1 Introduction

Affordable and reliable energy is critical to development worldwide. Coal is India’s most plentiful and vital energy source. Over half of the country’s energy needs are met by coal, which is predicted to account for 76% of the total capacity for thermal power. (Online document Coal India Ltd., website) It is expected that coal will play a large role in India’s energy security as a result of the country’s growing population, burgeoning economy, and desire to improve living standards. However, the impacts of coal mining operations pose risks to the environment and people surrounding the operating area, which have been widely documented in academic research [1,2,3,4,5]. Opencast mining has a negative impact on groundwater quality as well as flow regime [6]. Water pollution is mostly caused by runoff or discharge from mining activities, including extraction, processing, tiling, and disposal facilities, as well as repair and service facilities. An acid mine is a by-product of drainage, mining, and processing activities, and it is also another form of water pollution during active mining and after mine closure. Landslides, topography, erosion, and soil pollution are all caused by contamination or runoff from wastewater or leaching from tiling ponds, landfills, and treatment and repair workshops. Air quality is affected by emissions of particulate matter (PM10 and PM2.5) from dust during various mining operations [7,8,9,10]. There has been evidence of nearby populations being exposed to cancer-causing coal products such as silica dust and organic chemicals [11, 12]. In the coal mining area, there is a high mortality rate from heart, respiratory, and kidney diseases, as studied by Hendrix [13] and Werner et al. [14]. Coal mining is one of the most important sources of anthropogenic methane emissions [15,16,17]. The degradation of land and changes in its use pattern as a result of mining activities are forcing communities to move away from traditional economic activities such as agriculture, livestock, and fishing. In addition, occupational health and safety issues and adverse effects on vulnerable groups have increased the pressure on mining companies to integrate fundamentals of sustainability to better manage their social impact despite being the largest provider of energy and employment. Coal mining does not easily fit the definition of sustainability because it usually deals with a finite resource located in environmentally sensitive areas, leading to climate change risks. Although many studies, frameworks, and approaches are available in the literature for discussions on the definition, application, and implementation of sustainable development, scholarly arguments indicate that defining sustainable development in the context of different stages of coal mining remains a major challenge [18,19,20,21,22]. Even within the same units of study and context, the ongoing debate over mining sustainability reveals a lack of agreement on how to define mining sustainability [23, 24]. As a result, for the mining industry, creating an integration of sustainability concepts at the strategic level with sector-specific frameworks is crucial.

In response to the growing demands of society for accountability, there are Indian companies issuing sustainability reports. The GRI (Global Reporting Initiative) has been approved by a majority as a mechanism for disseminating sustainability reports (online sustainability reports). GRI rules were initially published in 2000. Their mission is to help businesses prepare sustainability reports that address social, environmental, and economic challenges. In 2016, the GRI went beyond making proposals to establish the first global standards for sustainability reporting known as “GRI standards.” The standard is broad, including important aspects such as organizational, management, economic, environmental, social, human rights, community, and product responsibility. The GRI has produced 78 indicators for the mining industry in the environmental, social, and economic domains. The environmental category has 11 subcategories, with a total of 33 indicators. There are three subcategories and 36 indicators in the social category and four subcategories and nine indicators in the economic category. Sustainability Reports based on GRI offers two benefits: assessing a company’s triple bottom line (economic, environmental, and social factors) and informing shareholders about sustainability efforts made and progress thereupon [25, 26].

Although GRI standards are recognized worldwide, they have a major drawback: instead of analyzing and reporting consistent performance at a site or mining site, mining companies are tasked with examining themselves and reporting on overall performance. Corporate reporting rather than reports on particular projects such as mining facilities might lead to confusion among shareholders based on the GRI guidelines. The studies of Böhling et al. [27], Fonseca et al. [28], Jeskins et al. [29], and Que et al. [30] are some examples of recent research in this field. GRI reports being voluntary, this flexibility gives companies the choice of which result indicators to disclose. A study by Boiral [31], Boiral and Henry [32] revealed that up to 90% of important adverse events were not disclosed in the sustainability reports of 23 corporations in the energy and mining industries, although they achieved A or A + application status in third-party assessment. Coal India Limited (CIL), a government-owned coal mine corporation in India, functioning in 84 mining areas with a total of 352 mines both open cast and underground, annually publishes sustainability reports highlighting the implementation of the GRI. However, few of its subsidiaries incorporate all 78 GRI indicators in it, and the reports lack representation of all 352 mining sites.

