According to Gardner, Avolio, Luthans, May and Walumbwa (2005), authentic leadership can be described as the process whereby leaders are aware of their thoughts and behaviours within the context in which they operate. Authentic leaders are often aware of their own leader and subordinate values, moral perspectives, strengths and knowledge (Avolio & Luthans, 2006). They are regarded as mindful of both their own personal authenticity and the manner in which they allow subordinates to achieve common goals and objectives (Clapp-Smith et al., 2009).

Recent studies suggested that authentic leadership is a ‘higher-order, multidimensional construct, comprised of self-awareness, balanced processing, the internalisation of moral and ethical perspectives and relational transparency’ (Walumbwa et al., 2008, p. 89). Self-awareness refers to leaders’ knowledge of themselves, their mental state and the perceived image they have of themselves (Gardner et al., 2005; Neider & Schriesheim, 2011), whereas balanced processing refers to a leader’s ability to consider and analyse all relevant facts objectively before making a decision (Gardner et al., 2005; Neider & Schriesheim, 2011). Possessing a moral perspective is when leaders rely on their own morals, values and standards to drive their actions, irrespective of external pressures (Gardner et al., 2005; Neider & Schriesheim, 2011). Finally, relational transparency refers to these leaders’ ability to express their true thoughts and motives, facilitating their ability to openly share information (Gardner et al., 2005; Neider & Schriesheim, 2011).

Authentic leadership is focused on a leader’s relationship with his or her subordinates (Avolio & Gardner, 2005; Wong & Cummings, 2009). A fair number of leadership theories emphasise a leader’s behaviours and characteristics; however, very few leadership theories focus on the relationship between leaders and subordinates (Wong & Cummings, 2009). Even though authentic leadership has a significant focus on the relational transparency and self-awareness of leaders, it also focuses on personal and social identification (Wong & Cummings, 2009). Authentic leadership views personal and social identification as processes through which the behaviour of a leader results in self-awareness among the leaders and subordinates (Wong & Cummings, 2009).

A positive moral perspective and balanced processing are important components of authentic leadership (Neider & Schriesheim, 2011). Leaders need to objectively consider all of the facts to engage in ethical and transparent decision-making; therefore, authentic leaders utilise their own moral capacity and resilience to confront and deal with ethical dilemmas and make moral decisions. Making decisions in a fair and moral manner is crucial given the nature of change in social, political and business environments (Sarros & Cooper, 2006). The nature of these environments makes it important to rely on leaders who are genuine and possess moral attributes.

Because of the high moral standards, integrity and honesty displayed by authentic leaders, subordinates may develop positive expectations as well as increased levels of trust and a stronger willingness to cooperate with leaders to the benefit of the organisation (Avolio, Gardner, Walumbwa, Luthans, & May, 2004; Wang & Hsieh, 2013). This is confirmed by Caldwell and Dixon (2010), who found that authentic leaders influence individuals at various levels in an organisation. Authentic leaders thus have a significant impact on both their subordinates and the organisations that they lead.

Although building stronger trust in leaders is required to address the many challenges faced by organisations, trust continues to be low in organisations (Gallup, 2012). To foster trust in supervisors, leadership that impacts the entire organisation in a positive manner is required. To achieve such leadership, loyalty, commitment and the willingness to take risks should be the important characteristics of organisational members. The above-mentioned characteristics can only be achieved if leaders instil extensive trust within their subordinates.

According to Agote et al. (2015), trust in a leader will have an effect on subordinates’ work attitudes and behaviours. Trust can be defined as the disposition of an individual to be vulnerable to the actions of another while believing that the other will conduct a specific action (with good intentions) – this action should be important to the trustor (Clapp-Smith et al., 2009; Roussin, 2008). Trust is also the willing exchange of actions between individuals. This exchange only takes place if the trustor believes that exploitation is unlikely and as a result is willing to display trust behaviours and to risk vulnerability (Eggers, 2011; Ferres, 2003).

The intentions and actions of an individual must be confidently perceived, while the expectation of ethical treatment should also be present for trust to exist (Eggers, 2011; Ferres, 2003). In order for trust to exist within a leader–follower relationship, it is necessary for a subordinate to observe the following characteristics within a leader: open communication, cooperation, willingness to sacrifice, confidence, predictability and fair treatment (Clapp-Smith et al, 2009; Ferres, 2003). As a result, unbiased processing as well as moral and ethical perspectives can be expected to nurture trust within a leader–follower relationship (Miniotaite, 2012).

