Introduction

Companies from the oil industry sector are at the heart of persistent debates around their lack of respect for the natural environment (e.g. oil drilling controversy in Nigeria or in the Arctic National Wildlife Refuge), and the need for these organizations to diminish their carbon footprint notably by developing cleaner and more sustainable energy sources (Frynas 2005). Gaining broader societal acceptance thus became a critical factor in such controversial industry sectors where organizations’ failure to meet stakeholders’ societal expectations result in perceptions of those organizations as illegitimate (Campbell 2007; Palazzo and Richter 2005; Sethi 1975). In such a context, corporate social responsibility (CSR) initiatives could appear as a useful tool to ensure organizations’ license to operate by deflecting stakeholders’ negative perceptions towards activities that might be perceived as unsustainable (Palazzo and Richter 2005; Yoon et al. 2006). For example, the Royal Dutch/Shell petrochemical company initiated its CSR engagement following criticism from pressure groups during the 1995 Brent Spar and Nigeria crisis (Frynas 2003, 2005), and mentions on its societal webpage that “the organization aims to address social concerns at its operations and works to benefit local communities, protecting our reputation as we do business”.Footnote 1

In the oil industry sector, CSR initiatives are often presented as protecting and promoting the natural environment, which may appear paradoxical, as this industry is known as one of the largest polluters. Frequent accusations of ‘greenwashing’, or perceived discrepancies between stated environmental intentions and an organization’s effective initiatives, crystallize stakeholders’ scepticism and distrust towards organizations’ environmental initiatives in this specific industry (Frynas 2005). To date, management literature has just begun to explore how CSR initiatives could mitigate stakeholders’ usual defiance towards controversial companies, and therefore better contribute to these companies’ long-term legitimacy and performance (Frynas 2005; Palazzo and Richter 2005; Yoon et al. 2006).

According to CSR literature, research should focus more intensively on the micro-mechanisms pertaining to stakeholders’ behaviours in order to understand the potential link between an organization’s societal investment and its overall performance (Bhattacharya et al. 2009; Hansen et al. 2011; Pivato et al. 2008). Among the various stakeholders considered in existing research it appears that “employees as the unit of analysis have received scant attention in the CSR literature” (Aguilera et al. 2007, p. 839). The few published studies in this field have essentially explored the relationship between perceived CSR and employees’ commitment (Brammer et al. 2007; Peterson 2004; Turker 2009a). However, although these studies indicate a potential influence of CSR on employees’ favourable attitudes, their findings do not illuminate the mechanisms that explain and predict such relationships (Bhattacharya et al. 2009; Jones 2010). In this respect, employees’ psychological responses to CSR still need to be further investigated in order to understand why, how and under which circumstances organizations’ investment in societal initiatives could favourably impact employees’ attitudes (Bhattacharya et al. 2009; Jones 2010).

The present study’s objective is to address these research questions in order to gain a better understanding of the mechanisms through which CSR initiatives foster employees’ support to their organizations’ goals, even in controversial industry sectors. Specifically, our study goes beyond previous research regarding employees’ attitudinal reactions to CSR in three ways. First, we investigate why perceived CSR can potentially influence employees’ attitudes. In this respect, our model tests CSR’s influences on employees’ organizational identification (OI), largely recognized as a critical psychological process for reinforcing employees’ relationship with, and support of an organization (Bhattacharya et al. 2009; Riketta 2005). In particular, we examine whether the previously hypothesized influence of perceived CSR on employees’ OI (i.e. Jones 2010; Kim et al. 2010) holds in the controversial oil industry sector where CSR initiatives are usually directed at protecting the natural environment. In so doing, we survey employees of an international oil company whose reputation has been put under great pressure due to various environmental disasters and which, at the same time, implements concrete actions of environmental management recognized by an ‘eco-dynamic’ label.Footnote 2

Second we develop an integrated framework that tests how perceived CSR can impact employees’ OI through two potential mediating mechanisms explaining this relationship: perceived external prestige (PEP) (Dutton et al. 1994; Smidts et al. 2001) and organizational trust (Colquitt et al. 2007; Mayer et al. 1995; Robinson 1996). Finally, based on marketing literature, we explore a contingency effect that can explain some of the circumstances under which CSR could more favourably impact employees’ attitudes. More specifically, we assume that employees’ attributions of the motives underlying an organization’s CSR initiatives play a moderating role on the hypothesized relationship between perceived CSR and organizational trust.

To the extent of our knowledge, this hypothetical framework represents the first empirical effort to explain and predict employees’ responses to CSR initiatives in controversial sectors.

Research Background

A Stakeholder-Centric Approach to CSR

Our paper adopts a stakeholder-oriented conceptualization of CSR and relies on Barnett’s (2007) definition of CSR initiatives as “a discretionary allocation of corporate resources towards improving social welfare that serves as a means of enhancing relationships with key stakeholders” (Barnett 2007, p. 801). Stakeholders are broadly defined as any individual, group or entity that can affect, or be affected by, an organization’s activity (Freeman 1984). According to stakeholder theory, organizations’ existence depends on their ability to integrate stakeholders’ expectations into their business strategy because stakeholders provide essential resources and returns for organizations’ successful functioning and survival (Donaldson and Preston 1995). As such, scholars have argued that stakeholders’ perceptions, attitudes and behaviours act as a mediating process in the relationship between an organization’s societal and financial performance (Barnett 2007; Bhattacharya et al. 2009; Pivato et al. 2008).