It is evident from the literature that in terms of measurement and management, a large number of key (core) sustainability indicators are often of concern. For target consumers, especially decision-makers and the general public, a large number of different indicators can be confusing [33]. It is also advocated that these indicators be easily comprehensible, with no complexity and a limited number for easy management [34]. Indicators should be adapted to their intended audience and preferably selected in tandem with them. A small number of 10–15 indicators that represent the most significant trends are generally considered by stakeholders such as decision-makers and the general public [35].

This study aims to explore the basic research question “How well do the 78 GRI standard indicators of the mining sector fulfill the requirement of mine site-specific sustainability using the Delphi method in the first stage and then prioritizing the reduced number of indicators using multicriteria decision-making (MCDM) techniques to obtain a manageable set that quantifies sustainability performance for individual coal mine sites.

To our knowledge, no studies have yet been conducted on the prioritizing of sustainability indicators of the mining sector using stakeholder participation and the MCDM approach. Nevertheless, Sitorus et al. [36] presented the application of MCDM methods in mining and mineral processing and even more specifically in mining technology, equipment, and site selection.

The present study intends to answer the basic research question mentioned above and thereby fill a gap in the literature and contribute in the following manner:

  1. 1.

    It provides a ranking of sustainability indicators by the multicriteria decision-making (MCDM) method and further classification by stakeholder perception mapping to obtain a prioritized set of indicators.

  2. 2.

    It will contribute to the literature on the comparability of MCDM methods.

  3. 3.

    Based on what was discovered within the limited scope of our research, there are no studies of the kind we conducted with stakeholder participation in the Indian coal mine on sustainability.

The remainder of this paper presents the following: Section 2 discusses the methodology adopted. Section 3 details the summary of findings including a discussion, Section 4 contains validation and sensitivity analysis, and Section 5 highlights limitations and future research along with conclusions.

2 Methodology

2.1 Delphi Method

The study utilized expert interviews and questionnaire surveys in two stages in line with the Delphi technique [37], which creates a structure for dealing with a complex problem effectively. The two stages of the study were conducted between 2017 and 2019.

For the first stage, which was a pilot study, expert interviews were chosen as the primary method to explore how 78 GRI sustainability indicators for the mining sector are viewed by the intended user in the Indian coal mining sector to fulfill the requirement of mine site-specific sustainability. The methodology adopted aims to focus on the understanding of the terms, their application, and the achievement of the intended goal within the regulatory framework of sustainable development in India, rather than on statistical generalization. The interviewees were selected based on their subject expertise and a minimum of 8 years of work experience in the subject matter. The sample size started with 30 individuals who were contacted for the discussion out of which 17 agreed to discuss the subject. The target population was mostly government officials involved in policy-making, and others were working in consulting firms who were directly involved in the development of the sustainability reports for industries (Table 1).

Table 1 Profile and code of the interviewee group in the first stage of the survey
Full size table

The interviews were semi-structured, and the interviewees had the freedom to elaborate on their views. GRI indicators for the mining sector were also provided for them to rate on the basis of importance with “yes” and “no” scales. The basis of the questionnaire was repeatable and systematic.

In the second stage, a group of professionals with working experience in mining sectors was considered an expert group for further rating of indicators under three pillars of the sustainability assessment framework developed by the GRI: environmental, social, and economic impacts. In this stage, 60 professionals were provided with the questionnaire, of which 40 responded. The sample size in the study is justified in the literature as the Delphi panel sample size ranged from three to eighty [38, p3].

Delphi’s first stage suggested revising the rating scale to emphasize comprehensibility, data accessibility, fit for purpose, measurement, and data usability.

Adopting the methodology developed by Feng and Joung [39] as cited by Popovic et al. [40], the rating of each of the indicators was performed based on the four attributes defined below:

  1. (i)

    Relevance: Is the indicator suitable for assessment from multidimensional standpoints, e.g., environmental, social and economic and technical, for the local coal mine? This attribute’s assessment will reveal the logical relationship between the GRI indicators and the respondent’s operating area.

  2. (ii)

    Practicality: How practical is the indicator for quantitative/qualitative measurement? Ratings on this attribute will shed light on ease of measurability of the GRI indicators with respect to the working area of the respondent.

