Image in the mining sector

Image is the mental construction of a public about a brand, company, industry, sector, or country based on the evaluation of information about certain attributes with which organizations are defined and differentiated (^{Capriotti, 2013}); in other words, the image is an idea or attitude generated by the receiver of the information. In the 1970s, amid the global economic crisis, corporate efforts focused on improving the image and reputation of the company to achieve social legitimacy (^{Said et al., 2022}), increasingly incorporating social responsibility policies and making it an imperative for entrepreneurial success (^{Hoelscher & Rustad, 2019}).

In the mining sector, a negative image of its socio-environmental performance prevails, which results in resistance to mining in emerging economies and those in post-conflict (Sydd et al., 2023); therefore, improving the global or supra-organizational image (i.e., that which influences the image of all companies in the sector and vice versa) became the focus of attention of various meetings worldwide during the 1990s (^{Dashwood, 2014}; ^{Said et al., 2022}). Thus, driven by leading companies, it led to a commitment to corporate social responsibility (CSR), framed in international standards² and voluntary initiatives (between society, public and private sector) in search of mining sustainability (^{Potts et al., 2018}). Even more than any other sector, mining has joined many CSR initiatives due to increasing pressures from external groups, such as local communities and Non-Governmental Organizations (NGOs) (^{Sun & Bahizire, 2023}).

Theoretically, mining is an emerging thematic field in the literature on CSR, widely studied by the administrative sciences (^{Barnett et al., 2020}; ^{Karakaya & Nuur, 2018}). In the mining field, CSR refers to the voluntary and mandatory corporate commitment, according to the regulations of each country, to improve the welfare conditions (e.g., economic, social, and environmental) of the communities; these efforts respond to the expectations of the stakeholders and seek to mitigate the negative effects of mining activity (^{Rodrigues et al., 2022}; ^{Said et al., 2022}). Therefore, CSR should be implemented with actions perceived as beneficial by the community, as it is the most important stakeholder (^{Said et al., 2022}).

Recently, there was a paradigm shift toward CSR 2.0 and the creation of shared value, which benefits both business and society (^{Hoelscher & Rustad, 2019}; ^{Barnett et al., 2020}),³ as well as community support for or acceptance of the company’s operations, known as social license to operate (SLO) (^{Meesters et al., 2021}), has been observed. Research on “CSR image” has emerged within this framework, which studies public perceptions of corporate responses to social concerns (^{Pérez & Rodríguez del Bosque, 2013}). Although CSR can provide relational, reputational, and financial benefits to companies, the benefits to society are rarely validated, so there is a lack of evidence as to whether CSR delivers the social benefit it promises (^{Barnett et al., 2020}).

One of the key social issues in the CSR literature is the environmental effects of mining (^{Karakaya & Nuur, 2018}), as well as its potential conflicts with the community (Pérez Falcón et al., 2022; Sydd et al., 2023). Given the relevance of the effects of mining on the environment close to communities, the use of CSR to improve the image, mitigate impacts, and gain acceptance of the activity has remained current in academic (^{Devlin, 2023}; ^{Alamgir & Nasir-Uddin, 2017}), business (^{KPMG, 2019}; ^{Deloitte, 2018}), and organizational (^{United Nations Development Programme [UNDP] 2018}) discourse. Nevertheless, only recently has interest been detected in broadening the conceptualization of CSR to study its effect on the environmental and social impacts of extractive activities (^{Barnett et al., 2020}).

Thus, the socio-environmental field has been the focus of CSR initiatives in the mining sector, but these have gradually expanded to other areas, such as infrastructure, health, education, sports, culture, and art (^{International Organization for Standardization’s [ISO], 2010}; ^{Deloitte, 2018}); toward policies of community participation and building relations with the community (^{Badera, 2014}); as well as economic and business development initiatives (^{Amin-Chaudhry et al., 2019}; ^{UNDP, 2018}; ^{ISO, 2010}). Nevertheless, mining companies often fail to implement CSR, as it responds to shareholder welfare rather than corporate engagement or community interest (^{Isacowitz et al., 2022}).