In this respect, marketing literature suggests that perceived CSR relates positively to consumers’ attitudes and supportive buying behaviours towards the company (Bhattacharya and Sen 2004; Maignan et al. 1999; Pivato et al. 2008; Sen and Bhattacharya 2001). More recently, an emerging stream of research based on social identity theory (SIT) started to investigate employees’ reactions to CSR initiatives and suggested a positive correlation between perceived CSR and employees’ attitudes such as organizational commitment (Brammer et al. 2007; Peterson 2004; Turker 2009a). For instance, Peterson’s (2004) empirical study testifies that perceived CSR could influence organizational commitment. He defends the argument that members will be prouder to identify with a respectful and socially well-regarded organization because such affiliation is supposed to boost their self-esteem (Peterson 2004). Similarly, Brammer et al. (2007) suggested that external CSR, mainly concerned with organizational reputation, is positively related to organizational commitment, even after controlling for variables such as procedural justice and job satisfaction. More recently, Turker’s study (2009a) on employees’ affective commitment demonstrates that CSR initiatives directed at different stakeholder groups (e.g. customers, employees, government, environment, NGO, and so forth) can influence employees’ attitudes.

However, it increasingly appears that in order to better explain and predict the returns of organizations’ societal investment, we need a deeper understanding of the psychological processes driving employees’ favourable responses to CSR (Bhattacharya et al. 2009; Jones 2010). In this respect, the reviewed studies indicate that SIT, and its underlying self-enhancement process, is a good framework for explaining CSR’s impact on employees’ attitudes (Brammer et al. 2007; Peterson 2004; Turker 2009a). Specifically, these studies argue that CSR initiatives can reflect a positive organizational image that enhances employees’ pride and willingness to be associated with such a reputable organization (Peterson 2004; Turker 2009a). However, none of these studies empirically investigates these mechanisms or the impact of CSR on employees’ OI, known as the central psychological process underlying SIT (Ashforth and Mael 1989; Hogg and Terry 2000).

Organizational Identification

According to SIT, social identification corresponds to the psychological process through which individuals classify themselves into various social groups of reference (e.g. nation, organization, political or religious affiliations, and so forth) in order to reinforce their self-esteem and overall self-concept (Hogg and Terry 2001; Tajfel 1981; Tajfel and Turner 1986). OI thus corresponds to a specific form of social identification where individuals develop a feeling of oneness with, or belongingness to an organization (Ashforth and Mael 1989, p. 34). More specifically, it corresponds to “the degree to which a member defines him or herself by the same attributes that he or she believes define the organization” (Dutton et al. 1994, p. 239). This psychological process explains how individuals’ attitudes can be influenced through the establishment or maintenance of a relationship with their social groups of reference (Kelman 1958; O’Reilley and Chatman 1986). In this respect, employees’ OI has been associated with a number of important outcomes (see Riketta 2005) which makes the investigation of its antecedents an important priority on the research agenda (Mael and Ashforth 1992; Smidts et al. 2001).

Social identification is guided by two underlying mechanisms that fulfil some of individuals’ psychological needs (e.g. belongingness, self-esteem, meaningful existence) (Bhattacharya et al. 2009; Cropanzano et al. 2001). The first mechanism is a categorization process which corresponds to individuals’ efforts to classify people, including the self (i.e. self-categorization), into social groups of reference that make more sense of their lives (Tajfel 1981; Turner et al. 1987). In particular, through self-categorization, individuals integrate social groups of reference that are congruent with their own attributes and values in order to satisfy their psychological needs for belongingness and meaningful existence. Concurrently, through a self-enhancement process, individuals also try to affiliate themselves with social categories that reflect a positive and distinctive external image in order to boost their self-esteem (Hogg and Terry 2000, 2001; Tajfel 1978). Therefore, if individuals believe that outsiders appreciate their social group, they tend to feel more pride in their affiliation with it, and thus identify more strongly with the group.

CSR and Employees’ OI

Few studies have provided empirical evidences of the psychological processes leading from CSR initiatives to employees’ favourable attitudes. Rodrigo and Arenas (2008) investigated qualitatively how employees’ relationship with their organization evolved as a result of their organizations’ CSR initiatives. Specifically, employees’ perceptions of their organization’s social role and image lead many of them, who formerly felt that their organization was just a place to work, to view their employer as an institution that shares their own social views and values (Rodrigo and Arenas 2008). This study suggests that CSR initiatives can help an organization to define its ethical stance with regard to the moral values of society and therefore drive employees’ OI. To our knowledge, only two quantitative studies have observed a relationship between CSR initiatives and employees’ feeling of belongingness with their organization (Jones 2010; Kim et al. 2010). Jones’ study (2010) on 162 employees of a publicly traded company supported that OI was a central psychological process that mediates the relationship between employees’ attitudes towards their organization’s volunteer program and their behaviours. Particularly, the author referred to employees’ self-enhancement process to demonstrate that the relation between CSR (i.e. the organization’s volunteer program) and employees’ OI was partly mediated by employees’ pride described as the psychological link between organizational membership and one’s self-concept (Jones 2010; Riketta 2005). In the same vein, Kim et al.’s study (2010) based on 109 Korean employees found that CSR participation and perceived CSR (i.e. perceived philanthropic CSR) were both positively related to employees’ OI.

As such, the reviewed literature presents sufficient theoretical and empirical evidences to expect a relationship between perceived CSR and employees’ OI. However, there is still a dearth of research looking at the mechanisms through which employees interpret and react to organizations’ CSR activities in a way that strengthen their OI, and thus relationship with their employer (Bhattacharya et al. 2009; Jones 2010).

Hypotheses Development

The Mediating Role of PEP

An organization’s PEP, or construed image (Dutton and Dukerich 1991; Dutton et al. 1994), corresponds to employees’ best guesses of what characteristics outsiders are likely to ascribe to their organization (Dutton et al. 1994; Smidts et al. 2001). In other words, it consists of employees’ reflected evaluations of the organization’s reputation (Brown et al. 2006). Employees care about the organization’s reputation because external stakeholders’ admiration or disregard for the organization’s image has implications for employees’ self-regard and OI (Dutton and Dukerich 1991). Specifically, an employee will tend to identify more intensively with an organization that he or she considers as well regarded by external stakeholders (Mael and Ashforth 1992; Smidts et al. 2001). The pride employees obtain by working in a respected and prestigious organization reinforces their self-esteem and therefore encourages them to identify with it (Dutton and Dukerich 1991; Dutton et al. 1994). Supporting this argument, empirical studies consistently reported a positive relationship between PEP and employees’ OI (Bartels et al. 2009; Mael and Ashforth 1992; Smidts et al. 2001).

