Master’s Thesis
The perception of organisational CSR initiatives as an instrument for motivating employees to engage in pro-environmental behaviours
Master’s Programme Communication Sciences Track Corporate Communication
Graduate School of Communication University of Amsterdam
Loek van der Wilt 10285938
Supervisor: Dr. James Slevin
2 Abstract
Awareness of the world’s natural degradation has led to scientific and organisational interest in seeking ways to fight it. Corporate social responsibility (CSR) initiatives are known as a way for organisations to show their environmental awareness. However, the consequences which CSR initiatives can have inside organisations and its contribution to a possible solution for organisations in meeting their CSR objectives and world’s degradation in general, are less familiar. The current study argues the internal influence of CSR initiatives is just as important as external benefits for organisations because positively perceived CSR amongst employees will have the power to positively influence employees’ environmental attitudes and pro-environmental behaviours. Data collected by a cross-sectional survey amongst 170 employees in the Netherlands revealed that employees’ positive perceptions of their organisations’ CSR initiatives directly influences their organisational identification, their pro-environmental attitude and their pro-environmental behaviour at their workplace in a positive way. Besides, the results of this study also demonstrate an indirect effect of perceived CSR by employees on their pro-environmental behaviour in their private lives through their pro-environmental attitude and pro-environmental behaviour at their workplace. The possible effects CSR supporting leaders could have on these relationships were also tested, but not proven. This study contributes to pro-environmental behaviour literature by investigating the effect of perceived CSR by employees on their private pro-environmental behaviour as a first, while demonstrating organisations the importance of strong internal communication of
3 Introduction
During past decades, awareness concerning the environmental damage caused by humans has grown exponentially. Luckily this has also ignited growing motivation to reduce this damage. The worldwide indignation about Trump pulling out of the Paris agreements (Rucker & Johnson, 2017) shows this will hopefully be an exception on the rising attention for sustainability. Organisations are affected by this trend as well because it seems generally accepted that organisations significantly contribute to the environmental degradation (Robertson & Barling 2013, Norton, Parker, Zacher & Ashkanasy, 2015). Therefore, corporate social responsibility (CSR) – as a process in which an organisation embraces responsibility for its actions and encourages a positive impact through its activities for the environment, consumers, employees, communities and all other members of the public sphere who can be considered as stakeholders (McWilliams & Siegel, 2001) – is getting more
important for organisations and their business strategies (Porter & Kramer, 2011).
Organisations’ ignorance of their footprint on the natural en social environment can lead to a bad CSR reputation and less consumer trust (regarding the private sector) (e.g. Leonidou, Kvasova, Leonidou, & Chari, 2013) but CSR involvement can also lead to tangible, financial and reputational benefits (Du, Bhattacharya & Sen, 2007; Arevalo & Aravind, 2017). As organisations nowadays accept this role, either out of sincere interests in the world around them or to improve the corporate reputation (Du, Bhattacharya & Sen, 2010), the questions that rise for organisations are how to accomplish good organisational pro-environmental performance and how to communicate this in a way it produces maximum organisational benefits.
Many studies show concepts at organisational level which are related to organisational environmental performance, such as profitability (Russo & Fouts, 1997), managerial
4
incentives and policy (Kock, Santaló & Diestre, 2012) and organisation’s capacity for change (Judge & Elenkov, 2005).Less research has been done on the antecedents of environmental performance at the individual level, which can be valuable because previous studies point out that engagement of employees in pro-environmental behaviour also has a positive relationship with the environmental performance of the organisation (Paillé, Chen, Boiral & Jin, 2014; Chen, Tang, Jin, Li & Paillé, 2015). Moreover, pro-environmental engagement of employees can have multiple positive impacts for organisations simultaneously. For example Norton et al. (2015) summarize financial performance, leader effectiveness and job satisfaction in a literature review as several benefits which organisations can experience from workplace pro-environmental behaviour by employees. Besides, it is widely suggested that an organisational reputation is influenced by both the experiences of internal and external stakeholders (Gotsi & Wilson, 2001; Gill, 2015). So in creating a strong external CSR reputation, which is seen as important for long-term business benefits (Du et al., 2007), the influence of internal factors cannot be forgotten (Johnston & Everett, 2012). When employees have trust in their
organisation’s CSR ideals and feel aligned with it, they can be strong CSR ambassadors for their company (Johnston & Everett, 2012; Gill, 2015). Vice versa an organisation’s CSR reputation can have positive internal effects, such as a high job satisfaction amongst
employees (de Roeck, Marque, Stinglhamber & Swaen, 2014; Du, Battacharya & Sen, 2015) which is also related to better job performance (e.g. Ziegler, Hagen & Diehl, 2012;
Alessandri, Borgogni & Latham, 2017). In sum: an internally and externally positively perceived CSR reputation and strong pro-environmental behaviours will have the strength to reinforce each other, therefore more research to find predictors of pro-environmental
behaviour is very useful for organisations to further enhance their pro-environmental performance.
5
Now that the internal and external importance of good organisational
pro-environmental performance and a strong organisational CSR reputation are examined shortly, ways of engaging employees in pro-environmental behaviour will be addressed. First it has to be mentioned that the involvement in CSR activities is not completely without risks or
problems for organisations. Obviously CSR activities need investments, which can be a big threshold for smaller organisations (Freisleben, 2011). Additionally, investing in initiatives that are not congruent with the organisational identity can be a danger for organisations. CSR activities which look dishonest can namely damage consumers’ attitudes towards the
organisation (Wagner, Lutz & Weitz, 2009), which most probably works in the same way for the judgement of fairness of CSR actions by employees (Aguilera, Rupp, Williams &
Ganapathi, 2007). A good fit between an organisation and their CSR activities can minimize scepticism and is therefore less dangerous for an organisational reputation (Elving, 2013). Nonetheless, when executed and communicated in the right way, organisational CSR
activities can help organisations engage their employees in pro-environmental behaviour. This study will argue that positively perceived CSR will boost employees’ pro-environmental attitude, which in turn helps engaging employees in pro-environmental behaviours and therefore positively affects the pro-environmental performance of the organisation, while simultaneously influencing employees’ identification with the organisation in a positive way. Not only will the organisational environmental performance rise, the increase of pro-environmental behaviours of employees on and off their workplace will also benefit the natural environment in general. While this study does not measure the externally perceived CSR reputation of organisations, the link between an internal and external organisational reputation and the possibility of employees becoming CSR ambassadors for their organisation (Gill, 2015) also makes positive effects on external organisational CSR reputations plausible.