  3. (iii)

    Reliability: Does the indicator provide trusted and useful information that is understandable by those who are not experts? Is such information readily accessible? Ratings on this attribute will shed light on ease of data accessibility and trustworthiness of measured data in accordance with the GRI indicators with respect to the working area of the respondent.

  4. (iv)

    Importance: How important is the indicator in addressing site-specific sustainability and its progress (trend) at the local level in the focal work area? Ratings on this attribute will shed light on the decision-making aspect of the GRI indicators even when measurability is directly or indirectly low from the perspective of the working area of the respondent.

The rating scale for attributes’ relevance, reliability, and practicality is given as “low, medium, high, and no opinion.”

The response to the questionnaire survey was changed to low = 1, medium = 2, and high = 3 according to the 3-point Likert scale, and the scale option “no opinion” was also provided in case the respondent had no idea about a particular indicator in the context of his working area. A three-point scale was selected for the respondents to generate the simplest and easiest response. Owing to the large number of GRI indicators in the questionnaire survey, the “importance” attribute is rated using a slightly different scale (1–9) to have a broader view of how one indicator dominates another under the “importance” attribute. Instead of cardinal data, the study employs a questionnaire survey with ordinal data (low, medium, and high) having assigned numerical values 1, 2, and 3 for the attributes practicality, relevance, and reliability to rate the 78 GRI Sustainability Indicators. For the TOPSIS application, this approach was taken in order to make it easier, more convenient, and quicker for the participants to fill out the questionnaire. Some of the earlier studies on TOPSIS application have adopted this approach [41,42,43,44,45,46,47,48,49,50,51,52].

2.2 Reliability of Scale

To determine the reliability of the scale, an appropriate statistical method, Cronbach’s alpha coefficient (α), or McDonald’s omega coefficient (ω), is used. Reliability coefficients can vary from 0 to 1.

The mathematical expression for Cronbach’s alpha coefficient (α) is as follows:

$$\alpha = \left(\frac{k}{k-1}\right)\left(1-\frac{{\sum }_{1=1}^{k}{\sigma }_{yi}^{2}}{{\sigma }_{x}^{2}}\right)$$
(1)

where k is the number of items/factors, \({\sigma }_{yi}^{2}\) refers to the variance associated with item i, and \({\sigma }_{x}^{2}\) refers to the sum of item variance.

The McDonald Omega coefficient (ω) is calculated as the ratio of variation owing to a common property (i.e., factor) to overall variance.

$$\omega = \frac{{\left({\sum }_{i=1}^{k}{\lambda }_{i}\right)}^{2}}{{\left({\sum }_{i=1}^{k}{\lambda }_{i}\right)}^{2}+{\sum }_{i=1}^{k}{\delta }_{ii}}$$
(2)

where \({\lambda }_{i}\) is the standardized factor loading, and \({\delta }_{ii}\) is the standardized error variance (i.e., \({\delta }_{ii}\) = 1 − \({\lambda }_{i}^{2}\)).

2.3 Multicriteria Decision-Making Approach (MCDM)

MCDM methods are the most widely used and well-known technique for ranking a group of alternatives. Stojčić et al. [53] reviewed and reported through online journal databases that more than four thousand articles in the field of science and technology covering sectors such as construction, supply chain management, transport and logistics, and energy make use of MCDM approaches. The basic procedure for MCDM methods requires three steps [54]:

  1. (i)

    Identifying decision problems, defining goals to be obtained, analyzing and selecting alternatives that will be evaluated against a set of criteria, and identifying decision-makers.

  2. (ii)

    Applying the MCDM method suitable to the defined objective by normalizing and weighting the alternatives and criteria chosen.

  3. (iii)

    Evaluating the relative performance of the alternative with respect to the chosen criteria and finally ranking the alternatives based on the overall weighted score. The higher the ranking is, the more preferable the alternative.

Technique for order preferences by similarity to ideal solutions (TOPSIS), simple additive weighting (SAW), multi-objective optimization based on ratio analysis (MOORA), preference ranking organization method for enrichment evaluations (PROMETHEE), elimination et choix traduisant la realite´ (ELECTRE), vlsekriterijumska optimizacija i kompromisno resenje (VIKOR), and multi-attribute utility theory (MAUT) are some of the most frequently applied methodologies in MCDM problems.

A diagram with the workflow describing the research methodology is given in Fig. 1.