It should not be forgotten that groups respond to CSR to the extent that they derive and perceive personal benefits, according to the relation established with the company (^{Bhattacharya et al., 2009}), hence the importance of perceptions.⁴ Nonetheless, a sustainable development approach is appropriate for the study of perceptions due to 1) the extractive nature of the sector, with a high potential to cause damage to communities and the environment; 2) the public scrutiny to which mining companies are subject, given the negative effects generated; and 3) the fact that it is a Business to Business (B2B) sector. Therefore, the analysis of the image of mining should leave aside the scales proposed by the literature based on the “product or service” and focus on the company’s contribution to sustainable development in mining communities.

Proposal for a mining social image scale

The social image of the mining sector is defined in this study as a set of ideas, beliefs, and impressions that community members share about the mining activity operating in their communities. Consequently, the image is evaluated externally, that is, based on the attributes that society perceives of the sector, which are regularly manifested through the CSR actions undertaken by the companies in the mining communities. To this end, the study of images includes individual and social impressions of the perceptions of a set of attributes, which, at an appropriate level, develop more favorable attitudes.

At the supra-organizational level, there are numerous studies validating an image scale for the tourism (^{Marinao-Artigas & Barajas-Portas, 2021}) and financial (Marinao-Artigas, 2018) sectors. Nevertheless, after reviewing the academic literature, no studies were identified concerning the construction and validation of image scales in the mining sector, except for the work of ^{Ruíz-Martín et al. (2014.)}; instead, qualitative research or case studies of mining companies where images or CSR are studied are more frequent. Although the literature on image measurement is fragmented and there is little consensus on the dimensions to be evaluated (^{Pérez & Rodriguez del Bosque, 2013}; ^{Peloza & Shang, 2011}), after reviewing these studies, the inclusion of two approaches is recognized: image-attitude and perceived benefits.

The image-attitude approach is evaluated, from social psychology, from the cognitive-affective-conative (CAC) model and its components: a) cognitive, based on knowledge, thoughts, beliefs, and ideas that one has about it (i.e., rational); b) affective, i.e., feelings or emotions that it provokes when perceived (i.e. irrational); and c) behavioral, predisposition to act in a certain way (i.e. conative) (^{Yang et al., 2022}; ^{Kim & Chen, 2021}; ^{Capriotti, 2013}). On the other hand, it is suggested that benefits are evaluated as part of the rational component of the image (^{Marinao-Artigas & Barajas-Portas, 2021}; ^{Marinao-Artigas, 2018}; ^{Capriotti, 2013}) through the perceived benefit or value; that is, what each person believes that an attribute or object can do for them.

CAC is a hierarchical cause-and-effect model (^{Yang et al., 2022}) in which image components are activated one after another, forming a psychological decision-making process (^{Kim & Chen, 2021}). Nevertheless, it should be considered that the behavioral component is an outcome factor of Image-Attitude, framed in predisposition and not in performance (i.e., final action); therefore, an assessment from the affective and cognitive components is suggested (^{Capriotti, 2013}). Thus, if the Image-Attitude is positive, the individual’s actions when faced with the object (e.g., mining sector) may generate or strengthen a favorable behavior.

Thus, a social image scale is proposed that captures perceptions of CSR as a cognitive component since these are the attributes that companies offer to communities and those who perceive their value (^{Said et al., 2022}; ^{Peloza & Shang, 2011}) and this is conceived, together with affective valuations, as the constructs of the image, which outlines the attitude of the inhabitants of the communities toward mining (see Figure 1). Based on the theoretical and conceptual elements presented, the following dimensions of the scale are proposed: 1) emotional perception, related to an individual’s feelings toward mining, which may be favorable, unfavorable, or neutral; and 2) cognitive perception, based on the personal assessment of social, economic, and environmental CSR policies, thus addressing a sustainable development approach. Finally, the behavioral component will be evaluated outside the scale, as it shows an intention or probability of accepting or rejecting mining in the community.

Source: created by the authors based on the review of the literature

Figure 1 Theoretical model of the mining social image scale 

Construction of the measurement scale

Table 1 shows a summary of the AFE. Initially, 3 items with factor loadings < .50 were discarded, leaving 17 items for the analysis,¹¹ ordered and renumbered by size; four components were identified, which the authors called subscales, and which explain 67.0% of the total variance (see Table 1). The KMO statistic provides evidence of the correlation between items (value close to unity). Hence, Bartlett’s test (p − value ≤ .05) shows the feasibility of simplifying the information. Reviewing the rotated factor loadings (λ), each of the subscales registers a good degree of unidimensionality according to the items attached to them and a good internal consistency according to Cronbach’s Alpha (α > .70) (see Table 1). The subscales identified, ordered according to the variance explained, are labeled as environmental CSR (AMB), emotional perception (EMO), economic CSR (ECO), and social CSR (SOC).