6
Theoretical and managerial implications about the effects of the independent variable perceived CSR, the antecedents of the dependent variable pro-environmental behaviour and the reciprocal relationship between these two concepts will be based on the analysis of data collected with an online survey. The results will strengthen recent studies which prove that positively perceived CSR by employees has positive effects on their pro-environmental behaviour (Vlachos, Panagopoulos & Rapp, 2014; Tian & Robertson, 2017) but will discredit some that indicate that organisational identification (OI) mediates this relationship (e.g. Tian & Robertson, 2017). Instead the pro-environmental attitude of employees will be brought forward as mediator in the relationship between perceived CSR and pro-environmental behaviour. For the first time expected spill-over effects of workplace pro-environmental behaviours to private pro-environmental behaviours (Nik Ramli & Naja, 2012) will be added to a model like this, to prove it to be affected by perceived CSR. The two tested double-mediation models of the effect of perceived CSR on private pro-environmental behaviours – the first through OI and workplace environmental behaviours, the second through pro-environmental attitude and workplace pro-pro-environmental behaviours – give new insights into the motivations of employees to behave pro-environmentally. In practice this can be used to engage employees more in pro-environmental behaviours at work and at home. For
organisations this means they can make an even bigger impact for the natural environment by positively changing pro-environmental behaviours, while improving their own environmental performance. Because communication plays a big part in spreading a CSR reputation both internally and externally (Du et al., 2010; Gill, 2015), information about communication tools as possible catalysts for the effects of CSR practices is valuable as well. Existing literature shows influence of CSR attitudes and behaviours of managers or supervisors on employee CSR attitudes and behaviours (e.g. Vlachos et al., 2014; Kim, Kim, Han, Jackson & Ployhart, 2017). Therefore the presence of CSR supporting managers or supervisors in organisations, is
7
tested as a moderator on the relationship between perceived CSR and OI and the relationship between perceived CSR and pro-environmental attitudes. A moderating role of CSR
supporting leaders on the effects of perceived CSR on OI and pro-environmental attitude will give insights into how communication can strengthen (or weaken) attitudinal change of employees caused by perceived CSR.
Theoretical framework
The definition for CSR given in the introduction has to be specified because this study focusses on the effect of the perception of CSR by the employees of organisations that perform it. Perceived CSR is used as the independent variable and can be understood as the perceived presence of socially and environmentally responsible practices and policies that aim to enhance the welfare of various stakeholders (Turker, 2009). The concepts that are used in this research as related concepts to perceived CSR will be defined in this section. First the internal effects of CSR are addressed and organisational justice theories will be put forward as a basis theory for understanding the effects of perceived CSR on individuals inside
organisations. Then, respectively the concepts voluntary workplace pro-environmental behaviour, organisational identification, environmental attitude, private
pro-environmental behaviour and CSR supporting leaders will be defined, literature about them will be discussed and their expected relationship with perceived CSR will be formulated in hypotheses.
8
Although research on the effects of CSR related actions on individual organisational members is more recent and less extensive than research about the outcomes on organisational level (Robertson & Barling, 2013), it is proposed that the way employees (as internal stakeholders) judge CSR, influences their emotions, attitudes and behaviours (Rupp, Ganapathi, Aguilera & Williams, 2006). The past years, empirical studies have shown that the way employees perceive the CSR policy of their organisation has interesting effects. Examples are the effects on the attractiveness of an organisation, organisational commitment, intentions to stay, organisational citizenship behaviours (Rupp, Shao, Thornton & Skarlicki, 2013), and more specifically on the extent to which employees engage in voluntary CSR behaviour (Vlachos et al., 2014).
To understand the effects of perceived CSR for employees the basics of organisational justice theory have to be addressed. Organisational justice theories pose that to what extend individuals within an organisation judge the social actions of the organisation as fair, will influence their attitudes and behaviours towards the organisation (Cropanzano, Byrne, Bobocel & Rupp, 2001). This need for fairness arises from instrumental (e.g. economic benefits and the need for control), relational (i.e. need for belongingness) and moral (i.e. need for a meaningful existence) motivations (Cropanzano et al., 2001; Rupp et al. 2006). Based on models of organisational justice, CSR actions of an organisation can be seen as an instrument for employees to judge the fairness of their organisation to fulfil their needs (Aguilera, Rupp, Williams & Ganapathi, 2007).Employees judge the organisation’s actions, the outcome of those actions and the way people, outside and inside the organisation, are treated
interpersonally as a result of these actions (Rupp et al., 2006). How they judge CSR and how this satisfies their needs influences their emotions, attitudes and behaviours (Rupp et al., 2006). Employees are influenced by both how fairly they are treated themselves (first party perspective) by an organisation and how the organisations treats external groups and
9
environments (third party perspective)(Rupp et al., 2013). Note that the way perceived CSR is operationalized in this study does not measure the perception and judgement of fair treatment of individuals inside organisations. So it is investigated how perceived CSR raises a feeling of third-party justice under employees and how this influences their attitude and behaviour.
This power of CSR to change employee attitudes and behaviours asks for new insights in its effects on the pro-environmental behaviour of employees. To investigate if the effect of CSR reaches even further than the pro-environmental behaviours of employees at the
workplace like proven by Vlachos et al. (2014) and Tian and Robertson (2017), private environmental behaviour is added to the model as dependent of voluntary workplace pro-environmental behaviour. With two double mediated models this study tests whether
perceived CSR influences employees’ engagement in voluntary pro-environmental behaviour at their workplace and at home, and which mechanisms enable this influence. Organisational identification and pro-environmental attitude are tested in separate double mediation models as possible links in this chain, because according to literature they can logically be placed between perceived CSR and workplace pro-environmental behaviour by employees. CSR supporting leaders are tested as possible communication ‘tool’ in both models. By uncovering the mechanisms which work between perceived CSR and pro-environmental behaviour, more knowledge of how to motivate employees for engaging in pro-environmental behaviour is gained.
Direct effects of perceived CSR
Perceived CSR and workplace pro-environmental behaviour
In the literature a distinction is made between two classes of individual CSR performance: in-role specific performance and extra-in-role specific performance. In-in-role
CSR-10
specific performance refers to CSR-related behaviours which are formally prescribed by organisations and are therefore expected and mostly directed by the organisation (Vlachos et al. 2014). Voluntary pro-environmental behaviour of employees is a type of extra-role CSR-specific performance, which is described in literature as behaviours that originate within the individual but also support the effective functioning of a CSR program (Vlachos et al. 2014; Tian & Robertson, 2017). In this research the definition of voluntary workplace
pro-environmental behaviour (workplace PEB) is used to measure the activities employees undertake at their workplace that harm the environment as little as possible or contribute to the environment, and that are not obligated by their organisation. It is expected that voluntary pro-environmental behaviour of employees is positively influenced by perceived CSR. This expectation can be underpinned with the individual motivations that derive from perceived CSR actions according to organisational justice theory (e.g. Rupp et al., 2006). Previous studies have already proven that positive CSR judgements positively influence organisational citizenship behaviours by employees (Rupp et al., 2013), their own initiatives for
pro-environmental innovation in the workplace (Ramus & Steger, 2000) and their extra-role CSR performance in specific (Vlachos et al., 2014).
H1: Perceived CSR positively influences the voluntary workplace pro-environmental behaviour of employees.
Perceived CSR and organisational identification
The concept of organisational identification can be explained as the degree to which a member defines him or herself by the same attributes that he/she believes define the
organisation (Dutton, Dukerich & Harquail, 1994). It finds its existence in the theory of social identity, which states that people have the need to highlight their distinctiveness in
11
interpersonal contexts (Dutton et al., 1994). Individuals tend to categorize themselves and others in social groups, like organisational membership, so they can order the social environment and locate themselves in it (Mael & Ashforth, 1992). The organisational perceptions of employees and how they think outsiders see their organisation can enhance their self-continuity, distinctiveness and self-enhancement when perceived positive (Dutton et al., 1994). Individuals will identify themselves with an organisation if the organisations’ internal and external reputation will help them to fulfil their desires to enhance their self-concept and self-esteem needs (Farooq, Rupp & Farooq, 2017).