Fig. 1
figure 1

Research design

Full size image

The MCDM approach—TOPSIS [55,56,57]—is used and justified in the present study due to its easy adaptability. MOORA [58, 59] and SAW [60] are applied to the results of TOPSIS as validation methods. The normalization method of TOPSIS is also used in MOORA and SAW to maintain uniformity.

Weight is a numerical number that shows the relative relevance of the criteria used to assess options in all of these multicriteria decision models. The entropy approach is used to determine the weights [61, 62].

A diagrammatic approach known as “perceptual mapping” is used to show how stakeholders perceive their decision with respect to the mean and standard deviation (SD). It is a pictorial representation of data to classify them into four quadrants by shifting its axis from the origin to a specified point. The four quadrants are labeled high mean high SD, high mean low SD, low mean high SD, and low mean low SD, by relocating their axis away from the origin to its average mean and average SD.

For successful optimization, the rankings from the three selected MCDM approaches along with stakeholder perception mapping are compared.

Expressions for the determination of weight and MCDM methodologies are summarized as follows:

  1. (I)

    Entropy method for the determination of weights [58]

    $${\mathrm{Matrix \;normalization }\;p}_{ij} = \frac{{x}_{ij}}{{\sum }_{i = 1}^{m}{x}_{ij}}\;i = 1,\dots \mathrm{m};\, j = 1\dots n$$
    (3)
    $$\mathrm{Output\; entropy} \;{E}_{j} = \frac{{\sum }_{i = 1}^{m}{p}_{ij }\,ln\left({p}_{ij}\right)}{ln (m)} \;j = 1,\dots . n$$
    (4)
    $$\mathrm{Weight\; entropy }\;{w}_{j} = \frac{1-{E}_{j}}{{\sum }_{i = 1}^{n}\left(1-{E}_{j}\right)}\;j = 1,\dots n$$
    (5)
  2. (II)

    TOPSIS method [63]

    The TOPSIS approach is broken down into the following steps:

    $${\mathrm{Normalized \;decision \;matrix} \quad n}_{ij}=\frac{{r}_{ij}}{\sqrt{{\textstyle\sum }_{i = 1}^{m}{r}_{ij}^{2}}}$$
    (6)
    $$\mathrm{Weighted \;normalized\; decision\; matrix}\quad {v}_{ij} = {w}_{ij}{r}_{ij}$$
    (7)

    The positive and negative ideal solutions are as follows:

    $${v}_{j}^{+}=\left\{{v}_{1,}^{+},{v}_{2}^{+}, \dots \dots ,{v}_{n}^{+}\right\} = \left\{{max}_{j} \left({v}_{ij}\right)\right\}$$
    (8)
    $${v}_{j}^{-}=\left\{{v}_{1}^{-}, {v}_{2}^{-},\dots \dots .,{v}_{n}^{-}\right\}=\left\{{min}_{j}\left({v}_{ij}\right)\right\}$$
    (9)

    The distance between each criterion and the positive and negative ideals is calculated as follows:

    $$D^{+}_{i}= \sqrt{\textstyle\sum_{j=1}^{n}{\left({v}_{ij}-{v}_{j}^{+}\right)}^{2}}$$
    (10)
    $$D^{-}_{i} =\sqrt{\textstyle\sum_{j=1}^{n}{\left({v}_{ij}-{v}_{j}^{-}\right)}^{2}}$$
    (11)

    Relative closeness to the ideal solution is reflected as

    $$C_{i} =\frac{{D}_{i}^{-}}{{D}_{i}^{-}+{D}_{i}^{+}}$$
    (12)

    The order of preference is ranked.

    The following expressions of MOORA and SAW are used for validation purposes:

  3. (III)

    MOORA method [58]

    $$\mathrm{Normalized \;decision \;matrix }\;{X}_{ij}=\frac{{x}_{ij}}{\sqrt{{\textstyle\sum }_{i=1}^{m}{x}_{ij}^{2}}}$$
    (13)
    $$\mathrm{Weighted\; normalized\; decision\; matrix }\;{W}_{ij}={w}_{j }\times {X}_{ij}$$
    (14)

    \({w}_{j}\) is the weight of the jth criterion

    $$\mathrm{Priorities }\left({Q}_{i}\right)={\textstyle\sum }_{j=1}^{n}{W}_{ij}$$
    (15)

    Ranking of preferences: The maximum value of \({Q}_{i}\) is ranked first and so on.