The first component, i.e., the AMB scale, comprises six items concerning the perception of whether these CSR actions are carried out by mining (at the sector level.) These are ordered according to representativeness: care for natural resources (P1), concern for local species (P2), protection from dangerous toxins (P3), reduction of water pollution (P4), nature protection programs (P5), and reforestation campaigns (P6) (see Table 1). The second subscale consisted of four items of emotional perception about the informant’s feelings regarding the presence of mining activity in the region, both positive (P7 and P8) and negative (P9 and P10); among these, the positive emotions expressed were the most loaded in this subscale (see Table 1).

The third subscale is composed of four items of economic perception (ECO), so that the associations between mining and the creation of sources of employment (P11), improvements in the economy (P12), and the economic performance of society contribute more to the construction of ECO than the promotion of the growth of economic activities (P14) (see Table 1). Finally, the fourth subscale, called SOC, is made up of 3 items of perceptions of CSR social actions: sponsorship of health programs (P15), educational and cultural support (P16), and donations to social causes (P17); among these, health programs have a stronger association with SOC (see Table 1).

In order to build the scale, it is hypothesized that the image is constructed from the covariation of the subscales, each of which has at least three indicators and satisfies the order condition (^{Vargas & Mora-Esquivel, 2017}). Thus, it starts from a first-order measurement model (MDM) (see Figure 2, left). Nevertheless, the degree to which this satisfactorily predicts the covariance matrix, through the Chi-square statistic (χ²), is not the desired χ113gl2=315. 4 (p-value≤ .01), but alternative measures suggest a good fit.¹² Furthermore, the subscales of CSR (i.e., AMB, SOC, and ECO) show a more robust correlation (between .33 and .62) than that between EMO and other subscales, except ECO (see Figure 2, left). This information provides a guideline for evaluating a second-order MDM (see Figure 2, right) that conforms to the theoretical proposal in Figure 1.

Source: created by the authors based on estimates in Amos^® (v.23.0).

Figure 2 Mining social image scale (Standardized coefficients) 

Figure 2 shows that when moving from a first-order to a second-order multidimensional model, the χ² value decreases (χ111gl2=208.5), but remains statistically meaningful (p − value ≤ .01).¹³ Nevertheless, given that the increase in χ² between models is ∆χ22gl=106.9, being greater than the critical value of 5.991 (p − value ≤ 0.05), the second-order multidimensional scale is more appropriate than a first-order one (see Figure 2, right). It is also observed that all fit measures improve and suggest: 1) a good overall fit of the data to the model (RMSEA = .049; GFI > .90); 2) an adequate incremental fit, both CFI and NFI have values > .90; and 3) a good degree of parsimony, as (χ²/gl = 1.878) and is within the suggested rank of 1 to 3, and PGFI has ‘acceptable’ values, even lower than other indices (^{Byrne, 2010}).

Continuing with Figure 2 (right), the factor loading (λ_i) is the representativeness of the observed variable or item (P_i) in the latent construct (ξ_i), while e_i and δ_i are measures of error in P_i and ξ_i respectively, for i = 1, 2, 3...373. The estimates of λ_i range between .58 and .94 and the item variance is explained by its respective subscale in a significant ratio (. 31 < R² < .86), meeting the criteria of ^{Jöreskog and Sörbom (1993)}, namely: 1) p − value ≤. 05; 2) λ_i > .50; and 3) R² > .30; and showing evidence of the unidimensionality of the subscales, with items associated with each other and representing a single construct (^{Hair et al., 2014}).

Assessment of the reliability and validity of the scale

The items of the scale have good construct reliability; that is, the ratio of the item variance explained by the construct is above the suggested value of > .70 in most cases (see Figure 2); in others, the values are between .50 and .60, which are acceptable factor loadings in incipient stages of a scale (^{Hamid et al., 2017}; ^{Chin, 1998}). Regarding reliability, the first and second-order subscales show internal consistency, with Composite Reliability (CR) greater than the critical value (CR > .70); but without exceeding the threshold of .90, which is not desirable (^{Hamid et al., 2017}). These data are also supported by the McDonald reliability index [MaxR(H)] (see Table 2).