Leaders can develop collaborative relationships, build credibility and gain the respect of subordinates when they act authentically, thereby building trusting relationships with subordinates (Avolio et al., 2004). A subordinate’s trust stems from judgements of authenticity which are based on consistent leader actions (Coxen et al., 2016). Dirks and Ferrin (2002) suggest that when a subordinate is treated fairly and respectfully, he or she is more likely to display positive attitudes and commitment to a leader.

Studies have suggested that trust plays an important role in the relationship between leadership constructs (e.g. authentic leadership) and follower outcomes (e.g. employee behaviours) (Clapp-Smith et al., 2009; Coxen et al., 2016).

Both psychological safety and trust involve vulnerability or the perception of risk through choices which seek to minimise negative consequences (Edmondson, Kramer, & Cook, 2004). Psychological safety is conceptualised as an individual’s view of the risks and consequences associated with his or her work environment, stemming from a subconscious conviction of how others will respond when an individual finds him/herself in a particular situation (Edmondson et al., 2004; Roussin, 2008). The presence of psychological safety creates confidence in an individual that others will accept and not reprimand his or her actions (Edmondson, 1999).

The difference between psychological safety and trust stems from choice. The trustor’s conscious decisions to trust an individual cannot be a choice to feel psychologically safe, but can be a choice to place his or her trust in someone (Edmondson et al., 2004). Psychological safety is thus defined as an individual’s perception of the consequences of taking an interpersonal risk in his or her job environment, without the fear of negative consequences to his or her image, status or career (Edmondson & Lei, 2014).

The relationship between a supervisor and subordinate has a direct influence on the feeling of psychological safety that the subordinate experiences within the work environment (Edmondson, 1999; Newman, Donohue, & Eva, 2017). When a supervisor supports rather than controls the subordinate, the subordinate will experience a sense of psychological safety. Such supervisors show a sense of concern for their subordinates’ feelings and needs, providing them with positive feedback which not only enables them to develop new skills, but also encourages them to share their opinions without any fear of negative consequences (Edmondson, 1999; Roussin & Webber, 2011).

Employees might be less willing to take risks or express themselves if they perceive these risks to result in negative consequences or even if these risks may lead to embarrassment (Detert & Burris, 2007). Examples of such interpersonal workplace risk include (1) harm resulting from opportunism; (2) identity damage as a result of social interactions and (3) neglect of an individual’s interest by others (Williams, 2007). Trust in supervisors can serve to mitigate these interpersonal risks in the workplace, which could result in increased psychological safety (Ning Li & Hoon Tan, 2012).

An increased experience of psychological safety may result in work engagement (Lyu, 2016), as psychological safety reflects upon the belief that an individual can be engaged without any fear of negative consequences (Edmondson, 1999; Eggers, 2011; Roussin & Webber, 2011). Where a work environment displays ambiguity, unpredictability and is threatening, the opposite would be true as subordinates would perceive the environment as being psychologically unsafe. Subordinates working in a perceived psychologically unsafe work environment may disengage from their work and may be reluctant to attempt new things (May, Gilson, & Harter, 2004).

Authentic leaders utilise their own self-awareness as well as the self-awareness of their subordinates to lead (Avolio & Gardner, 2005). Displaying an awareness of their own strengths and shortcomings as well as those of their subordinates helps them to lead more transparently. They motivate their subordinates by inspiring them and displaying charisma (Eggers, 2011). Authentic leaders further motivate subordinates by being intellectually stimulating and considerate of subordinates’ individuality (Eggers, 2011). Through this, leaders assist their subordinates in developing leadership skills by helping them become more aware of their own feelings, behaviours and thoughts. Leaders and subordinates must be aware of one another’s expectations, needs and wants. This leads to positive change in an organisation through developing psychological safety. Psychological safety and trust will lead to transparency, positive self-awareness, a positive moral perspective and a willingness to continually learn (Eggers, 2011).

Work engagement is defined as the ‘harnessing of organisational members’ selves to their work roles: In engagement, people employ and express themselves physically, cognitively, emotionally and mentally during role performances’ (Kahn, 1990, p. 694). Bakker, Schaufeli, Leiter and Taris (2008) describe work engagement as ‘a positive, fulfilling and effective motivational state of work-related well-being that is characterised by vigour, dedication and absorption’ (p. 187). Vigour refers to increased levels of energy and mental resilience at work, whereas dedication refers to an individual’s involvement and fulfilment in his or her work (Bakker et al., 2008). Finally, absorption refers to an individual’s happiness and concentration at work which allows time to pass quickly (Schaufeli, Salanova, Gonzales-Roma, & Bakker, 2002).