For employees, perceived CSR is an important influencer of the evaluation of the organisation (Farooq et al., 2017) and identification with the organisation (Jones, 2010). The strong influence emanates from the help a positive CSR reputation offers employees to meet their psychological needs (Aguilera et al., 2007). Both externally (by enhancing perceived prestige) and internally (by enhancing perceived respect) a positive CSR reputation can lead to organisational identification of employees (Farooq et al., 2017). Employees can use their perception of the fairness of CSR policy to judge the fairness of their organisation (Aguilera et al., 2007). Perceived fairness of the CSR policy of organisation subsequently leads to stronger organisational identification (De Roeck, El Akremi & Swaen, 2016). Therefore the following is expected:
H2: Perceived CSR positively influences the organisational identification of employees.
Perceived CSR and pro-environmental attitude
When an organisation is judged positively on social and environmental actions, this can satisfy employees’ need for a meaningful existence (Rupp et al., 2006; Aguilera et al., 2007). The perceptions of CSR actions by employees goes further than the fulfilling of self-focused
12
needs because it is about how the organisation treats the environment and others (third-party perspective). The way organisations help to fulfil employees’ needs (e.g. the need for
meaningful existence) influences their attitudes towards their organisation (e.g. Jones 2010; Rupp et al., 2013; Farooq et al., 2017). The moral standards employees use to judge CSR actions of organisations (Aguilera et al., 2007) of course also exist outside their professional life. These moral standards employees use to judge organisations’ CSR actions seem likely to be confirmed and strengthened when their organisation acts like them. An additional
argument for the strengthening of moral standards can be found in the cognitive dissonance theory which implies that when a person simultaneously has two cognitions (thoughts or beliefs) that are psychologically inconsistent, a state of dissonance will occur (Festinger, in Nik Ramli & Naja, 2012). A consequence is that people will try to reduce this dissonance, for instance by attitudinal change (Nik Ramli & Naja, 2012). With this in mind, it seems logic that in the case of perceived pro-environmental actions of organisations this will positively influence the pro-environmental attitude of employees, because an non-pro-environmental attitude while their organisations show them pro-environmental actions, would cause a state of dissonance. No direct link between perceived CSR and the pro-environmental attitude as defined in this research are found in previous literature. However, a recent study by
Wesselink, Blok and Ringersma (2017) proves a positive relationship between perceived organisational support for the environment and employees’ intention to act
pro-environmental.
H3: Perceived CSR positively influences the pro-environmental attitude of employees.
Organisational identification and pro-environmental attitudes as predictors of workplace pro-environmental behaviour
13 Organisational identification and pro-environmental behaviour
Organisational identification is a known predictor of employees’ attitudes towards their organisation (Jones, 2010). Previous studies have revealed links between organisational identification and the behaviours of employees. For example, positive effects have been found between the organisational identification of employees and their creative efforts at the
workplace (Brammer, He & Mellahi, 2015), fair and kind treatment of co-workers and exerting extra effort (Farooq et al., 2017). These studies suggest employees will perform pro-organisational behaviours when they feel aligned with the values of their organisation. An explanation can be found in that the differences between employees’ own interests and that of their organisation reduce when they see the values of their organisations as part of their own (Ashforth, Harrison & Corley, 2008). Strong organisational identification can lead to
employees “enacting the organisation’s identity” (Ashforth et al., 2008, p.337). A relationship between organisational identification and extra-role CSR behaviour of employees in specific was also proven recently (Tian & Robertson, 2017). It is therefore expected that:
H4: Organisational identification positively influences employees’ voluntary workplace pro-environmental behaviour.
Pro-environmental attitude and workplace pro-environmental behaviour
According to the theory of planned behaviour, behavioural intention is a really strong
predictor for future behaviours (Ajzen, 1991). Intention for behaviour in its turn, is formed by three components: attitude towards the behaviour, perceived social pressure to perform the behaviour and whether an individual controls estimated skills and resources to perform the behaviour (Ajzen, 1991). The personal considerations attitude and estimated control, seem in
14
many cases even dominant in predicting intentions over the perceived social pressure (Ajzen, 1991; Harland, 1999). Also in the prediction of pro-environmental behaviour, attitudes seem an important predictor (Harland, 1999) and therefore the theory of planned behaviour is still relevant to predict pro-environmental behaviour of employees. Recent literature about the subject underpin this argument with empirical evidence showing employees’
pro-environmental attitudes as having a direct effect on their pro-pro-environmental behaviour at the workplace (e.g. Bissing-Olsen, Iyer, Fielding & Zacher, 2013; Manika, Wells, Gregory-Smith & Gentry, 2015; Wells, Taheri, Gregory-Smith & Manika, 2016; Wesselink et al., 2017). The expectation is that this relationship will be confirmed again.
H5: Pro-environmental attitude positively influences employees’ voluntary workplace pro-environmental behaviour
Spill-over effects of workplace to private pro-environmental behaviour
Cognitive dissonance theory can in addition to explaining changes in attitude, also explain the basis of changes in behaviour. A drive is created to avoid the stage of cognitive dissonance when performing actions that can support beliefs or attitudes (Thogersen, 2004). In the case of environmental concern, a drive should be created to perform more actions that are positive for the environment and therefore decrease the actions that contradict it (Thogersen, 2004).
Cognitive dissonance playing a role in changing people’s pro-environmental behaviour has previously been demonstrated with experiments (e.g. Dickerson, Thibodeau & Aronson, 1992), and recently supported again (Priolo et al., 2016). With the basic principles of the cognitive dissonance theory it is possible to argue that when employees that are focussed on minimizing environmental impact of their organisation at their workplace, this will impact
15
their personal environmental attitudes as it clashes to pursue different environmental beliefs at the workplace and at home (Nik Ramli & Naja, 2012). Additionally, work is a highly social activity whereby the feeling of belonging to a group and the inter group behaviour can influence the behaviours of an individual (Tajfel, 1974). With the workplace as one of the primary settings of almost every adult life it is a space were attitudes and behaviours can be shaped (Salancik & Pfeffer, 1978). As group interaction has also been demonstrated to be useful to promote environmental behaviour (Duke, 2010), group interaction about pro-environmental behaviour at the workplace expectedly has influence on the pro-pro-environmental behaviours of employees at home as well. It is expected that workplace pro-environmental behaviour of employees ‘spills-over’ to their private lives, resulting in the performance of private pro-environmental behaviour (private PEB):
H6: Employees’ voluntary workplace pro-environmental behaviour influences employees’ private pro-environmental behaviour.
Indirect effect of perceived CSR on private pro-environmental behaviour
In the previous paragraphs, theoretical support for the direct relations of perceived CSR on several dependant variables have been provided. All these variables can be placed in a model between perceived CSR an private pro-environmental behaviour which will prove perceived CSR as a possible solution for engaging employees in more pro-environmental behaviour. With two double mediation models the indirect effect of perceived CSR on private pro-environmental behaviour through organisational identification, pro-pro-environmental attitudes and workplace pro-environmental behaviour is investigated.