Construct validity (i.e., how well the measure defines the concept) is assessed by the average variance extracted (AVE). Table 2 shows that the subscales have convergent validity (AVE > .50) and are reliable as items converge or share a high ratio of common variance (^{Fornell & Larcker, 1981}; ^{Hair et al., 2014}). In the case of ECO, with an AVE _ECO < .50 and a CR _ECO > .70 (above the critical value), its construct validity is still considered adequate, as the AVE is a more conservative or restrictive reliability measure (Fornell & Larcker, 1981, p. 46).¹⁴ Thus, irrespective of scale grade, the items accurately represent the concept of interest.

On the other hand, discriminant validity is assessed following the ^{Fornell and Larcker (1981)} criteria: 1) the construct must explain the variance of its items better than that of other constructs, and 2) the AVE of the construct must be greater than the squared inter-construct correlations (AVE). The results in Table 2 show that discriminant validity is fulfilled for the subscales, as MSV < AVE and (AVE) greater than the absolute value of the inter-construct correlations (the diagonal of the correlation matrix in Table 2 shows the value of (AVE); therefore, each subscale differs, and the items distinctively represent only one construct.

Predictive validity and scale discussion

The proposed scale was constructed from cognitive (COG: AMB, SOC, and ECO) and emotional (EMO) perceptions, representing two dimensions of a formative scale of the social image of mining. Nevertheless, as a test of the scale’s predictive validity, items of community satisfaction with the presence of mining are taken from the same instrument used to collect data on perceptions. The aim is to construct an image-attitude model to explore the relations between the image dimensions and the predisposition toward social conduct or behavior concerning mining (see Figure 1).

This behavioral element that is added to the analysis is manifested through three items, coded on a Likert scale of five positions, where the informants respond on the degree of agreement or disagreement with the following statements: 1) canceling mining activity in the community; 2) mining as a main characteristic of the community; and 3) mining gives prestige to Sonora. It should be noted that the construct ‘Behavior’ is analyzed outside the image scale and is included in the model for SEM analysis to have an over-identified model and estimate the structural relation within a multiple indicators multiple causes (MIMIC) model (^{Diamantopoulos, 2011}; Bagozzi & Yi, 2012).

The IMAGE-ATTITUDE model, which includes a second-order mining image scale, is illustrated in Figure 3; for ease of interpretation, factor loadings and other estimated parameters of the subscale items are omitted, corresponding to Figure 2, right. The SEM results indicate a good model fit, as RMSEA and χ²/gl register desired values of .034 and 1.423, respectively. Likewise, the incremental fit and parsimony statistics are in the desired ranges (see Figure 3). The results of the model confirm the existence of a direct and positive relation between emotional (EMO) and cognitive (COG) perceptions in the construction of the mining IMAGE-ATTITUDE, the combination of both explaining 83.0% of the variance of the formative construct (see Figure 3).

Source: created by the authors based on estimates in Amos^® (v.23.0)

Figure 3 Structural image-attitude model (SEM) of mining in the Sonora River (Standardized coefficients) 

Figure 3 shows that IMAGE-ATTITUDE positively profiles the behavior of individuals in the community, with a positive direct effect of β = .69 (p − value ≤ 0.01); furthermore, this construct shows a statistically meaningful relation with the behavioral dimension, indicating its ability to predict it and confirming the predictive validity of the image scale. Thus, an increase in IMAGE-ATTITUDE by one standard deviation (hereafter s. d.) causes behavior toward mining activity to vary by .69 s. d. The ‘Behavior’ items explain 47.0% of their variance, where the perception that mining brings prestige to the entity was the most important (λ₁ = .46), followed by the predisposition to cancel mining in the community (λ₂ = .41) and mining as a main characteristic of the community (λ₃ = .37); nevertheless, the λ (statistically relevant) and R² register low values.¹⁵

While the SEM shows evidence that COG is more influential than EMO in constructing the IMAGE-ATTITUDE, it is interesting to observe the degree to which these dimensions influence, for example, the predisposition toward behavior in favor of or against the cancellation of mining in the community. This effect is standardized and indirect (i.e., mediated) and statistically meaningful, so that when COG and EMO increase by 1 s. d. the predisposition varies .206 s. d. (.73 ∗ .69 ∗ .41), and .152 s. d. (.54 ∗ .69 ∗ .41), respectively.