Engaged employees are energetic and experience a feeling of enthusiasm for their work, and as a result, are completely absorbed by their work to the extent that time flies while working (Bakker et al., 2008; Hassan & Ahmed, 2011). The improvement of work engagement levels in the workforce has become critical for organisational success (Harter, Schmidt, & Hayes, 2002; Du Plessis & Boshoff, 2018). This is because work engagement has a positive impact on business, financial and in-role performance, as well as employee productivity (Bakker & Demerouti, 2008; Du Plessis & Boshoff, 2018).

Various studies suggest that leadership is an important factor that positively contributes to work engagement, either directly or indirectly through other constructs (Coxen et al., 2016; Ebrahim, 2017; Harter et al., 2002; Heyns & Rothmann, 2018). Authentic leaders lead by example – they lead through their values and strive for truthful relationships (Gardner et al., 2005; Kernis, 2003). Leading by example illustrates one’s commitment to work and provides guidance to subordinates (Bandura, 1977), allowing subordinates to remain emotionally and physically connected as well as cognitively vigilant in their work roles.

A key element of leadership effectiveness is having trust in supervisors (Coxen et al., 2016). According to Hsieh and Wang (2015), trust in supervisors is linked to favourable organisational outcomes, such as job satisfaction, employee commitment and organisational citizenship behaviour. Studies indicate that trust in supervisors has an indirect effect between the leader’s actions and employee behaviours (Coxen et al., 2016; Hsieh & Wang, 2015). When subordinates perceive their supervisors as being trustworthy, it will positively affect their psychological well-being (Lee, 2017). As a result, these subordinates will experience higher levels of work engagement (Wang & Hsieh, 2013). Furthermore, a subordinate’s desire to voluntarily return authenticity is increased when the subordinate perceives his or her supervisor to have authenticity. This, in turn, creates an environment of trust and dependency which enables subordinates to be engaged and fully immersed in their work (Wang & Hsieh, 2013). Therefore, trust and employees’ attitudes are inextricably linked with leadership being an antecedent.

Authentic leaders’ behaviour stems from their own values and such leaders are driven to display truthfulness and openness in relationships (Gardner et al., 2005; Kernis, 2003). These leaders can be said to lead by example through the demonstration of transparent decision-making (Avolio & Gardner, 2005). Supervisors’ commitment to their work is displayed when leading by example. This provides subordinates with guidance to become cognitively vigilant as well as emotionally and physically connected with their work performance (Bandura, 1977). According to the research of Kahn (1990), leaders have an influence on the levels of work engagement displayed by subordinates. In a psychologically safe environment, an individual feels accepted and supported as well as able to provide input without any fear of negative consequences or social embarrassment (Kahn, 1990).

Based on the above discussion, the hypotheses of this study were formulated as follows:

The hypothesised model is illustrated in Figure 1.

This study was quantitative in nature. A cross-sectional survey design was used as the data were collected at one single point in time.

Mplus 7.4 (Muthén & Muthén, 1998–2016) was utilised to perform the statistical analysis. Descriptive statistics (e.g. means, standard deviations, skewness and kurtosis) and inferential statistics (e.g. correlations) were used for data analysis. Confirmatory factor analysis (CFA) was used to test the factorial validity of the measuring instruments. Raykov’s rho coefficients were used to assess the composite reliability of the measuring instruments and a cut-off value of 0.70 was used (Raykov, 2009). Pearson product-moment correlation coefficients were used to measure the relationships between the proposed variables. Cohen’s effect sizes were used to determine the practical significance of the results, with cut-off values of 0.30 (medium effect) and 0.50 (large effect). A value of 95% (p ≤ 0.05) was set for the confidence interval level for statistical significance.

Structural equation modelling (SEM) was used to test the measurement and structural models. The following fit indices were used: chi-square (χ²), comparative fit index (CFI), standardised root-mean-square residual (SRMR) and root-mean-square error of approximation (RMSEA). Acceptable model fit was indicated by non-significant χ² values, GFI and CFI (≥0.90) and RMSEA (≤0.08) (Byrne, 2012). The Akaike information criterion (AIC) and Bayes information criterion (BIC) were used. The smallest value of the AIC and BIC indicates good model fit (Kline, 2011).