16
Employees are stakeholders whose beliefs about their organisation’s reputation play a prominent role in identification theory, which makes organisational identification particularly useful to understand employee responses to CSR (Jones, 2010). In the past years several studies showed organisational identification as concept that (partially) explained the effect that perceived CSR has on employee behaviour. For example, studies support the mediating role of organisational identification between: the effect of CSR perceptions on
pro-environmental behaviours (Norton et al. 2015; Raineri & Paillé, 2016), the effect of CSR on employee creative effort (Brammer et al., 2015), the effect of CSR perceptions on
organisational citizenship behaviours (Rupp et al., 2013) and the effect of perceived CSR on voluntary pro-environmental behaviour (Tian & Robertson, 2017). Additionally the expected effect of voluntary workplace pro-environmental behaviour on private pro-environmental behaviour (H6), legitimizes the expectation that perceived CSR will have an indirect effect on private pro-environmental behaviour of employees through double mediation by
organisational identification and workplace pro-environmental behaviour (first double mediation model).
H7: Perceived CSR is indirectly positively related to employees’ private pro-environmental behaviour through organisational identification and voluntary workplace pro-environmental behaviour.
The expected direct effects from perceived CSR on pro-environmental attitude (H3), pro-environmental attitude on workplace pro-environmental behaviour (H5) and from workplace pro-environmental behaviour on private pro-environmental behaviour suggest a second double mediation model. An indirect effect of perceived CSR on private pro-environmental behaviour of employees through pro-pro-environmental attitude and workplace
17
pro-environmental behaviour is expected. Pro-environmental attitude is not found as a
mediator between perceived CSR and pro-environmental behaviour in previous literature, but seems really plausible:
H8: Perceived CSR is indirectly positively related to employees’ private pro-environmental behaviour through pro-environmental attitude and voluntary workplace pro-environmental behaviour.
The moderating role of CSR supportive leaders
It is important to also investigate managerial influence on the antecedents of
pro-environmental behaviours, because this can give managers information about how strong internal communication can strengthen effects of perceived CSR. Possible influence of CSR supportive managers or supervisors can enable managers to use this as a communication tool for engaging employees in pro-environmental behaviour. Both supervisors that belong to the top management and every supervisor further down in the organisation can be defined as CSR supportive leaders, as long as the employees notice their proactive attitude towards CSR.
The social information processing (SIP) theory asserts that work attitudes and behaviours are strongly influenced by the processing of information from the social environment, even stronger than individual characteristics (Salancik & Pfeffer, 1978). The social learning theory complements the SIP theory because it further focusses on the change of behaviours by the influence of the social environment (Kraus, Ahearne, Lam & Wieseke, 2012). According to social learning theory the influence on behaviours of employees at the workplace can be the results of the direct learning from experiences and learning through modelling of the behaviour of others (Bandura, in Vlachos et al., 2014). Both approaches (SIP
18
and social learning theory) suggest that subordinates see their superiors as important social referents and therefore are likely to develop attitudes and perform behaviours that copy that of their superiors (Kraus et al., 2012). This mechanism of influence across hierarchical levels is also known as the ‘trickle-down effect’ (Vlachos et al., 2014).
The judgement of CSR activities of an organisation is also argued to be socially
constructed inside organisations, for example by role models and influential persons (Aguilera et al., 2007), which raises the expectation leaders can be important influencers for the CSR perceptions of employees. Existing studies that did find support for claims about managerial pro-environmental behaviour on employee pro-environmental behaviour exist. For example a direct relationship between the green behaviours of leaders and the green behaviour of
subordinates at an individual level has been demonstrated (Kim et al., 2017), just as a positive relationship between the in-role CSR specific behaviour of managers and the in-role CSR specific performance of employees (Vlachos et al., 2014).
It is attempted to find a moderating effect of the influence of perceived CSR
performance of leaders on the relationships of perceived CSR on organisational identification and environmental attitudes. The strengthening of these effect on antecedents of pro-environmental behaviours, will indirectly prove more influence of CSR supporting leaders on the pro-environmental behaviour of employees. Kraus et al. (2012) prove that managers can enhance organisational identification of their employees. Also, Robertson and Barling (2013) provide empirical proof that leaders’ pro-environmental leadership style and workplace behaviours positively influence employees’ emotions and motivations towards pro-environmental behaviours. Hence the assumption is made that the presence of a CSR supporting manager positively moderates the effects on employees’ organisational identification and on their pro-environmental attitudes. The following hypotheses are formulised:
19 H9: The perceived presence of CSR supporting leaders will positively moderate the effect of perceived CSR on organisational identification (H2).
H10: The perceived presence of CSR supporting leaders will positively moderate the effect of perceived CSR on pro-environmental attitude (H3).
Figure 1 Conceptual model (all hypotheses expect positive effects)
Method section
20
A cross-sectional online survey was used to operationalise all variables (see appendix II for the complete survey). Because perceived CSR, organisational identification and the presence of CSR supporting managers/supervisors are variables that have to be perceived at an
individual level, a survey is a solid measurement instrument. A survey is self-reported and therefore reflects opinions accurately. Employees can perceive CSR initiatives and their importance differently, which is dependent for employees to eventually engage in behaviours related to the CSR initiatives (Singhapakdi, Vitell, Rallapalli & Kraft, 1996; Groves & LaRocca, 2011). Additionally CSR actions and CSR supporting managers/supervisors can only influence behaviours if employees themselves experience and evaluate them (Brammer et al., 2015). The use of a self-reported survey for all variables in this study does mean that the possibility of the occurrence of common method bias has to be taken into account (Podsakoff, MacKenzie & Podsakoff, 2003). In the opening letter and the consent message (see appendix II) anonymity and the possibility to stop the survey at any time were promised to keep the social desirability bias low (Richman, Kiesler, Weisband & Drasgow, 1999).
All five main variables which were included in the conceptual model are measured on 7-point scales with the answers ranging from ‘completely disagree’ to ‘completely agree’ which makes them ordinal variables. It is however widely accepted that variables measured with Likert scales are interpreted as interval measurement level variables (Jameson, 2004) which make them usable in regression analyses. A 7-point scale has been chosen because it has enough options for respondents to clearly discriminate but the difference between options stay clear (Matell & Jacoby, 1972). It has been indicated that scales with a middle point can cause distortions in results because of social desirability bias (Garland, 1991), but a 7-point scale is in most cases wide enough so the middle-point is not chosen a lot (Matell & Jacoby, 1972)
21 Sample & procedure
An online survey was created in the online survey platform Qualtrics. The survey was distributed online via messages on Social Media platforms Facebook and LinkedIn.
Additionally, organisations were contacted online to spread the survey within their
organisation in exchange for the results of the study and an analysis of the results specifically for the employees within their organisation. Because the survey was distributed on social media, a reasonable part of the respondents will be acquaintances of the researcher. The sample can be defined as a convenience sample. Although a convenience sample can benefit the generalizability because of an heterogeneous sample, it can also lead to invariance of measures (Wheeler, Shanine, Leon & Whitman, 2014). Nevertheless, due to the short time (approximately 2 -3 weeks) that was reserved for data collection, the online distribution was chosen because it has the power to easily reach a lot of people. A disadvantage of this is that the respondence rate was really low as it is impossible to obligate the respondents to
participate.
Two question had an answer option that directed respondents out of the survey because they did not fit the target group. If respondents answered ‘No’ to the question ‘Do you currently have a job’ they were excluded because the research is focussed on the influence on behaviour by a current employer. This question excluded 6 respondents. If respondents answered ‘0 – 6 months’ on the question ‘How long have you approximately been working for this organisation?’ respondents were excluded because employment time under six months is perceived as too short for employees to be genuinely influenced by organisational factors, which is based on the positive influence organisational tenure can have on organisational identification (e.g. Riketta, 2005) and organisational citizenship behaviours (e.g. Ng & Feldman, 2011). This question excluded 10 more respondents.
