Supervisory coaching and performance feedback as mediators of the relationships between leadership styles, work engagement, and turnover intention

ABSTRACT

Human resource development (HRD) is key to organizational success. With some HRD roles devolving to leaders in recent years, a gap in understanding is now evident in how leaders’ leadership styles shape development-oriented behaviours that may effectively assist them in fulfilling their HRD roles, and the corresponding effects that this has on employee work engagement and turnover intention. This study compared the effects of transformational and transactional leadership styles on employee attitudes (i.e. work engagement and turnover intention) through leaders’ behaviours (i.e. supervisory coaching and performance feedback). This study used a multilevel approach (i.e. matching leaders to multiple subordinates) with 500 employees, nested in 65 workgroups from private organizations in Malaysia. As hypothesized, we found a link between transformational (but not transactional) leadership and higher levels of supervisory coaching and performance feedback, and that these job resources mediate the relationship between transformational leadership and work engagement. Furthermore, we found that work engagement mediates the relationships of both supervisory coaching and performance feedback to turnover intention. Overall, the study results reveal one way in which Asian leaders can effectively facilitate some aspects of HRD through development-focused behaviours which serve as job resources to boost work engagement and reduce turnover intention.

KEYWORDS:

• Supervisory coaching
• performance feedback
• transformational leadership
• transactional leadership
• work engagement

Introduction

Human resource development (HRD) is defined as ‘the field of study and practice responsible for the fostering of a long-term, work-related learning capacity at the individual, group and organizational level of organizations’ (Watkin 1989, 427), particularly in relation to employee learning and development (Werner and DeSimone 2011). Leadership has been shown to play a more influential role in employees’ behaviour compared to other sources within the organization (Lapointe and Vandenberghe 2017). As a result, some aspects of the HRD role have devolved to leaders in recent years (Alfes et al. 2013). Leadership styles, in particular, transformational leadership, have been considered in relation to HRD functions (Gong, Huang, and Farh 2009; Scandura and Williams 2004), highlighting the importance of leaders in employee learning and development.

Supervisory coaching and performance feedback are pivotal leader behaviours that help organizations to create a competitive advantage (Albrecht et al. 2015). It has been suggested that supervisory coaching is at ‘the heart of managerial and leadership effectiveness’ (Hamlin, Ellinger, and Beattie. 2006, 328), mainly through daily routine interactions between leaders and their followers (Ellinger and Kim 2014). In the employee development aspect, performance feedback contributes to individual and organizational effectiveness (McCarthy and Garavan 2006). For employees, performance feedback allows them to reach a higher level of understanding of their job requirements and enhances their knowledge and abilities in effectively carrying out tasks (Sommer and Kulkarni 2012). While both behaviours relate closely to leader–member interaction, and while both may be performed by leaders, the literature has distinguished them as different aspects of employee development (Kochanowski, Seifert, and Yukl 2010) and revealed that leaders may provide supervisory coaching but not performance feedback (Gregory, Levy, and Jeffers 2008).

Both leaders’ supervisory coaching and performance feedback are important in activating employees’ external and internal motivation and shaping employee work attitudes (Alfes et al. 2013; Lonsdale 2016). There are of course a range of factors that shape motivation and attitudes, including external factors (e.g. organizational cultures and policies) (Parker, Van Den Broeck, and Holman 2017) and internal factors (e.g. work passion and personal empowerment) (Mcallister et al. 2017). We focus on supervisory coaching and performance feedback in particular because, in the context of employee development, they represent vital leadership behaviours that may explain the link between leadership styles and important employee outcomes – an under theorized and researched pathway.