However, although a reputation partly results from an organization’s capacity to successfully respond to its stakeholders’ expectations (Fombrun and Shanley 1990), the impact of CSR on PEP still remains under investigated (Kim et al. 2010). Literature already demonstrates that CSR initiatives can be an effective tool for reinforcing organizations’ reputation and attractiveness towards desirable and highly qualified applicants (Backhaus et al. 2002; Morsing 2006; Turban and Greening 1997). Furthermore, Morsing’s (2006) conceptual study based on auto-communication theory suggested the potential role of CSR initiatives directed at external stakeholders in members’ OI. This study argues that positive reactions to CSR among external stakeholders should reinforces members’ organizational pride, and thus identification (Morsing 2006).

A few field studies have specifically addressed the potential mediating role of PEP in the relationship between perceived CSR and employees’ OI (Dutton and Dukerich 1991; Kim et al. 2010). For example, Dutton and Dukerich’s (1991) qualitative study highlighted that employees’ PEP and identification were influenced by organizational events related to societal issues. This study supports that employees do pay great attention to CSR initiatives because these societal actions are used by external stakeholders to make character judgements about the organization’s reputation and, by association, that of its members. More recently, Kim et al.’s (2010) study emphasized the role of PEP in understanding the link between employees’ reflected evaluations of CSR and their OI process. Although the authors did not analyse the intervening role of PEP through mediated regression analyses, their findings showed that perceived CSR (i.e. perceived philanthropic CSR) was positively related to PEP which in turn was related to employees’ OI. On the contrary, employees’ participation in CSR activities was directly associated with employees’ OI, without being related to PEP (Kim et al. 2010). The reviewed literature thus suggest that CSR can foster employees’ self-enhancement process through its impact on PEP, and therefore support employees’ efforts to define themselves through organizational affiliations that reflect attributes able to reinforce their self-esteem and overall self-concept. Therefore, we suggest the following hypothesis:

Hypothesis 1

PEP mediates the positive relationship between perceived CSR and employees’ OI.

The Mediating Role of Trust

Trust involves one’s intention to accept vulnerability to a trustee’s actions based on the expectation, or belief, that these actions will be beneficial, favourable, or at least not detrimental to one’s interest (Colquitt et al. 2007; Robinson 1996). More specifically, it corresponds to the trusting party’s (i.e. trustor’s) evaluation of the exchange partner’s trustworthiness (or credibility—see Mayer et al. 1995) consisting of the trustee’s level of ability, integrity and benevolence (Bhattacharya et al. 2009; Mayer et al. 1995; Morgan and Shelby 1994). Ability corresponds to the trustor’s belief in the trustee’s competencies and expertise in a specific domain that ensures the trustor that the expected outcome will be effectively delivered by the trustee. Integrity is usually associated with the trustor’s belief in the trustee’s honesty, openness, fairness and/or consistency. It usually involves a perception of congruency, or the belief that the trustee adheres to a set of principles and values considered as acceptable by the trustor (Mayer et al. 1995; Sitkin and Roth 1993). Finally, benevolence corresponds to the trustor’s belief that the trustee’s actions are not guided by egocentric or opportunistic motives (Bhattacharya et al. 2009; Mayer et al. 1995). In this respect, positive trust-based evaluations have been argued to reduce uncertainty and perception of risk in relationships, which in turn enhances cooperation and effective teamworking (Colquitt et al. 2012; Morgan and Shelby 1994; van den Bos 2001).

In organizational settings, the degree of trust is largely reflected by an organization’s capacity to create and preserve strong relationships with its stakeholders and, as a result, to improve intra and inter-organizational cooperation and performance (Colquitt et al. 2007; Zaheer et al. 1998). Organizational trust thus appears as a key variable in any fruitful and long-term organization–employee relationship by acting as social glue in the establishment of employees’ long-term attachment to their organization (Atkinson and Butcher 2003; Ertürk 2010; Mishra and Morrissey 1990). Accordingly, it has been argued that organizational trust is an important component to fostering employees’ OI and empirical studies brought further evidence that organizational trust relates to employees’ attachment and feeling of belongingness to their organization (Dirks and Ferrin 2002; Ertürk 2010; Lee 2004; Mulki et al. 2006; Perry 2004; Straiter 2005).

However, although the literature showed that perceived CSR plays a key role in forging consumers’ trust in an organization and its products, the impact of CSR initiatives on employees’ trust-based evaluations remains largely unexplored (Hansen et al. 2011). Hansen et al. (2011) seem to be the first to have investigated the benefits of CSR initiatives in terms of employees’ trust evaluations and subsequent behaviours. Particularly, authors demonstrate that organizational trust mediates the effect of perceived CSR on employees’ behaviours because perceived CSR sends important signals to employees about an organization’s ethical stance and moral values, and therefore the extent to which it can be trusted. Moreover, the above-reviewed literature highlights that employees with a high degree of organizational trust develop a strong perception of similarity and attachment to their organization (Dirks and Ferrin 2002; Ertürk 2010; Harris and Wicks 2010; Lee 2004). These mechanisms can thus facilitate and support employees’ self-categorization efforts to define themselves through organizational affiliations that reflect the same attributes as those that define their own self-concept. Therefore, we suggest the following hypothesis:

Hypothesis 2

Organizational trust mediates the positive relationship between perceived CSR and employees’ OI.

The Moderating Role of CSR Attributions

CSR attributions refer to the causal reasoning stakeholders engage in when trying to understand the underlying motives of an organization’s CSR activities (Becker-Olsen et al. 2006; Bhattacharya and Sen 2004). Research on the role of CSR attributions in marketing literature revealed that despite the clear potential of CSR initiatives to drive organization-favouring outcomes on the part of consumers, the expected benefit from CSR investment is anything but guaranteed (Bhattacharya et al. 2009; Bhattacharya and Sen 2004; Ellen et al. 2006; Yoon et al. 2006). In particular, research highlighted that perceptions of the underlying motives of an organization’s societal engagement may moderate the effect of perceived CSR on consumers’ affective, cognitive and behavioural responses (Becker-Olsen et al. 2006; Sen et al. 2006; Swaen and Chumpitaz 2008).