Concerning the image scale, the COG dimension reports a standardized coefficient of β = .73 (p − value ≤ 0.01), so that when COG increases by 1 s. d., IMAGE-ATTITUDE increases by .73 s. d. Similarly, COG was assessed as a second-order reflective component, according to the perceived benefits of environmental (AMB), social (SOC), and economic (ECO) CSR. Among these subscales, SOC influences COG the most (λ = .78), followed by ECO (λ = .70) and AMB (λ = .67), being statistically meaningful parameters. Thus, an increase of 1 s. d. in COG implies increases of .78 s. d. in SOC; .70 s. d. in ECO; and .67 s. d. in AMB. The predictor items of SOC and AMB explain, in both cases, 67.0% of their variance and 49.0% in the case of ECO.

Regarding the EMO dimension, this exhibits a standardized coefficient of β = .54 (p − value ≤ 0.01). It shows that an increase in EMO of 1 s. d. causes IMAGE-ATTITUDE to increase by .54 s. d. The EMO subscale is manifested through the sensations provoked in the individual when mining is evoked, so a favorable perception, where mining makes the individual feel calm/comfortable or less fearful/less vulnerable, contributes to constructing a positive image-attitude. On the other hand, the calibration detected a direct and positive relation between EMO and the subscales AMB (β = .47) (p − value ≤ 0.01) and SOC (β = .25) (p − value ≤ 0.01) (see Figure 3), which was not initially considered.

Thus, the structure of the image models is under debate; on the one hand, the suggestion of the CAC model and, on the other hand, the proposal of the Image-Attitude model evaluated here. The CAC model (^{Yang et al., 2022}; ^{Kim & Chen, 2021}) suggests a hierarchical causality (EMO → COG → ATTITUDE), where a particular environment or situation triggers individuals’ cognitive reactions, which, in turn, will stimulate feelings and emotions that will influence their behavior. Nevertheless, in the case of the mining sector, in contrast to the tourism or financial sector (^{Marinao-Artigas & Barajas-Portas, 2021}; ^{Marinao-Artigas, 2018}), the Image-Attitude model in the context of the Sonora River communities (see Figure 3) resulted in a better fit than one of CAC structure. Nonetheless, it can be observed that emotional perceptions impact the perception of cognitive benefits, measured by CSR in the environmental (AMB) and social (SOC) domains.

The main finding is the validation of the proposed scale, which confirms the direct and positive relation of COG and EMO with IMAGE-ATTITUDE. Thus, people show a favorable attitude toward mining, concerning image, based on the cognitive benefits of CSR and positive emotions (e.g., peace of mind, safety); consequently, CSR is a key element that builds the image, shapes attitudes, and mediates behavior toward mining. Thus, despite the prominence of CSR, the emotional dimension caused by catastrophic environmental events in the sector, which are rooted in the collective memory of the community, such as the case of the ‘toxic spill’ in the Sonora River, should not be forgotten (^{Orozco & Rodríguez, 2020}, ²⁰²²; ^{Lugo-Gil & Lara, 2022}), which could decrease the likelihood of ‘pro-mining’ behavior.

Accordingly, the formation of perceptions and attitudes contributes to the operationalization of the social license to operate (SLO), which, despite its conceptual ambiguity, is conceived as a tool for local risk management based on legitimacy, credibility, and trust (^{Devlin, 2023}; ^{Meesters et al., 2021}). While SLO is not a compulsory license in Mexico, obtaining community and other stakeholder support is recognized. It is part of the strategy of mining companies in the country, i.e., companies are encouraged to pursue SLO to operationalize CSR (^{Brunet et al., 2023}). Thus, the proposed image scale contributes significantly to implementing and monitoring SLO, which, together with CSR, are crucial factors in discussions about ‘community mining relations.’

Therefore, managing a positive image in mining will require a much more efficient CSR strategy than in other sectors (^{Barnett et al., 2020}), as the negative impacts of mining activity can undermine the benefits of CSR. Furthermore, CSR needs to be grounded in the reality of the communities, as the CSR discourse raises high expectations (e.g., employment and development; social services such as education, health care, road, and electricity infrastructure, among others). Hence, mismanagement of CSR strategy leads to 1) community skepticism toward CSR actions, 2) lack of commitment, 3) mistrust of the company in the eyes of the community, 4) divergent interpretations of CSR benefits between community and company, and 5) CSR initiatives that do not respond to the “real needs” or socio-cultural contexts of the community (^{Devlin, 2023}).