To determine whether common method variance (CMV) influenced the results, Harman’s single-factor test, a common method, was used. This technique is a post hoc test conducted to determine whether a single factor is responsible for variance in the data (Tehseen, Ramayah, & Sajilan, 2017). This test thus aims to determine the presence or absence of CMV (Tehseen et al., 2017).

The bootstrapping method was used to test for indirect effects. The method was set at 5000 draws (Hayes, 2012) and the bias-corrected confidence level (BC CI) was set at 95%. When zero is not in the 95% CI, one can conclude that the indirect effect is significantly different from zero at p < 0.05.

Confirmatory factor analysis was used to estimate the factor structure of the variables. Using SEM, a four-factor measurement model and three alternative models were tested to assess possible relationships between the latent variables.

Model 1 consisted of four first-order latent variables, namely, authentic leadership (measured by 13 observed variables), work engagement (measured by eight observed variables), trust in supervisors (measured by eight observed variables) and psychological safety (measured by three observed variables). All the latent variables were allowed to correlate.

Model 2 consisted of three first-order latent variables, namely, authentic leadership (measured by 13 observed variables), trust in supervisors (measured by eight observed variables) and psychological safety (measured by three observed variables) as well as a second-order latent variable consisting of vigour combined with dedication (measured by six observed variables) and absorption (measured by two observed variables).

Model 3 consisted of three first-order latent variables, namely, work engagement (measured by eight observed variables), trust in supervisors (measured by eight observed variables) and psychological safety (measured by three observed variables) as well as a second-order latent variable of authentic leadership, consisting of self-awareness (measured by three observed variables), relational transparency (measured by three observed variables), balance processing (measured by four observed variables) and moral perspective (measured by three observed variables).

Model 4 consisted of seven first-order latent variables, namely, self-awareness (measured by three observed variables), relational transparency (measured by three observed variables), balanced processing (measured by four observed variables), moral perspective (measured by three observed variables), work engagement (measured by eight observed variables), trust in supervisors (measured by eight observed variables) and psychological safety (measured by three observed variables).

Table 1 presents the fit statistics for the four competing measurement models described above.

Further analyses were conducted in an exploratory mode to improve the fit of the selected model even more. The item errors were allowed to correlate which could improve model fit. According to Byrne (2012), correlated errors could be representative of the respondent’s characteristics that reflect bias and social desirability as well as a high degree of overlap in the item content. The revised model (model 1) indicated that the fit improved once the errors were allowed to correlate. A comparison of the AIC and BIC values indicated that model 1 fitted the data the best with a χ² of 636.23. The fit indices for CFI and Tucker–Lewis index (TLI) were acceptable (>0.90), as well as the model fit for the RMSEA (<0.05). The SRMR for model 1 was 0.05; values lower than 0.08 indicate an acceptable fit.

Table 2 contains the descriptive statistics which include the descriptive statistics (e.g. means and standard deviations), Raykov’s rho reliability coefficients and a correlation matrix.

As per the results in Table 2, it is evident that the Raykov’s rho coefficients of all the measuring instruments were considered acceptable, ranging from 0.70 to 0.94. Raykov’s rho coefficients have the same acceptable cut-off points as Cronbach’s alpha coefficients, recognising values of ≥0.70 as acceptable (Wang & Wang, 2012).

Table 2 further provides the correlation coefficients of the variables which were all statistically significant at either 0.01 or 0.05. Authentic leadership was practically and significantly related to work engagement (r = 0.49) (medium effect); psychological safety (r = 0.45) (medium effect) and trust in supervisors (r = 0.74) (large effect). Work engagement was practically and significantly related to psychological safety (r = 0.37) (medium effect) and trust in supervisors (r = 0.52) (large effect). Trust in supervisors was practically and significantly related to psychological safety (r = 0.39) (medium effect).

The measurement model formed the basis of the structural model. The hypothesised relationships shown in the model were tested. An acceptable fit of the model to the data was found: χ² = 636.23, df = 455, CFI = 0.95, TLI = 0.95, SRMR = 0.05 and RMSEA = 0.04. Table 3 shows the fit statistics and path coefficients of the three models.

In the above calculations, the maximum likelihood estimation with robust standard errors (MLR) estimator was utilised, taking into account the skewness and kurtosis of frequencies. The χ² values for MLR cannot be directly compared (Satorra & Bentler, 1999). Chi-square difference testing had to be done to determine how the χ² would change between the different models. Table 4 shows the difference testing for competing structural models. The results in Table 4 indicate that both models 1 and 2 had a significant p –value, which suggests a significantly worse fit than model 3. Therefore, model 3 was the best-fitting model.