Two primary types of motives have been investigated in CSR literature: (1) an organization’s self-centred (self-serving, egoist or strategic) motives where CSR initiatives are perceived as increasing the organization’s bottom line; and (2) an organization’s other-centred (public-serving, genuine or social) motives where CSR initiatives are perceived as acting out of genuine concern and as increasing social welfare (Becker-Olsen et al. 2006; Bhattacharya and Sen 2004; Ellen et al. 2006). In this respect, Sen et al. (2006) found that the effect of perceived CSR on consumers’ OI was significantly stronger for the consumers’ group with high CSR genuine attributions. In the same vein, Swaen and Chumpitaz’s study (2008) shows that the impact of perceived CSR on consumers’ trust and intent to purchase was significantly lower for the consumers’ group with high CSR strategic attributions. Of particular importance, this study supports other authors’ argument that consumers’ CSR attributions interact with their trust-based evaluations (Ellen et al. 2006; Lafferty and Goldsmith 1999; Osterhus 1997). Furthermore, marketing literature suggests that consumers’ attribution reasoning process is more salient whenever consumers’ scepticism and suspicion about organization’s activity and reputation is higher than average, as in the case of controversial industry sectors (Becker-Olsen et al. 2006; Bhattacharya et al. 2009; Yoon et al. 2006). Yoon et al.’s study (2006) even observed that CSR initiatives can backfire on an organization’s reputation when consumers are sceptical about the genuine motives of the organization’s societal actions and have reasons to doubt the validity of the information offered to them. Exploring these contingency effects can therefore support CSR literature efforts to understand under which circumstances CSR initiatives in controversial industry sectors can reinforce organizational trust, and subsequently strengthen the employee–organization relationship (Bhattacharya et al. 2009). Based on the sum of these arguments, we propose the following hypothesis:

Hypothesis 3

Employees’ self-centred CSR attributions moderate the strength of the positive relationship between perceived CSR and organizational trust.

Method

Sample and Procedure

In 2010 we surveyed employees working at the European headquarters of the petro-chemical division of an international oil company. Employees’ perceptions and attitudes were collected through self-report questionnaires. In order to reduce common method variance due to social desirability bias (Spector 1994), we first assured participants that replies to the questionnaire would be anonymous and that their answers would be processed in the utmost confidentiality by an independent researcher. We stressed that there were no right or wrong answers and that researchers were only interested in participants’ spontaneous opinions. Additionally, as recommend by the literature, we also separated the independent and dependent variables’ questions in the survey (Podsakoff and Organ 1986). Finally we conducted a single factor model for these data (see Harman’s single factor test; (Harman 1967) which revealed a poor fit (χ 2 [152] = 838.38, NNFI = .78, CFI = .80, RMSEA = .17; SRMR = .12), suggesting that no single factor model accounts for a majority of the variances explained. These precautions provide some evidence that common method bias was not a serious issue in our data.

Of the 206 surveys mailed to all employees working in the European headquarters of the petro-chemical division under study, 155 usable questionnaires were collected, corresponding to an overall response rate of 75 %. This high response rate rules out any concern for nonresponse bias. The final sample displays the following socio-demographic characteristics. About 60 % of participants were female and only 15 % had to manage a team of employees. The median age of respondents was 35–39 years and 55 % have been employed for more than 10 years within the organization. Therefore, most of them had experienced some of the environmental and social scandals targeting the organization over the past few years. Respondents occupied a wide variety of positions. None of these variables displayed a significant correlation with the dependent variables of our framework. Therefore, following literature’s recommendations, we omit to control for these demographic variables in the subsequent analyses in order to reduce the model complexity (Becker 2005).

Measures

All items were translated into French using a standard translation-back-translation procedure (Brislin 1980). Responses were made on a Likert-type scale ranging from 1 (strongly disagree) to 5 (strongly agree).

Perceived CSR: Petro-chemical organizations are part of the controversial oil industry sector where CSR issues gained considerable public and political attention in the context of the increasing threat of global warming and because of various environmental disasters created by this industry (Frynas 2005). In this overall climate of environmental pressure, we could reasonably expect that CSR environmental initiatives would reflect some organizational norms and values that are considered as attractive, or at least socially acceptable, for individuals working in this specific controversial industry sector. Moreover, at the time of the survey, the petro-chemical headquarters under study had just implemented an environmental program in order to obtain an “eco-dynamic” quality label.Footnote 3 In this overall context, the present research specifically focused on employees’ perceptions of their organization’s initiatives designed to protect and promote the natural environment (i.e. perceived environmental CSR).

To measure this specific dimension of CSR we relied on the scale recently developed by (Turker 2009b) which distinguishes four types of CSR initiative according to the type of stakeholders targeted: CSR to customers, CSR to employees, CSR to government, and finally CSR to secondary social stakeholders (i.e. not involving human relationship) and non-social stakeholders (i.e. environment, society, NGOs and future generations). For the purpose of our study, we adapted this last dimension of CSR to specifically address the company’s environmental responsibility and retained the four following items: “(Name of Organization) implements special programs to minimize its negative impact on the natural environment”, “(Name of Organization) targets sustainable growth which considers future generations”, “(Name of Organization) supports nongovernmental organizations working on the preservation of the natural environment”, “(Name of Organization) encourages its employees to participate in voluntary activities design to protect the environment”. The scale was unidimensional (items loaded on the predicted factor ranging from .60 to .79) and reliable (Cronbach’s alpha = .71).

PEP was assessed with the three items developed by Bartels et al. (2009) designed to measure employees’ evaluations of their organization overall reputation. Items are: “(Name of Organization) has a good reputation”, “(Name of Organization) is seen as an example organization”, and “(Name of Organization) is appreciated”. The scale was unidimensional (items loaded on the expected factor ranging from .72 to .88) and reliable (Cronbach’s alpha = .80).