  4. (IV)

    SAW method

    $$\mathrm{Normalized\; decision \;matrix }\quad{n}_{ij}=\frac{{r}_{ij}}{\sqrt{{\sum }_{i=1}^{m}{r}_{ij}^{2}}}$$
    (16)

    Select the best option using Eq. (17)

    $$A_{i} = {\textstyle\sum }_{j=1}^{m}{w}_{j}{r}_{ij}$$
    (17)

    where \({A}_{i}\) is a performance value of the \({i}^{th}\) alternative; \({w}_{j}\) denotes the weights of the \({j}^{ih}\) criterion; and \({r}_{ij}\) is the normalized value of the \({i}^{th}\) alternative with respect to the \({j}^{ih}\) criterion.

2.4 Correlation of the MCDM Methods

Analysis of Spearman correlations may further establish the comparability of rankings derived by the various approaches of MCDM. No effort is made to establish one variable as dependent and the other as independent; rather, it is interpreted as an association among the MCDM approaches utilized in the study.

3 Results and Discussion

3.1 First Stage (Expert Interview)

The study explores the basic research question “How well do the GRI standard indicators of the mining sector fulfill the requirement of mine site-specific sustainability.”

We base our discussion on five main criteria:

  1. 1.

    To what extent is sustainability effectively incorporated into the current practice of environmental assessment in the Indian coal mining sector.

  2. 2.

    The current state of implementation of sustainability during operation.

  3. 3.

    A system to monitor and system to report sustainability.

  4. 4.

    A system to show progress/trend among coal mines.

  5. 5.

    Usefulness of global/national sustainable development assessment procedures for the performance assessment of local sustainable development.

Most of the interviewees who are directly involved in policy-making on sustainable development and GRI reporting and auditing stated that sustainable development is a major term in reference to mining projects in India to obtain environmental clearance (GC-2, GC-3, GE-1, GE-2, GE-4, CF-1, CF-2, CF-4, CF-5). The GRI standard is robust in guiding mining organizations to publish reports that provide relevant information about their contributions to sustainability (GC-1, GC-2 GE-4, CF-1, CF-4, CF-5, CF-6, CF-7, CF-9). However, the majority of interviewees agreed that “definition and interpretation of sustainable development/sustainability is biased by the organization” (GE-2, GE-4, GE-5, CF-1, CF-2, CF-3, GC-3) and “applicability to every stage of the project is difficult” (GC-1, GC-2, CF-6). Some respondents cited the voluntary nature of sustainability reporting as a barrier to adopting the standard in letter and spirit (GE-2, CF-1, CF-4, CF-5). A GRI reporting consultant (CF-8) suggested that “The current project scenario necessitates the development of methods to mitigate a variety of effects. Without a complete image of the project's output on multiple measures, the ability to provide a true picture of the project’s impacts as a whole would be jeopardized.”

Policy decision-makers have shown concern about monitoring and reporting systems for sustainability based on GRI standards, as “less focus on the quantitative aspect of GRI indicators is of concern. Most of the indicators require either a qualitative datum and sometimes a combination of qualitative and quantitative data” (GE-1, GE-3, CF-4, GE-5, GE-6). One GRI consultant (CF-7) revealed another problem associated with measuring sustainability: the “measuring scales are not identical for an indicator among two sustainability reports; rather, they are mixed.” Furthermore, on the question of whether post-monitoring of the mining project shows progress/trend, the opinion of policy decision-makers is that “overemphasis on positive statements undermine the problematic issues of an organization.” Regarding the usefulness of global/national sustainable development assessment procedures for local sustainable development performance assessment, participants from consulting firms (CF-1, CF-2, CF-7, CF-8, CF-9) viewed “less involvement of the operational workforce in discussion while working on sustainability issues for their work area” as a concern, while the majority of policy decision-makers stated that “challenges, such as climate change, poverty alleviation, biodiversity, energy access, etc., currently are very less incorporated in terms of geographic locations for meaningful sustainable progress” (GC-2, GE-1, GE-2, GE-3, GE-4, GE-5).

The Delphi technique helped to define the attributes and response scale for the second stage of the questionnaire survey to arrive at a holistic view of sustainable development in the coal mining sector.