Figure 2 shows the path coefficients estimated by Mplus 7.4 (Muthén & Muthén, 1998–2016) for model 3.

From Figure 2, it is evident that authentic leadership is a significant predictor of both trust in supervisors (β = 0.74; p < 0.01) and psychological safety (β = 0.36; p < 0.01). Hypotheses 1 and 2 were therefore accepted. Authentic leadership thus explains 54% of the variance in trust in supervisors and 21% of the variance in psychological safety. Hypothesis 4 was rejected as authentic leadership was not a significant predictor of work engagement (β = 0.16). This is apparent in Table 3.

Based on the results from Harman’s single-factor test, the fit statistics of loading the model onto one factor were as follows: χ² = 1756.65, df = 464, CFI = 0.66, TLI = 0.63, SRMR = 0.11 and RMSEA = 0.11. The fit statistics show that the model did not fit, which indicates that CMV was not a problem (Tehseen et al., 2017). If there was model fit to the one factor, CMV could have posed a problem to the study.

To determine whether indirect effects were present in this study, the procedure, as explained by Hayes (2012), was used. Bootstrapping was used to construct two-sided bias-corrected 95% CIs to evaluate indirect effects.

Table 5 shows that the indirect effect of authentic leadership on work engagement through trust in supervisors was significant (p < 0.01) and did not include zero. This suggests that authentic leadership did have an indirect effect on work engagement via trust in supervisors. Based on these results, hypothesis 5 was accepted. Trust in supervisors did not have a statistically significant indirect effect on psychological safety; therefore, hypothesis 3 was rejected. Psychological safety did not have a statistically significant indirect effect on work engagement. Therefore, hypothesis 6 was also rejected. All the hypotheses were thus accepted, except for hypotheses 3, 4 and 6.

This study had several limitations. The first limitation of the study was the use of a cross-sectional design, which restricts the determination of causal relationships among the study variables. The second limitation of the study was the use of the convenience sampling approach which could influence the generalisability of the results obtained. Thirdly, the research was conducted during a time of uncertainty in the mining industry of South Africa, which may be perceived as a limitation. Finally, the research was conducted on a single operation in the mining industry; as a result, generalisation of the findings to other contexts may not be possible.

The following recommendations can be made for future research. Firstly, future research should use longitudinal research designs or diary studies to determine the causal relationships among the study variables. Secondly, future research should expand the study to other organisations and industries as well as other provinces because of the fact that each of these factors may pose its own unique set of challenges and may yield a different result. Thirdly, future research may improve on this study by gathering data from additional sources within organisations, over and above the supervisors and subordinates. Fourthly, future research may improve on this study by utilising a mixed-method approach which includes both quantitative and qualitative data collection. This would allow the researcher to establish the authentic style of the leader as well as the prevailing variables and employee outcomes such as trust and psychological safety. The researcher would then be able to mitigate the close relation between constructs such as trust and psychological safety. Lastly, future research could also include other related leadership constructs into the data collection. This would allow the researcher to determine if the outcomes were exclusively related to authentic leadership, to exclude potential outcomes from other positive leadership constructs such as ethical, transformational, leader–member exchange and empowering leadership.

It is important for employees, leaders and the human resources department to understand the impact of authentic leadership on outcomes such as supervisor trust, psychological safety and work engagement. The lack of ownership, strict rules and regulations, as well as other challenges in the mining sector such as economic uncertainty, may have a likely effect on the leaders’ willingness to display authentic behaviours which may have an impact on the feelings of psychological safety, trust and engagement experienced by subordinates. Feeling psychologically safe is important as it decreases ‘barriers to engagement’ (Wanless, 2016, p. 6). The benefit of having an engaged workforce is enhanced employee performance (Markos & Sridrevi, 2010), which is an important factor in the mining industry.

Trust is an important component of organisational interventions. Therefore, it is vital that employees and the organisation understand that the only way to remain viable is to support one another. When an organisation builds an environment of trust, its employees will reciprocate by becoming more engaged in their work. Both the organisation and employees should participate in a give-and-take relationship. This will help both parties feel confident as well as foster a positive work environment which enhances work performance, psychological safety and work engagement. Authentic leadership plays a key role in creating this positive work environment. Leadership development programmes could be designed in the mining sector to develop authentic leaders who could have a positive impact on the experience of trust, psychological safety and work engagement.