Organizational trust evaluations were measured along three items developed by Dulac and colleagues (2008) and specifically designed to measure employees’ degree of trust in their organization. Items are “I believe (Name of Organization) has high integrity”, “(Name of Organization) is always honest and truthful”, and “In general, I believe (Name of Organization) motives and intentions are good”. The scale was unidimensional (items loaded on the predicted factor ranging from .64 to .79) and reliable (Cronbach’s alpha = .89).

CSR Attributions was measured along a three-item scale based on the work of Ellen et al. (2006). Items are “(Name of Organization) is taking advantage of its environmental initiatives to help its own business”, “(Name of Organization) implements environmental initiatives to get publicity”, “(Name of Organization) hopes to increase its profit by implementing environmental initiatives”. The scale was unidimensional (items loaded on the predicted factor ranging from .73 to .82) and reliable (Cronbach’s alpha = .71).

Organizational identification was measured using the famous six-item scale developed by Mael and Ashforth (1992). A sample item is “When someone criticizes (Name of Organization), it feels like a personal insult”. The scale was unidimensional (items loaded on the predicted factor ranging from .60 to .89) and reliable (Cronbach’s alpha = .88).

Construct Validity Analysis

We examined the distinctiveness of our five constructs using confirmatory factor analysis via LISREL 8.72 and the maximum likelihood method of estimation (Jöreskog et al. 2001). We tested a sequence of 15 nested measurement models, ranging from the hypothesized five-factor model to a one-factor model (Bentler and Bonett 1980). In line with Kline’s (2011) recommendations, model adequacy was assessed using fit indexes describing the measurement model from different perspectives. In so doing, we relied on two incremental fit indexes (CFI and NNFI), on two absolute fit indexes (SRMR and RMSEA), and on one parsimony index (normed Chi-square or [χ2/df]) as these indices are widely reported and recommended in SEM literature (Kline 2011; Williams et al. 2009). The most restrictive cut-off values are .95 for CFI, more than .90 for NNFI, .05 or less for RMSEA, .08 or less for SRMR, and less than 2.00 for χ2/df (Hu and Bentler 1999; Kline 2005; Tabachnick and Fidell 2007). Additionally, the Chi-square difference test was used to compare the fit of the hypothesized measurement model with that of the fourteen more constrained models. As shown in Table 1, the results indicate that the five-factor model has a very good fit (χ2 = 181.88; df = 142; χ2/df = 0.78; RMSEA = 0.04; NNFI = .98; CFI = .98; SRMR = .07) and was significantly superior to all more constrained models. All the individual items loaded reliably on their predicted factors and the t-Student estimate associated to each factor loadings were all significantly related to their construct ranging from 5.17 to 12.01 for environmental CSR, 7.01 to 8.2 for CSR attribution, 7.32 to 12.90 for PEP, 11.50 to 14.01 for organizational trust and 8.10 to 13.91 for OI which support the convergent validity of our measures.Footnote 4

Table 1 Confirmatory factor analyses fit indices for measurement models
Full size table

Results

Descriptive Statistics

The means, standard deviations and correlations among the variables used for this study are displayed in Table 2. These analyses reveal that perceived CSR positively relates to employees’ OI (r = .36; p < .01). Moreover, these correlations are consistent with our multiple mediation hypotheses. First, results show that perceived CSR is positively associated with PEP (r = .43; p < .01) and that PEP is positively associated to employees’ OI (r = .22; p < .01). Second, it appears that perceived CSR positively relates to organizational trust (r = .43; p < .01), which is positively associated with employees’ OI (r = .62; p < .01). It is also worth noting that perceptions of self-centred motives justifying organizational CSR actions towards the environment appear to correlate negatively with employees evaluations of organizational trust (r = −.30; p < .01) which suggests that self-centred CSR attributions could diminish the potential effect of perceived CSR on organizational trust.

Table 2 Descriptive statistics and correlations among latent variables
Full size table

Structural Model and Hypothesis Testing

Structural model were also examined using LISREL 8.72. The overall fit of this model is good (χ2 = 205.33; df = 162; χ2/df = 1.27; RMSEA = 0.04; NNFI = .98; CFI = .98; SRMR = .07). A partially mediated model was tested where a direct path from perceived CSR to employees’ OI was added. Although this model resulted in a relatively similar fit to the hypothesized model, the direct link between perceived CSR and employees’ OI was non-significant (β = .17, p > .05) suggesting that the fully mediated model is more parsimonious than, and should be preferred over, the partially mediated model (Raykov and Marcoulides 1999).

Concerning our mediation analyses, Fig. 1 first attests that employees’ perceptions of environmental CSR positively influence PEP (β = .54; p < .01) and organizational trust (β = .62; p < .01). Moreover, it appears that employees’ evaluation of organizational trust positively influence their OI (β = .78; p < .01). However, although correlation analyses support the association of PEP with employees’ OI, the structural model attests that PEP does not significantly influence employees’ OI when controlling for the other variables in the model (β = −.06; p > .05) suggesting that PEP does not mediate the link between perceived CSR and employees’ OI when integrating organizational trust in the model.

Fig. 1
figure 1

Structural model: Standardized path coefficients and R 2 for the hypothesized model. Note: *p < .05, **p < .01. Interaction variable: perceived CSR multiplied by self-centred CSR attributions

Full size image

However, in order to surpass the limitations of Baron and Kenny’s (1986) causal steps approach for assessing mediation effects (e.g. no evaluation of the indirect effect) and its lack of guidance for testing multiple mediation models (see limitations of the causal step approach in Hayes 2009; MacKinnon et al. 2002; Preacher and Hayes 2008; Williams et al. 2009), we followed recent literature developments and recommendations in research methodology to use a joint approach of product-of-coefficients using Sobel’s test (1982) with bootstrapping technique (Preacher and Hayes 2008). In our case this strategy presents two main advantages.