3.2 Second Stage of Survey

3.2.1 Reliability of Opinions of Participants

The 78 indicators of GRI mining sector indicators representing the environmental, social, and economic pillars of sustainable development were used on the questionnaire for the participants to rate them on four attributes. Data analysis was conducted with Microsoft Excel 2016 version and Statistical software Jamovi version 1.6.23.

To validate the reliability of the questionnaire, Cronbach’s alpha coefficient was computed. The obtained overall Cronbach’s alpha coefficient value (Table 2) for the aspects evaluated was α > 0.70 and very proximate to 1. The Cronbach’s α was further compared with McDonald’s ω (Table 2), which in all cases was greater than Cronbach’s α, demonstrating a high degree of internal consistency and reliability.

Table 2 Overall Cronbach’s α and McDonald’s ω
Full size table

3.2.2 Ranking of GRI Indicators

As mentioned earlier, the responses of participants on the questionnaire were analyzed in the decision matrix for the purpose of MCDM methods, and weight computation was obtained from the mean response of participants for each indicator of sustainability on four attributes.

The entropy weight was computed and is presented in Table 3.

Table 3 Weight factor (Wj) of criteria by entropy method
Full size table

The attribute “practicality” was given a higher weight (0.411) for social (LPDW) indicators. For social (prod res) indicators and economic indicators, the attribute “relevance” was given a higher weight (0.380, 0.285). The attribute “reliability” was given a higher weight (0.305) for social (society) indicators. The environmental indicators were given a higher weight (0.361) for the attribute “importance.”

The TOPSIS method is considered the fundamental method for selecting the best-ranked indicators.

The ranking of each indicator was obtained utilizing the mathematical expressions shown in Eqs. (6) and (7).

Table 4 presents the TOPSIS ranking of the environmental sustainability indicators. The top five TOPSIS rankings include environmental protection expenditures and investments (EN 30), water withdrawal by source (EN 7), water discharge by quality and destination (EN 21), NOx, SOx, and other significant air emissions (EN20), and total weight of waste by type and disposal method (EN 22). These are perceived as most significant by the respondents.

Table 4 TOPSIS ranking GRI environmental sustainability indicators
Full size table

Greenhouse gas emissions, consumption, and reduction within and outside the organization (EN 14, EN 15, EN 16, and EN 17) ranked in the middle, suggesting that these are seen as significant, but respondents were not confident that they could be measured with the current methods available.

Biodiversity indicators (EN 10, EN 11, EN 12, EN 14) ranked at the bottom could be interpreted as operations outside the eco-sensitive area.

The subcategory of labor practices and decent work in social sustainability indicators (Table 5) ranks rates of injury, occupational diseases, lost days, and absenteeism and total number of work-related fatalities, by region and by gender (SL 6) first, emphasizing the need to monitor and report workplace-related safety in mining operations.

Table 5 TOPSIS ranking GRI social (labor practices and decent work) sustainability indicators
Full size table

Ranks 2, 3, 4, and 13 indicate interesting outcomes, as respondents perceive inclusive participation in health and safety meetings as essential but place less emphasis on training employees’ gender-wise (SL 9).

Skill management (SL 10) and equal remuneration (SL 13) find bottom-order ranking.

Minimum notice periods regarding operational change (SL 4) at rank 16 is a surprising observation as it is related to avoiding risks in the mining operation.

Local community engagement, impact assessments, development programs and political contributions (SS1, SS 2, and SS 6) are ranked in the top 3 under social (society) sustainability indicators (Table 6), suggesting that respondents consider it important to have transparent monitoring of the commitments to local communities in and around mining operations.

Table 6 TOPSIS ranking GRI social (society) sustainability indicators
Full size table

Formal grievance mechanisms and risk assessment for corruption in operations (SL 4 and SL 5) at the 4th and 5th rank reflect respondents’ view that there should be a foundation for responding early to concerns with the mining company and resolving community conflict.

In Table 7 below, the top-ranking assessment of significant product and service categories having health and safety impacts (SP 1) and provisions for penalties on violations of regulations and codes (SP 2) suggests that respondents are concerned about human wellbeing and are advocates for penalties if violations by the mining organization occur. The customer satisfaction survey (SP 9), ranking last in order, reflects its limited to no effect on sustainability.

Table 7 TOPSIS ranking GRI social (product responsibility) sustainability indicators
Full size table

In the economic sustainability category (Table 8), the top-ranking indicator, direct economic value generated and distributed (EC 1), reinforces its role in sustainability. Another interesting insight is that the middle-order rankings of infrastructure investment (EC 7) and procurement from local suppliers (EC 9) show heterogeneous opinions with the important attached to social indicators.