22
The requests to organisations to spread the survey internally resulted in the
cooperation of the RAVU, which is the regional Ambulance service in the Dutch province Utrecht. A total of 60 respondents indicated RAVU as their employer, which is 35,3% of the total sample. The survey was send to approximately 400 employees of whom 60 answered (15%). Because the RAVU would be responsible for a big part of the respondents it was checked if the respondents would not modify the results heavily. A short preliminary
interview with a region manager showed the RAVU contributes to some external and internal CSR actions, after which the assumption was made that respondents within the RAVU would not differ too heavily from other respondents. Because the RAVU is a public/governmental organisation and the working field of an organisation can possibly influence employees’ perceptions (Groves & LaRocca, 2011; Vlachos et al., 2014), the organisational sector was added as possible control variable in the survey so it could potentially be used during analyses.
A total of 326 respondents started the survey. Data of 140 respondents was removed because the respondents did not finish the survey and data of 16 respondents was excluded because they did not fit the target group. Added up, a total of 156 respondents was removed, after which 170 valid responses remained (N = 170). The sample included 105 males (61.8%) and 65 females (38.2). Respondents were between 21 years and 65 years old, with an average of 39.8 (SD = 13.1). The highest achieved degree of education amongst respondents was distributed as followed: high school/college degree (5.3%), MBO (15.3%), HBO (41.8%), University bachelor degree (7.1%), University master’s degree or higher (30.6%). The sample is highly educated compared to the average in the Netherlands (Onderwijs in Cijfers, n.d.). The time respondents indicated as being employed at their current organisation was
distributed as: 6 – 12 months (17.1%) for 1 – 2 years (18.8%), for 2 – 5 years (14.7%) and most of them already worked at that organisation for 5 years or more (49.4%). The biggest
23
part of the sample – 97 respondents – indicated their organisation worked in the private sector (57.1%). Another 68 respondents said they work in the public sector (40%) and 5 in the voluntary sector (2.9%).
Measurements
Perceived CSR
The independent variable perceived CSR is measured with an 8-item scale. An original scale for perceived CSR from Turker (2009) was used as a starting point because it has recently proven to measure perceived CSR successfully (e.g. Farooq et al., 2017; Tian & Robertson, 2017). Five questions about internal CSR were excluded because this study only wants to measure the perceived external CSR. Also three questions about actions towards consumers were removed because CSR towards customers did not match the definition of CSR which is used in this study. Two items from the original scale from Turker have been removed, because they did not seem to fit the definition of CSR that is used here, were not in line with pro-environmental behaviour or the questions ask for knowledge that most employees
probably do not have any knowledge of (e.g. ‘Our company always pays its taxes on a regular and continuing basis’). Farooq, Payoud, Merunka & Valette-Florence (2014) previously added the question ‘Our company gives adequate contributions to charities’ because nowadays charity gifts plays a role in external CSR of organisations (Maignan and Ferrell, 2000). The scale still showed reliable afterwards (Farooq et al., 2014) and was therefore also added to this survey. This leaves a total of 8 questions in the scale (see appendix II).
A principal component analysis (PCA) showed all variables load on Perceived CSR (lowest factor loading: .62). The total scale showed to be reliable (α = .86). Reliability could not be improved by omitting items. The first component had an Eigenvalue of 4.08, the
24
second of 1.22 and they explained 66.25% of variance together. After a Varimax rotation four items loaded on environment-targeted CSR (lowest factor loading .78) and four items loaded on community-targeted CSR as expected (lowest factor loading .70).
Organizational Identification
For the measurement of organisational identification a six item-scale developed by Mael and Ashforth (1992) was used. This scale has had highly homogenous measurements of
organisational identification in different studies and is demonstrated as a strong scale for measuring extra-role behaviours that result from organisational identification (Riketta, 2005). The scale has also been used and proven reliable in recent studies (e.g. De Roeck et al., 2016; Tian & Roberston, 2017). It was concluded that the PCA showed clearly one scale, after observing only one Eigenvalue above 1 (Eigenvalue 3.48) and the scree plot. This factor explains 65,78% of the variance. The scale has good reliability (α = .89).
Pro-environmental attitude
An 8-item scale created by Manika et al. (2015) was used to operationalize general
environmentally friendly attitudes. All items in the scale were phrased negatively and were recoded before analyses. PCA with Varimax rotation showed five items that loaded the strongest on component 1 (Eigenvalue 4.09) and three items that loaded the strongest on component 2 (Eigenvalue 1.07). The two components could not be clearly labelled at first glance. Because the scree plot showed a clear point of inflexion after component 1 and all items correlate strongly with component 1 without rotation (lowest factor loading .60), the scale was accepted as measurement instrument of pro-environmental attitude. The scale showed to be reliable (α = .86) and could not be strengthened if items were to be deleted.
25 Voluntary workplace pro-environmental behaviour
The 12 item ‘Organizational Citizenship Behavior Toward the Environment Scale’ of Lamm, Tosti-Kharas and Williams (2013) was taken as an example for this scale because the
definition of environmental organisational citizenship behaviour matched the definition that was used for voluntary workplace pro-environmental behaviour. To keep the questionnaire short, some questions that were pretty similar to others (e.g. I am a person who turns off the lights in a vacant room) and questions which were expected not suitable for each respondent (e.g. I am a person who makes sure all of the lights are turned off if I am the last to leave.) were left out. This eventually left a 8 item scale (see appendix II).
A PCA showed only one component overtopping an Eigenvalue of 1 (Eigenvalue 3.07). Also the scree plot clearly showed one factor, which explains 38.32% of the variance. The scale was reasonably reliable (α = .76), which could be increased with .01 by omitting the item ‘I am a person who prints double-sided’. Because the difference was quite minimal and this item was adopted from the already proven scale by Lamm et al. (2013), it was decided to keep the item in.
Private pro-environmental behaviour
To operationalise private pro-environmental behaviour three items that could be applicable to the daily life of employees were copied of the scale used for voluntary workplace
pro-environmental behaviour (e.g. ‘I am a person who recycles used paper’). Two questions from a scale of Manika et al. (2015) were added (e.g. ‘I add or remove clothing rather than turning heating or air conditioning up when it’s hot or cold’). And two questions that contain topics that are becoming a bigger part of living pro-environmental were added (e.g. ‘I am a person who (sometimes) eat less meat, because I know it is better for the environment’). It left a 8 item scale (see appendix II).
26
PCA with Varimax rotation showed two components with an Eigenvalue above 1. They were indicated as low-intensity pro-environmental behaviours (Eigenvalue 2.83) and really low-intensity pro-environmental behaviours (Eigenvalue 1.33). Five items loaded best on component 1 and three items on component 2. Because without rotation all eight items loaded pretty well on component 1 (lowest factor loading .54) and analysis of the scree plot showed a clear point of inflexion, the scale was accepted for measuring private
pro-environmental behaviour. The scale showed reasonable reliable (α = .74).