First, this technique allows the investigation of multiple mediation hypotheses as is the case in our two mediators’ model (Preacher and Hayes 2008). In this respect, analysing a multiple mediation hypotheses should involve two steps: (1) investigating the total indirect effect which consists of testing the total indirect effect of perceived CSR on OI through our set of mediators (i.e. PEP and organizational trust) and (2) testing the hypotheses regarding individual mediator in the context of a multiple mediators model (i.e. investigating the specific indirect effect associated with each of our two mediators while controlling for the other mediator) (Preacher and Hayes 2008).

Second, bootstrapping is an analytical strategy that overcomes the Sobel test’s assumption that the indirect effect follows a normal distribution under the null hypothesis, this being questionable in the case of small sample such as that of the present study (Preacher and Hayes 2004, 2008; Sobel 1982). As such, bootstrapping is a nonparametric resampling strategy that “makes no assumptions about the shape of the distributions of the variables or the sampling distribution of the statistic” and “produces a test that is not based on large-sample theory, meaning it can be applied to small samples with more confidence” (Preacher and Hayes 2004, p. 722).

Results of our two mediators’ model are presented in Table 3. Taken as a set, PEP and organizational trust do mediate the effect of perceived CSR on employees’ OI, as the total indirect effect is significant (β = .32; p < .01). Moreover, as the total direct effect is non significant (β = .18; p > .05), we can conclude to a complete mediation. Looking at the specific indirect effects using the Sobel test (see MacKinnon et al. 2002; Sobel 1982) indicates that the link between perceived CSR and employees’ OI is not mediated by PEP (Sobel test z = −.71, p > .05) while employees’ evaluations of organizational trust appears to fully mediate the link between perceived CSR and employees’ OI (Sobel test z = 4.92; p < .01). As depicted in Table 3, the 95 % BCa bootstrap CI of PEP [−.11 to .05] and of organizational trust [.22 to .52], support the product-of-coefficient strategy suggesting that the total indirect effect of perceived CSR on employees’ OI is due exclusively to employees’ organizational trust evaluations. This is also consistent with the examination of the contrast of indirect effect that shows that the specific indirect effect through organizational trust is larger than the specific indirect effect through PEP (BCa bootstrap CI [−.58 to −.22]).

Table 3 Indirect effect of perceived CSR on organizational identification through PEP and organizational trust
Full size table

Finally, concerning the moderated regression analysis, Fig. 1 attests that perceived CSR is positively related to organizational trust (β = .62, p < .01), and that self-centred CSR attributions is negatively related to employees’ organizational trust evaluations (β = −.23, p < .05) (both were mean centred; Aiken and West 1991). Finally it appears that the interaction variable (i.e. perceived CSR multiplied by self-centred CSR attributions) is also significantly related to organizational trust (β = .14, p < .05). These results bring a first support to our hypothesis 3 specifying that the effect of perceived CSR on organizational trust is moderated by employees’ self-centred CSR attributions. Additionally, we conducted a stepwise approach through separate hierarchical regression analyses (Cohen and Cohen 1983) in order to report the progression of our R 2 and thus assert the moderating effect of perceived self-centred CSR attributions on the relationship between perceived CSR and organizational trust. In this respect, the predictor variable (perceived CSR) was entered in Step 1. The potential moderator variable (self-centred CSR attribution) was entered in Step 2 (both were mean centred; Aiken and West 1991). Finally, the interaction variable (perceived CSR multiplied by self-centred CSR attributions) was entered in Step 3. To interpret the moderation result, regression coefficients for predictor effects were obtained from Step 1, moderator effects from Step 2, and interaction effects from Step 3 (Cohen and Cohen 1983).

Table 4 shows that self-centred CSR attributions moderated the effects of perceived CSR on organizational trust. From step 1 we can see that perceived CSR was related significantly to organizational trust (β = .61, p < .01), and accounted for 36 % of its variance (p < .01). From step 2 we can see that self-centred CSR attributions was significantly related to organizational trust (β = −.21, p < .05), and accounted for 5 % of its variance (p < .05). Finally, from step 3 we can see that the interaction variable was significantly related to organizational trust (β = .14, p < .05), and accounted for a more modest 2.1 % of its variance (p < .05).

Table 4 Results of stepwise moderation regression analyses (standardized regression coefficients and R 2)
Full size table

To facilitate interpretation of the significant moderator effects, in Fig. 2 we plotted the relationship between employees’ perceived CSR and organizational trust, across high and low self-centred CSR attributions (Aiken and West 1991). Perceived CSR was significantly and positively related to organizational trust within both groups of employees, whether they had low or high perceptions of self-centred CSR attributions and both slopes were significantly different from each other (t test = 3.44) (see post hoc test Aiken and West 1991). Plotting the interaction terms illustrates that when self-centred CSR attribution is high, as perceived CSR increases, the increase in organizational trust is much steeper (the slope of the line depicting the perceived CSR—organizational trust relationship) than under the condition of low self-centred CSR attribution. Therefore, under low self-centred CSR attributions the level of organizational trust increases much less as perceived CSR increases.

Fig. 2
figure 2

The moderating effect of employees’ self-centred CSR attributions on the relationship between perceived CSR and organizational trust

Full size image

Discussion

Overall, findings suggest that perceived CSR can reinforce employees’ OI in the controversial oil industry sector through its effect on organizational trust, which should in turn foster employees’ supportive attitudes and behaviours such as their role of ambassador towards external stakeholder (see Jones 2010; Riketta 2005). Additionally, our study highlights that the mediating mechanism through organizational trust is not as straightforward as one might expect. It rather suggests that employees’ level of self-centred CSR attribution acts as a contingency factor that moderates the intensity of the link between perceived CSR and organizational trust in such a controversial industry sector. Finally, our study also reveals that PEP does not significantly mediate the relationship between perceived CSR and employees’ OI in the controversial oil industry, even though our results attest that perceived CSR is positively associated with PEP, usually recognized as an important predictor of employees’ OI (Bartels et al. 2009; Smidts et al. 2001). Together, these results lead to important implications for theory and practice.