Table 8 TOPSIS ranking GRI economic sustainability indicators
Full size table

4 Validation and Sensitivity Analysis

4.1 Comparison of MCDM as Validation Instrument

Due to the difference in numerical algorithms and other reasons, such as a change in criteria weight, including and excluding alternatives, MCDM methods can exhibit different rankings. This necessitates the validation and sensitivity analysis of the outcome. In this study, MOORA and SAW are considered for validating the results and checking the correlation between the three MCDM methods. The rankings with MOORA and SAW were performed (Table 9) with the help of Eqs. (13) and (14).

Table 9 Rank comparison of TOPSIS with MOORA and SAW for 33 GRI environmental sustainability indicators by different methods of MCDM
Full size table

It is evident from Table 9 that the three MCDM methods have given almost similar rankings within one rank up or down between the three methods for most of the environmental indicators. Indicator EN27 shows a disparity of four places down in ranking from MOORA.

Similar results have been obtained for the GRI social and economic indicators and are shown in Appendix Tables 19, 20, 21 and 22.

4.2 Exchange of Entropy Weight as a Sensitivity Analysis Instrument

Any MCDM technique must conclude with a process called sensitivity analysis, which is used to determine how much uncertainty exists in the source data, process step, and criteria weights (either individually or in combination) and has an impact on the ranking of alternatives [64, 65]. Unlike existing approaches of increasing/decreasing criteria weight or removing input, this study utilizes the exchange of criteria weight to investigate the stability of rankings.

Tables 10 and 11 present the exchange of entropy weight and changes in the TOPSIS ranking of GRI environmental sustainability indicators.

Table 10 Exchange of weight for environmental indicators for four attributes
Full size table
Table 11 TOPSIS ranking of environmental indicators after sensitivity analysis
Full size table

Entropy weight exchange tables and the results of sensitivity analysis for social and economic indicators are shown in the Appendix (Tables 23, 24, 25, 26, 27, 28, 29 and 30).

It is clear from Table 11 that the exchange of entropy weight did not alter the ranking for most of the indicators for the environmental, social (society), social (product responsibility), and economic categories. However, in social (labor practices and decent work), exchange 1 and exchange 3 exhibit minor displacements of rank.

4.3 Stakeholder Perceptual Mapping as an Optimization Instrument

Stakeholder perceptual mapping also serves as a validation method as well as an optimization tool.

Stakeholder perceptual mapping for 33 GRI environmental sustainability indicators for the attribute “practicality” is shown below (Fig. 2). Indicators falling in the “high mean–low SD” quadrant of the stakeholder perception map are further selected for use with the three MCDM methods to validate their ranking against the original ranking.

Fig. 2
figure 2

Stakeholder perceptual mapping of GRI environmental sustainability indicators for the attribute “practicality”

Full size image

Individual stakeholder perception mapping is shown in Appendix Figs. 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, and 21.

The three MCDM methods are again applied to the prioritized indicators of all sustainability categories under all attributes. Tables 12, 13, 14, 15, and 16 show the ranking of the original and prioritized numbers of indicators (shown in parentheses) under environmental, social, and economic categories:

Table 12 Rank comparison of the prioritized number of indicators and original number of indicators (shown in parentheses) under environmental sustainability
Full size table
Table 13 Rank comparison of the prioritized number of indicators and original number of indicators (shown in parentheses) under social (labor practices and decent work) sustainability
Full size table
Table 14 Rank comparison of the prioritized number of indicators and original number of indicators (shown in parentheses) under social (society) sustainability
Full size table
Table 15 Rank comparison of the prioritized number of indicators and original number of indicators (shown in parentheses) under social (product responsibility) sustainability
Full size table
Table 16 Rank comparison of the prioritized number of indicators and original number of indicators (shown in parentheses) under economic sustainability
Full size table

The similarity in the final ranking of GRI indicators by TOPSIS, MOORA, and SAW validates the results of these MCDM methods. It also confirms their applicability in real time for large as well as small data. The methodologies are simple and flexible to execute without any complications in the calculation or use of dedicated software.