CSR supporting leaders
To operationalise the variable CSR supporting leaders, three questions which were used earlier by Vlachos et al. (2014) and turned out to be reliable as a scale to measure extra-role CSR specific performance, were adapted in a way that they were suited to
managers/supervisors. A PCA showed that three items only loaded strongly on one
component (Eigenvalue 2.70) which explains 90.04% of the variance. The scale is assuredly reliable (α = .95).
Control variables
Although not all literature showed to be conclusive on the influences of basic demographic variables such as age, gender and education on environmental concern (Park et al. 2013), demographics were measured as control variables because previous studies also mentioned possible links to the main variables of this study. For example women have shown to be more reactive to CSR (Greening & Turban, 2000) and age is linked to organisational identification before (Riketta, 2005). Lamm et al. (2013) suggest potential associations between age, gender, education and workplace pro-environmental behaviours. Gender was operationalised by asking ‘What is your gender?’ with the answers-options ‘male’ and ‘female’. Age was
27
operationalised with the question ‘What is your age?’, which respondent could answer by a slide bar reaching from 1 – 100. Education was operationalised as ‘What is your highest achieved degree of education’ with five possible answers reaching from ‘High school/College degree’ to ‘University master degree (WO) or higher’. The sector in which an organisation is active is considered as control variable before (Groves & LaRocca, 2011; Vlachos et al., 2014) and was therefore added as a question in the survey as well. Respondents could answer private sector, public sector and voluntary sector on this question.
Choice of analysis
For the quantitative analyses the statistic analyse program SPSS has been used. Principal component analyses (PCA’s) were used to examine validity and reliability. The outcomes of the PCA’s have already been discussed in the method section, so will not be addressed again. A correlation matrix was constructed and a one-way analysis of variance (ANOVA) was performed to check which of the possible control variables that were operationalized in the survey based on existing literature, showed correlation with the main variables (Field, 2009, p224). Multiple regression analyses were used to separately test H1 – H6. To test the
mediated effects (H7, H8) PROCESS model 6 (Hayes, 2013) was used within SPSS. To test the moderated effects (H10, H11) PROCESS model 1 (Hayes, 2013) was used. The outcomes of the tests for control variables and the hypotheses testing will be addressed in the results section below.
Results
28
Before starting the analyses, data was cleaned by omitting data from the 156 respondents which showed incomplete results, as has been explained in more detail in the method section. The data was manually screened for unusual results. Some respondents with the same score on all items of a scale were found. After checking the variance in the answers of these
respondents on other scales and consideration of possible explanations, it was decided no data had to be removed because of this.
For the control variable ‘organisational sector’ the answers of respondents were checked to see how it could best be recoded for analysis. It turned out that the option ‘Voluntary sector’ was only picked 5 (2,9%) times within the sample. Inspection of the answers of these respondents showed 2 of them assumedly misinterpreted this question because the organisation which they mentioned as their employer were actually in the public sector (both RAVU). Because this made the voluntary sector group even smaller and because the definition of the voluntary sector is more closely related to the public sector than that of the private sector, it was decided to combine the ‘public sector’ group and the ‘voluntary sector’ group into the group ‘public sector’ (n = 73).
In order to assess if variables could causemulticollinearity, all main variables and control variables were compared in a correlation matrix. The means, standard deviations and the correlations of all variables are displayed in Table 1. For the dichotomous control
variables gender and organisational sector, dummy variables were created, to include them in the correlation matrix and for the use in regression and PROCESS analyses later on. The ordinal variable highest level of education (five categories) was also added in the correlation mix to have a first look at its possible correlations, but was also tested with a one-way ANOVA. The correlation matrix showed there was no multicollinearity between any of the variables because no Pearson’s R score was bigger than 0.8 (Field, 2009, p224) (see table 1).
29
The matrix also shows the control variables gender, organisational sector, level of education and age have some significant correlations with each other, but less with the main variables (see table 1). Gender had a significant correlation with voluntary workplace
pro-environmental behaviour and age with pro-pro-environmental attitude (see table 1). Because a one-way ANOVA showed no significant effect of level of education on any main variables (Table 15 and 16 in appendix I), just like the correlation matrix (see table 1), it is decided to drop education as control variable. The other assumed control variables are used in all the regression analyses as control variables and in all tests conducted with PROCESS as covariates, because they showed a significant correlation with at least one main variable and/or are suggested as possible influencers of the main variables in literature.
Hypotheses testing
Direct effects
Table 1 Means, Standard deviations, correlations of all main variables and control variables (N = 170)
M SD 1 2 3 4 5 6 7 8 9 10 1. Gender a .38 .49 - 2. Organisational Sector b .43 .50 .12 - 3. Level of education c 3.42 1.22 .03 -.24** - 4. Age 39.80 13.12 .07 .44** -.40** - 5. Perceived CSR 4.22 1.17 -.13 .13 .12 -.06 - 6. Organisational Identification 4.69 1.23 -.13 .04 .01 -.02 .28** - 7. PE attitude 5.35 .96 .12 .05 .12 .15* .21** -.05 -
8. Voluntary workplace PE behaviour 4.96 1.08 .15* .14 -.07 .12 .22** .05 .51** -
9. Private PE behaviour 4.94 1.06 .10 .02 -.03 .06 .10 -.01 .58** .72** -
10. CSR supporting supervisors 3.66 1.48 -0.5 -.01 .14 -.08 .54** .24** .14 .20* .16* -
Note. * significant at <.05 (two-tailed) ** significant at <.01 (two-tailed)
Variables 5 – 11 are measured on a 7-point scale (1 = Completely disagree, 7 = Completely agree), a Gender: 0 = Male, 1 = Female, b
Organisational Sector: 0 = Private Sector, 1 = Public Sector, c Level of education is an ordinal variable (1 = High school/college, 5 = University
master or higher)
30
A total of 6 multiple regression analyses were conducted to test the direct effects (H1 – H6). For all tests the autocorrelation of residuals and heteroscedasticity were checked by looking at the residual scatterplots and normality plots. No irregularities were found, so the regression analyses could be performed. In all hierarchical regression analyses the control variables gender, organisational sector and age were added in de first step, in the second step the dependent variable was added. In none of the regression analyses the control variables had a significant effect on the dependent variable (see model 1 in tables 2-7).
A regression model with perceived CSR as independent variable significantly explained 10% of the variance in voluntary workplace PEB (see table 2). Perceived CSR turned out to have a moderately strong positive relationship with the voluntary workplace PEB of employees (b* = .25, p < .01, see table 2). This result supports H1.
Table 2 Regression analysis of perceived CSR on workplace pro-environmental
behaviour (H1) (N = 170) Model 1 Model 2 Variable t SE B b* t SE B b* (constant) 16.74 .27 8.28 .42 Gender 1.80 .17 .14 2.29 .17 .17 Organisational sector 1.10 .18 .09 .48 .18 .04 Age .88 .01 .07 1.32 .01 .11 Perceived CSR 3.24 .07 .25** R2 .04 .10** R2 Change .04 .06**
Note. * p <.05. ** p <.01. *** p <.001. The significant direct effect is written in bold
Control variables: gender, organisational sector, age. Dependable variable: Workplace pro-environmental behaviour.
CSR: Corporate Social Responsibility
H2 predicts a positive effect of perceived CSR on OI. Table 3 showed that a model with perceived CSR predicts 9% of the variance of OI. A moderately strong positive effect of perceived CSR on OI was found (b* = .26, p < .01, see Table 3), thereby supporting H2.