Theoretical Implications

Taking a micro-level analysis perspective, the present research suggests that a strong engagement in CSR initiatives can, even in controversial industry sectors, support organizational legitimacy and long-term performance, by impacting employees’ OI recognized as a strong driver of employees’ favourable outcomes at work (Jones 2010; Riketta 2005). More specifically, the present research contributes in several ways to the emerging field of literature analysing employees’ responses to CSR initiatives.

First, our findings testify that employees’ OI is a key psychological process for understanding why CSR initiatives can reinforce employees’ favourable relationship with, and attitudes towards an organization. In line with the few quantitative studies testing this assumption (Jones 2010; Kim et al. 2010), our findings bring strong empirical support that SIT is thus a fruitful framework for understanding employees’ responses to CSR. In so doing, our study investigates an overlooked dimension of CSR, namely environmental CSR, while previous studies on OI have essentially focused on the impact of employees’ participation in CSR (e.g. through volunteer program) or of perceived CSR towards the community (Jones 2010; Kim et al. 2010). Additionally, our research also extends prior findings as we demonstrated that even in controversial industries, CSR initiatives could be favourably perceived by internal stakeholders. In this regard, our results do not support the existence of an inevitable backfire effect of environmental CSR on employees’ supportive attitudes in the controversial oil industry sector. This study rather testifies that an organization with a poor reputation in the environmental realm can still strengthen its legitimacy in the eyes of internal stakeholders by engaging in concrete initiatives that preserve and promote the natural environment.

Second, the present study also contributes to the understanding of how perceived CSR impacts employees’ willingness to feel a sense of oneness with their organization. Based on OI theory, we investigated two potential mediating mechanisms: PEP and organizational trust. Contrary to our hypothesis, PEP does not play a significant mediating role in the relationship between perceived CSR and employees’ OI. These results contradict the view that perceived CSR fosters employees’ OI through a self-enhancement process which could be based on the organization’s external reputation, as usually suggested by CSR literature in this field (Brammer et al. 2007; Kim et al. 2010; Peterson 2004; Turker 2009a). A potential justification for this unexpected result could reside in the specific industry sector under study. Indeed, organizations’ reputation from the oil industry sector usually suffer from an inherent negative bias in external stakeholders’ perceptions because of their persistent failure to meet stakeholders’ environmental expectations (Frynas 2005). Therefore, in order to rationalize their long-tenured engagement in an oil company, employees may be led to ignore its external image, and thus stop monitoring its PEP. Concerning the second mediating mechanism in our framework, findings support that organizational trust fully mediates the relationship between perceived CSR and employees’ OI. These findings suggest that CSR initiatives directed at other stakeholders (i.e. the natural environment) have the potential to signal an organization’s ethical stance to its members, and therefore inform them of the potential risks and benefits of being strongly associated with it. Specifically, the present study supports the previous literature’s argument according to which CSR initiatives signal to employees some organizational traits and values (e.g. level of benevolence, competencies, honesty) that inform and reassure the former about the organization’s trustworthiness and overall character (Hansen et al. 2011). This mechanism should consequently support employees’ self-categorization process through which they affiliate themselves with organizations that share their values in order to strengthen their self-concept and make more sense of their lives. In so doing, we responded to Jones’ call (2010) to explore mechanisms other than individuals’ need for self-esteem, to explain and predict the impact of perceived CSR on employees’ OI.

Finally, this study also contributes to a better understanding of the conditions under which the relationship between CSR and organizational trust could be maximized. In this respect, results show that employees’ self-centred CSR attributions moderate the link between perceived CSR and organizational trust. Specifically, our findings suggest that under high self-centred CSR attributions, as perceived CSR increases, the increase in organizational trust is more pronounced than under the condition of low self-centred CSR attributions. This result could partly be explained by the unique position of employees as internal stakeholders of their organization. More specifically, it appears that when CSR perceptions are highly evaluated, employees with a high level of self-centred CSR attributions could more easily consider the company as doing better by doing good. On the contrary, when CSR perceptions are low, and thus badly evaluated, the potential influence of perceived CSR on organizational trust is less important for employee groups with high self-centred CSR attributions, as they may think that the organization is merely greenwashing. These results are consistent with Becker-Olsen et al.’s (2006) study on consumers’ CSR attributions attesting that when a firm is viewed as profit-motivated, there is no systematic reduction in the perceived credibility of the company. Therefore, taking an internal viewpoint, our study support their argument according to which stakeholders’ suspicion and scepticism is not always driven by a firm being profit-motivated, “but rather by a discrepancy between the stated objectives and firm action” (Becker-Olsen et al. 2006, p. 50).

Limitations and Further Research

Despite the implications of the present study, there still remain several limitations opening a number of new directions for future research. First, although we took precautions to address common method variance and social desirability bias (e.g. confidentiality, survey’s structure and instructions, well established scales), all the variables measured come from self-reported questionnaires. Though, all the constructs present in our study reflect individual perceptions and cognitions. In this respect, self-report questionnaires have been argued to be among the most valid and useful data source for assessing individual perceptions, thoughts, feelings and attitudes (Glick et al. 1986; Spector 1994). Also, existing research demonstrated that individuals report organizational events quite accurately (Podsakoff and Organ 1986). Moreover, Harman’s single factor test provides some evidence that common method bias was not a serious issue in our data. Nevertheless, we encourage further research to replicate our findings in a longitudinal design integrating variables from other sources of measure (e.g. supervisor reports) in order to reduce potential bias and further support the causality of our observed relationships.

