model was used to explore the related constraints. The following conclusions are obtained:(1)The overall disaster resilience of Luoyang community is medium. Among them, community capital is the main factor influencing the construction of community disaster resilience. Government governance, community capacity, and community intelligence have a large weight and differ greatly among different communities, which need to be paid more attention to.(2)According to the results of cluster analysis, Luoyang community disaster resilience is decreasing from rural areas to urban areas. Moreover, communities with high disaster resilience are less than communities with low disaster resilience.(3)Based on the results of obstacle analysis, the constraints of community disaster resilience are mainly concentrated in demographic characteristics, economic development, and infrastructure. Among them, medical service ability is the biggest constraint to the construction of disaster resilience in Luoyang community. Economic development constrains almost all types of community disaster resilience construction. In addition, community trust, community dependence, popularization, and intellectualization of disaster prevention information, and disaster information sharing also have significant constraints on Luoyang community disaster resilience construction.(4)The analysis results of obstacle degree show that changing the weight method, weight value, and evaluation method will change the final disaster resilience level, but the regression of the ranking is rare. According to the Spearman rank correlation coefficient, each scheme is significantly correlated with the initial determined ranking of disaster resilience, indicating that the entropy weight TOPSIS method had high stability in the evaluation of community disaster resilience.

The method based on subjective perception is used to measure the disaster resilience of community residents. Then, the overall disaster resilience of community is assessed. As cognitive ability affects social cognitive ability, the relevant knowledge held by respondents can lead to vague judgments, as well as different lifestyles and knowledge levels, leading to differences in perceptions of community disaster resilience. The following research focuses on two aspects: (1) Through the optimization of community comprehensive disaster resilience index system and questionnaire, the applicability and accuracy in different regions are improved. (2) By verifying the difference of results and influencing factors of the measurement method based on the objective index and measurement method based on subjective perception, a set of evaluation methods for the comprehensive disaster resilience in accordance with China’s national conditions can be obtained.

Data Availability

The raw data supporting the conclusion of this article will be made available by the authors, without undue reservation.

Conflicts of Interest

The authors declare that there are no conflicts of interest.


This research was funded by National Social Science Fund of China, grant no. 15AGL013; China Earthquake Administration's major policy theory and practice problem research topic, grant no. CEAZY2019JZ09; Henan Provincial Social Science Planning Project, grant no. 2019BJJ030; Henan Higher Education Philosophical Social Science Basic Research Major Project, grant no. 2021JCZD04; and Research on the Construction of Disaster Prevention and Mitigation Support System in Large and Medium Cities in Henan Province, grant no. 222400410001.

Supplementary Materials

Appendices and supplementary materials will be provided as attachments. (Supplementary Materials)