31 Table 3 Regression analysis of perceived CSR on organisational identification
(H2) (N = 170) Model 1 Model 2 Variable t SE B b* t SE B b* (constant) 15.71 .31 7.53 .48 Gender -1.75 .20 -.14 -1.30 .19 -.10 Organisational sector .88 .21 .08 .24 .21 .02 Age -.50 .01 -.04 -.07 .01 -.01 Perceived CSR 3.39 .08 .26** R2 .02 .09** R2 Change .02 .06**
Note. * p <.05. ** p <.01. *** p <.001. The significant direct effect is written in bold.
Control variables: gender, organisational sector, age. Dependable variable: Organisational Identification
CSR: Corporate Social Responsibility
A positive effect of perceived CSR on PE attitude was hypothesized in H3. A model with perceived CSR as independent variable predicts 10% of the variance of PE attitude. A moderately strong positive effect of perceived CSR on PE attitude was found (b* = .25, p < .01, see Table 4). This result supports H3.
Table 4 Regression analysis of perceived CSR on pro-environmental attitude (H3) (N = 170) Model 1 Model 2 Variable t SE B b* t SE B b* (constant) 19.96 .24 10.33 .37 Gender 1.51 .15 .12 2.01 .15 .15* Organisational sector -.36 .17 -.03 -.99 .16 -.08 Age .16 .01 .16 2.35 .01 .20* Perceived CSR 3.28 .06 .25** R2 .04 .10** R2 Change .04 .06**
Note. * p <.05. ** p <.01. *** p <.001. The significant direct effect is written in bold.
Control variables: gender, organisational sector, age. Dependable variable: Pro-environmental Attitude
CSR: Corporate Social Responsibility
H4 and H5 respectively expected the OI and PE attitude of employees to have a positive effect on their workplace PEB. An effect of OI on workplace PEB could not be proven with the regression analysis (p = .40, see table 5). Therefore H4 is rejected. A pretty strong positive effect of PE attitude on workplace PEB was found (b* = .49, p < .001) and
32
therefore H5 could be confirmed. The model with PE attitude included as independent variable explained 28% of the variance in workplace PEB (see table 6).
Table 5 Regression analysis of organisational identification on voluntary
workplace pro-environmental behaviour (H4) (N = 170)
Model 1 Model 2 Variable t SE B b* t SE B b* (constant) 16.74 .27 9.95 .43 Gender 1.80 .17 .14 1.90 .17 .15 Organisational sector 1.10 .18 .09 1.03 .19 .09 Age .88 .01 .07 .91 .01 .08 Organisational Identification .85 .07 .07 R2 .04 .05 R2 Change .04 .00 Note. * p <.05. ** p <.01. *** p <.001.
Control variables: gender, organisational sector, age. Dependable variable: Workplace pro-environmental behaviour (PEB)
Table 6 Regression analysis of pro-environmental attitude on voluntary workplace
pro-environmental behaviour (H5) (N = 170) Model 1 Model 2 Variable t SE B b* t SE B b* (constant) 16.74 .27 4.26 .43 Gender 1.80 .17 .14 1.20 .15 .08 Organisational sector 1.10 .18 .09 1.50 .16 .11 Age .88 .01 .07 -.06 .01 -.00 PE Attitude 7.32 .08 .49*** R2 .04 .28*** R2 Change .04 .24***
Note. * p <.05. ** p <.01. *** p <.001. The significant direct effect is written in bold.
Control variables: gender, organisational sector, age. Dependable variable: Workplace pro-environmental behaviour.
PE: pro-environmental
The last direct positive effect that was expected, between employees’ workplace PEB and private PEB, was formulated by H6. The regression model with workplace PEB added as independent variable explained 52% of the variance in private PEB and a strong relationship between workplace PEB and private PEB was proven (b* = .73, p < .001, see table 7). This means H6 could be accepted.
33 Table 7 Regression analysis of voluntary workplace pro-environmental behaviour
on private pro-environmental behaviour (H6) (N = 170)
Model 1 Model 2 Variable t SE B b* t SE B b* (constant) 17.39 .27 4.69 .31 Gender 1.27 .17 .10 -.04 .12 -.00 Organisational sector -.15 .18 -.01 -1.34 .13 -.08 Age .06 .01 .06 .02 .01 .00 Workplace PEB 13.27 .01 .73*** R2 .01 .52*** R2 Change .01 .51***
Note. * p <.05. ** p <.01. *** p <.001. The significant direct effect is written in bold.
Control variables: gender, organisational sector, age. Dependable variable: Private pro-environmental behaviour.
PEB: pro-environmental behaviour
Mediation effects
Although perceived CSR did not show a significant correlation with private PEB in the correlation matrix, the two proposed double mediation models could be tested with a bootstrapping method, since no significant direct effect between an independent and dependent variable is needed (Hayes, 2013, p. 88). Two analyses were conducted in PROCESS macro model 6 with 10,000 bootstraps on a 95% confidence interval. The
confidence intervals that were used to check significance for both analyses are bias-corrected. The first double mediation model proposed an indirect effect of perceived CSR on private PEB through OI and workplace PEB, and is described in H7. Although the model with perceived CSR, OI and workplace PEB as independent variables showed as a strong significant model to predict private PEB (R2 = .53, p < .001, see table 8), the double mediation effect could not be proven because the confidence intervals of the total effect crossed zero (95% CI [-.03 , .03], see table 9). For this reason H7 had to be rejected. However, a single mediation effect between perceived CSR and private PEB, through workplace PEB, could be proven (.17, 95% CI [.06 , .29], see table 9). All correlations between the four variables in the first double mediation model are visualized in Figure 2.
34 Table 8 Double mediation model of perceived CSR on private pro-environmental
behaviour through organisational identification and workplace pro-environmental behaviour (H7) (N = 170)
Variable Estimate SE t p LLCI ULCI
(constant) Gender -.03 .12 -.28 .77 -.28 .21 Organisational sector -.15 .13 -1.11 .27 -.41 .11 Age -.00 .00 -.15 .88 -.01 .01 Perceived CSR -.05 .05 -.90 .37 -.15 .06 Organisation Identification -.03 .05 -.55 .58 -.12 .07 Workplace PEB .73 .06 13.11 < .001 .62 .84 Model summary R2 = .53, F (6, 163) = 30.20, p < .001
Note. Control variables: gender, organisational sector, age. Dependable variable: Private
pro-environmental behaviour.
CSR: Corporate Social Responsibility PEB: pro-environmental behaviour
The outcomes that prove significance are bold.
Table 9 Mediation effects of perceived CSR on private pro-environmental
behaviour through organisational identification and workplace pro-environmental behaviour (N = 170)
Indirect effect Estimate SE 95% CI
Total effect of P CSR on P PEB .11 .07 [-.03 , .25]
Direct effect of P CSR on P PEB -.05 .05 [-.15 , .06]
Total of indirect effects P CSR on P PEB .16 .06 [.05 , .28]
P CSR OI P PEB -.01 .01 [-.04 , .02]
P CSR OI W PEB P PEB .00 .01 [-.03 , .03]
P CSR W PEB P PEB .17 .06 [.06 , .29]
Note. P CSR: Perceived Corporate Social Responsibility, OI: Organisational Identification W PEB: workplace pro-environmental behaviour, P PEB: private pro-environmental
behaviour
The outcomes of significant indirect effects are bold.