The study also supports the argument of Salabun et al. (2020) for the applicability of TOPSIS for a large number of alternatives (indicators in our study). Similar rankings of TOPSIS with MOORA and SAW for economic, social (society), and social (product responsibility) indicators also indicate the applicability of these methodologies for smaller data (> 5).

4.4 Correlation of the MCDM Methods

The Spearman correlation coefficient was used to compare the MCDM procedures, taking into account the minimal differences in ranks between the three methods.

Table 17 indicates that all decision-making procedures at all levels have a positive correlation greater than 0.85. In the environmental sustainability category, TOPSIS has a strong correlation (> 0.9) with MOORA and SAW. For the social sustainability subcategory labor practices and decent work, TOPSIS showed slightly less correlation with MOORA (0.85) than SAW (0.965). The social sustainability subcategory society in TOPSIS has a strong correlation (> 0.9) with MOORA and SAW. In the social sustainability subcategory product responsibility, TOPSIS showed slightly less correlation with MOORA (0.85) than SAW (0.95). In economic sustainability, TOPSIS has a stronger correlation with MOORA (0.98) than SAW (0.966).

Table 17 Spearman’s rho correlation coefficient
Full size table

The results of the Spearman correlation coefficient exhibit an acceptable range of correlation, which is comparable to the work of Moradian et al. [59]. As a result, we can say that the ranks in the compared methodologies are statistically reliable with a probability of 95%, and the experimental results were statistically reliable.

4.5 Prioritized Set of Indicators

The comparison of MCDM methods and the results of sensitivity analysis enable us to obtain a set of prioritized numbers of indicators (Table 18) that are viewed as most significant by the participatory stakeholders. The prioritized indicator set can serve as common indicators for local coal mine sites. Furthermore, indicators ranked until the fourth position by all methods can be selected as priorities when beginning a study, supporting the arguments of Janoušková et al. [66] that instead of developing new indicators, the assessment of existing indicators should be used as a policy decision-making tool, as sustainability is a dynamic process that is always changing.

Table 18 Prioritized set of indicators
Full size table

5 Limitations and Future Research

Finally, there are some potential research avenues that may be explored to go beyond the constraints of our study.

While the TOPSIS method is suitable for cardinal data, due to practical considerations and simplification, the study employs ordinal data transformed into a numeric form using a 3-point scale to apply the TOPSIS technique.

The study considers entropy weight only for the decision-making matrices. The equal weight method is not performed separately, as lowering the entropy weight of this study to two decimal places yields an equal weight. The study also does not check the effect on ranking by other popular weighting methods, such as the analytic hierarchy process (AHP) method, since it was computationally intensive to perform pairwise comparisons for a large number of indicator sets. Rank reversal issues [67] in MCDM methods were also not studied, as the study started with a fixed number of indicators and criteria ratings after Delphi discussion rounds. This area can be an interesting direction for future research.

Other MCDM methods, such as complex proportional assessment (COPRAS), preference ranking organization method for enrichment evaluations (PROMETHEE), elimination et choix traduisant la realite´ (ELECTRE), and vlsekriterijumska optimizacija i kompromisno resenje (VIKOR), could also be compared with TOPSIS.

Despite this limitation, it is believed that the current study can serve as a foundation for future research.

6 Conclusion

The depletion of natural mineral resources is a major issue in the debate over the long-term viability of mining. Mining companies require a commitment to sustainable environmental and social development from exploration to extraction, as coal is still in demand worldwide as a low-cost energy source to meet the sustainable development goal of energy security.

Seventy-eight GRI indicators are prioritized to 20 indicators using entropy and MCDM approaches along with stakeholder perception mapping. Comparison of the MCDM approach exhibits close agreement between the approaches establishing the efficacy of these indicators singly or in combination to inform the decision-making process to prioritize indicators, assess sustainability at coal mining sites, and establish benchmarking between adjacent mining sites. To stay sustainable, the mining sector must focus on the challenges of climate change as well as preventing and reducing pollution.

The key insights from this study contribute to generating more consistent and comparable sustainability reports with a manageable set of GRI sustainability indicators, which will improve sustainable mining operations. The study suggests that MCDM approaches are a straightforward, effective, and efficient method for solving a variety of challenges involving the ranking and selection of sustainability indicators. Measuring and reporting on progress against the indicators will help coal mining companies in delivering the Sustainable Development Goals (SDGs) of the UN 2030 Agenda for Sustainable Development in the regions surrounding their mine sites and the nations in which they conduct business.