Further research is also warranted in order to extend the generalizability of our results. Indeed, our study takes place in the specific context of the controversial oil industry sector. The increasing media coverage of environmental scandals (e.g. TEPCO nuclear power station in Fukushima, Total and the sinking of the Erica, BP and the Exxon Valdez oil spill disasters, and so forth) has raised stakeholders’ scepticism and disregard towards organizations’ CSR initiatives in this sector (Frynas 2003, 2005; Handelman 2006; Palazzo and Richter 2005; Yoon et al. 2006). In this respect, the intervening role of organizational trust seems to be more salient than other mediating mechanisms (e.g. PEP) as stakeholders usually have more reasons to doubt the information conveyed by organizations pertaining to such a controversial sector. From another viewpoint, in a socially and well-regarded industry sector it might be observed that, through employees’ self-enhancement process, PEP plays a more preponderant role than organizational trust in mediating the impact of perceived CSR on employees’ OI. Further research should therefore intend to survey simultaneously samples of employees pertaining to both controversial and exemplary industry sectors in terms of societal responsibility in order to test the relative influence of these two potential mediating mechanisms in different contexts. In the same vein, Kim et al. (2010) study conducted in Korea suggested that the potential mediating effect of PEP between perceived CSR and employees’ OI is more salient in East Asian versus Western cultural environments as it seems that in collectivist cultures (i.e. Asian culture), employees’ self-conceptions are potentially more interdependent and less inseparable from the meaning others attribute to the organization (Kim et al. 2010). As such, cross-cultural studies are also warranted to clarify whether culture orientation plays a moderating role in the relationship between PEP and employees’ OI.

Beyond these limitations, we also encourage future research to test other mediating processes explaining the relationship between perceived CSR and employees’ OI. In this respect, organizational pride should be tested as a mediating mechanism together with PEP. Indeed, according to SIT, with the self-enhancement process, employees’ integrate organizations that are well-regarded because this external image bolsters their pride of membership thus satisfying their need for self-esteem and belongingness (Jones 2010). Pride could therefore represent the psychological link between an organization’s PEP and employees’ level of identification. Finally, further research could extend our framework by investigating the effects of other contingency factors on employees’ responses to CSR (Becker-Olsen et al. 2006; Bhattacharya et al. 2009). Specifically, marketing literature has started to explore the potential interactions between perceived CSR and CSR’s fit, or the perceived overlap between CSR initiatives and the organization’s operations, mission and overall positioning (Becker-Olsen et al. 2006; Bhattacharya et al. 2009). In this respect, CSR fit has been argued to moderate the relationship between perceived CSR and customer’s behavioural outcomes (Albert and Whetten 1985; Bhattacharya et al. 2009). In the same vein, Bhattacharya et al. (2009) proposed that some stakeholder-level contingency effects, attributable to individual factors, could also moderate the impact of perceived CSR on stakeholders’ outcomes. More specifically, they suggest that individual preferences and sensibilities regarding CSR issues could interact with stakeholders’ responses to CSR initiatives. Therefore, we encourage future research to further explore these contingency factors to better understand the circumstances under which CSR initiatives maximize employees’ support for their organization’s objectives and performance.

Practical Implications

For managers operating in the oil industry sector, our study suggests that their organization could gain more legitimacy and support from internal stakeholders by increased engagement in environmental CSR initiatives and associated communication efforts. In so doing, our study underlines that CSR initiatives in controversial sectors would be more effective in strengthening the organization–employee relationship if these actions were directed at reinforcing organizational trust rather than organization external prestige, or reputation. Moreover, our study indicates that in order to reinforce the relationship between perceived CSR and organizational trust, an organization initiating its CSR policies and actions should avoid over-emphasising the business case underlying these initiatives. Organizations should rather show benevolence and honesty when launching and communicating about new CSR initiatives in order to maximize the influence of perceived CSR on organizational trust. However, as far as the company gains experience in such societal practices and employees’ perceptions of CSR increase, the business case for CSR appears as much more legitimate in the eyes of internal stakeholders. Only then would organizations in these controversial industry sectors be able to use CSR initiatives to present themselves as doing well by doing good and deflect employees’ negative perceptions about the organization by increasing organizational trust. This should in turn encourage employees to identify more intensively with their organization and to advocate its legitimacy towards external stakeholders (Mitchell 2002). Consequently, by investigating these contingency effects our study supports managers’ efforts to better understand the circumstances under which CSR initiatives could hurt or improve the legitimacy of organizations operating in such controversial sectors.

The present study also has important implications for Human Resources managers who seek to build OI and convince employees to adhere to and integrate the organization’s values and mission. Our research indicates that CSR can have an effect on employees’ identity and strengthen identification because more than any other responsibility (e.g. economic or legal), societal engagement can reveal the values and soul of an organization (Becker-Olsen et al. 2006; Bhattacharya et al. 2009). In this respect, we support that the main objective for Human Resources managers in undertaking and promoting CSR strategies should be to strengthen the organization–employee relationship by designing CSR initiatives that increase organizational trust and employees’ OI. In this regard, initiatives that testify to the benevolence, integrity and competencies of an organization could have a stronger effect in reassuring employees about the organization’s overall character and trustworthiness. Moreover, our study supports research in auto-communication that examines the relationship among CSR, employees’ productivity, and the recruitment of high quality employees (Morsing 2006). In this respect, we encourages a growing integration of CSR communication within the multiple HR processes (e.g. recruitment, career and talent management, or in incentive design) in order to maximize the potential return of CSR on the employee–organization relationship.

Conclusion

This research confirms that OI theory is relevant to understand and predict employees’ attitudinal and behavioural responses to CSR initiatives. In this regards, our study attests that even in the oil industry sector, known as one of the biggest polluter, credible CSR initiatives directed at improving the natural environment can contribute to foster employees’ identification, and thus relationship with their organization. Specifically, findings suggest that in such controversial industry sector it is the reinforcement of employees’ organizational trust, rather than PEP, that best explain and predict the benefits of CSR on employees’ OI and subsequent attitudes. Moreover, our results further suggest that some contingency factors reflecting the organization’s motivation and underlying sincerity in its CSR engagement could interact in the process explaining the impact of CSR on employees’ feeling of belongingness within organizations pertaining to controversial industry sectors. As such, this study hopes to extent the understanding of the strategic contributions of CSR to organizational objectives by highlighting some of the mediating and moderating processes that drive employees’ psychological and behavioural responses to CSR initiatives.