Figure 2 The first double mediation model with the indirect effect of perceived CSR on
private pro-environmental behaviour through organisational identification and workplace
pro-environmental behaviour.
Note. * p <.05. ** p <.01. *** p <.001.
Interrupted lines are not significant, solid lines are significant. Control variables: gender, organisational sector, age
35
H8 predicted the second double mediation effect, which proposed an indirect effect of perceived CSR on private PEB through PE attitude and workplace PEB. The model with perceived CSR, PE attitude and workplace PEB as independent variables showed as a strong significant model to predict private PEB (R2 = .59, p < .001, see table 10). A double full mediation effect was proven because the confidence intervals of the total effect did not cross zero (.06, 95% CI [.02 , .12], see table 11). Therefore H8 was accepted. The results of the second double mediation model also displayed the same indirect effect of perceived CSR on private PEB through workplace PEB (see table 11), as the results of the first double mediation model showed. All correlations between the four variables in the second double meditation model are visualized in Figure 3.
Table 10 Double mediation model with the effect of perceived CSR on private
environmental behaviour through environmental attitude and workplace pro-environmental behaviour (H8) (N = 170)
Variable Estimate SE t p LLCI ULCI
(constant) .81 .37 2.20 <.05 .08 1.53 Gender -.08 .11 -.68 .50 -.30 .15 Organisational sector -.08 .12 -.65 .52 -.32 .16 Age -.00 .00 -.94 .35 -.01 .00 Perceived CSR -.09 .05 -1.93 .06 -.19 .00 PE Attitude .34 .07 5.24 <.001 .21 .47 Workplace PEB .59 .06 10.12 <.001 .48 .71 Model summary R2 = .59, F (6, 163) = 39.76, p < .001
Note. Control variables: gender, organisational sector, age. Dependable variable: Private
pro-environmental behaviour.
CSR: Corporate Social Responsibility, PE: pro-environmental, PEB: pro-environmental
behaviour. The outcomes that prove significance are bold.
Table 11 Mediation effects of perceived CSR on private pro-environmental
behaviour through pro-environmental attitude and workplace pro-environmental behaviour (N = 170)
Indirect effect Estimate SE 95% CI
Total effect of P CSR on P PEB .11 .07 [-.03 , .25]
Direct effect of P CSR on P PEB -.09 .05 [-.19 , .00]
Total of indirect effects of P CSR on PEB .20 .06 [.09 , .34]
P CSR PE A P PEB .07 .03 [.02 , .15]
P CSR PE A W PEB P PEB .06 .03 [.02 , .12]
P CSR W PEB P PEB .07 .04 [.00 , .16]
Note. P CSR: Perceived Corporate Social Responsibility, OI: Organisational
Identification W PEB: workplace environmental behaviour, P PEB: private pro-environmental behaviour
36 Moderation effects
The assumptions regarding the moderating effect of the presence of CSR supporting managers of supervisors were described by H9 and H10. The hypotheses were tested with PROCESS model 1 (Hayes, 2013) with 10,000 bootstraps on a 95% confidence interval. The confidence intervals that were used to check significance for both analyses were bias-corrected.
A continuous moderating role of CSR supporting supervisors on the relationship of perceived CSR on OI was not proven (p = .34, see table 12). This means H9 had to be rejected.However, when the level of a CSR supporting supervisor is low, the interaction effect turned out to be a lot bigger than in the medium and high condition (see table 14). This observation combined with a look at the slope plot of the outcome (see figure 5 in appendix I) led to the execution of an additional PROCESS model 1 analysis with CSR supporting leaders recoded into a dummy variable (1 = low CSR supporting leader, 0 = moderate and high CSR supporting leader), which showed a significant interaction effect on the relationship between
Figure 3 The second double mediation model with the indirect effect of perceived CSR on
private environmental behaviour through environmental attitude and workplace pro-environmental behaviour.
Note. * p <.05. ** p <.01. *** p <.001.
Interrupted lines are not significant, solid lines are significant. Control variables: gender, organisational sector, age
37
perceived CSR and OI between these groups (b = .37, p < .05, 95% CI [.02 , .72])(see Figure 4 for simple slope). The moderating role of CSR supporting leaders on the relationship of perceived CSR on PE attitude which is proposed by H10 is not proven (p = .96, see table 13). This means also H10 has to be rejected.
Table 13 Interaction effect of CSR supporting leaders with the direct effect of
perceived CSR on pro-environmental attitude (N = 170)
Variable Estimate SE t p LLCI ULCI
Perceived CSR .18 .17 1.09 .28 -.15 .51
CSR supporting leaders .01 .19 .01 .94 -36 .39
Perceived CSR x CSR supporting leaders .00 .04 .05 .96 -.08 .08
Model summary R2 = .10, R2 Change = .00, F (6, 163) = 2.92, p < .01
Note. Control variables: gender, organisational sector, age. Dependent variable: pro-environmental
attitude
Table 14 Interaction effects of CSR supporting leaders with the direct effects of perceived
CSR (N = 170)
CSR supporting leaders Estimate SE t p LLCI ULCI
Effect of P CSR on OI 2.18 .27 .12 2.20 <.05 .03 .51 3.66 .19 .10 2.04 <.05 .01 .38 5.15 .12 .12 .97 .34 -.13 .37 Effect of P CSR on PE attitude 2.18 .19 .10 1.95 .05 -.00 .38 3.66 .19 .07 .2.53 <.05 .04 .34 5.15 .19 .10 1.98 <.05 .00 .39
Note. Control variables: gender, organisational sector, age.
P CSR: Perceived Corporate Social Responsibility, OI: Organisational Identification W PEB:
workplace pro-environmental behaviour, P PEB: private pro-environmental behaviour The significant outcomes are written bold.
Table 12 Interaction effect of CSR supporting leaders with the direct effect of
perceived CSR on organisational identification (N = 170)
Variable Estimate SE t p LLCI ULCI
Perceived CSR .38 .21 1.78 .08 -.04 .80
CSR supporting leaders .33 .24 1.37 .17 -.15 .80
Perceived CSR x CSR supporting leaders -.05 .05 -.96 .34 -.15 .05
Model summary R2 = .10, R2 Change = .01, F (6, 163) = 3.09, p < .01
Note. Control variables: gender, organisational sector, age. Dependent variable: organisational
38 Conclusion
The main target of this research was finding antecedents of employees’ pro-environmental behaviour on and off the workplace, so more would become clear about what motivates employees to engage in pro-environmental behaviour. Employees’ perception of the CSR actions of their organisation is proposed as an important influencer of their pro-environmental behaviour here. Perceived CSR is also proposed as a ‘safe’ option for organisations to try to enhance the organisational environmental performance, due to its other positive effects on employees and the organisational reputation.
Theoretical implications
Studies investigating desired employee behaviour mainly focussed on their behaviour towards the organisation (e.g. Rupp et al., 2013; Brammer et al., 2015), focussed only on specific behaviours (e.g. water and energy saving, Wells et al., 2016) or only investigated the
pro-Figure 4 Interaction effect of CSR supporting leader on the relationship between perceived
corporate social responsibility (CSR) and organisational identification.
1 2 3 4 5 6 7
Low Perceived CSR High Perceived CSR
Organi sation al id en tif icat ion Moderate/high CSR leader Low CSR Leader