We utilize the job demands-resources (JD-R) theory (Bakker and Demerouti 2017) to form hypotheses about how these two leadership behaviours function as resources that stimulate a motivational process, evidenced in greater work engagement and reduced turnover intention. These relationships are of interest as work engagement is considered to be a focal point of talent management in retaining employees (Christensen Hughes and Rog 2008) while, more importantly, ensuring organizational sustainability and success (Shuck and Herd 2012), and reducing turnover (Timms et al. 2015). While leadership style has been linked to work engagement (e.g. Kim and Barak 2015) and turnover intention (e.g. Tse, Huang, and Lam 2013), little is known about the way that different leadership styles are experienced by employees in shaping their everyday working life outcomes (Behrendt, Matz, and Göritz 2017). In other words, in the language of the JD-R theory, what resources do leaders create through application of their different leadership styles to boost work engagement?

Our research contributes one answer to this question. Specifically, we offer insight into why some leadership styles are more effective than other styles in affecting employee attitudes, by explaining how leaders influence employee outcomes (cf. Lee, Idris, and Delfabbro 2016) through the provision of important job resources. Different types of job resources have different functions, roles, and effects (Bakker et al. 2011). In our study, we extend the motivational pathway of the JD-R theory by incorporating leadership style as an antecedent of job resources. In other words, leaders can influence work engagement by optimizing job resources (cf. Breevaart, Bakker, and Demerouti 2014; Tuckey, Bakker, and Dollard 2012). We propose that job resources are an important antecedent of employee development provided by the leader. Based on the principle that the influence of leadership styles occurs through concrete behaviours (Behrendt, Matz, and Göritz 2017), we focus on understanding two forms of leader behaviour (i.e. supervisory coaching and performance feedback) as mediators of the relationship between transformational and transactional leadership styles and employee attitudes (i.e. work engagement and turnover intention).

We examine these pathways in a multilevel survey study, undertaken in Malaysia. In Malaysia, transformational and transactional styles are utilized by organizational leaders (Kamisan and King 2013; Lo et al. 2010), which is important for examining variations in these leadership styles as initiators of the motivational process that we are studying. By employing a multilevel study, with employees nested within workgroups that have a common leader, we examine leadership styles as an organizational agent (Coyle-Shapiro and Shore 2007) associated with leader behaviours relevant to their HRD role and, in doing so, enhancing work engagement. Overall, this study sheds light on how these two leadership styles relate to leadership behaviours and how they may, or may not, effectively carry out part of the HRD role (Zhu, Chew, and Spangler 2005). Further, as a collectivist society wherein human relationships (guan-xi) are more close-knitted (Brewer and Chen 2007) and people are more group-oriented (Wasti 2003), Malaysia offers an Asian context that emphasizes relationship quality (e.g. Song et al. 2012). However, in Malaysia, there is also high power distance between the leader and the employees (Lee and Idris 2017) and a tendency towards conformism and subordination of the individual (Leung, Koch, and Lu 2002). Together, these features of the cultural context present how leaders, from a top-down perspective, may manage their employees successfully. In sum, Malaysia offers a good opportunity to explore the transmission of leadership styles through leadership behaviours to employee outcomes, from work engagement and turnover intention aspects.

Review of the literature and hypotheses development

Leadership behaviours as job resources and the link to work engagement

The JD-R theory presents an overview of job characteristics through two aspects: the health impairment pathway (i.e. job demands to burnout) and the motivational pathway (i.e. job resources to work engagement) (Bakker and Demerouti 2017). The health impairment pathway proposes that job demands will lead to higher strain in employees, thwarting organizational outcomes, whereas the motivational pathway proposes that job resources will lead to higher work engagement in employees, resulting in better organizational outcomes. Of the two pathways, the motivational pathway has gained wider attention in recent years due to its capacity to increase employees’ intrinsic and extrinsic motivation, in this case, in the form of work engagement (Fernet, Austin, and Vallerand 2012) that is able to increase job performance. Therefore, the current study focuses on job resources as they concurrently denote positive leadership behaviours.

Job resources are defined as the ‘physical, psychological, social, or organisational aspects of the job that are either/or: (1) functional in achieving goals, (2) reduce job demands and the associated physiological and psychological costs, and (3) stimulate personal growth, learning, and development’ (Bakker and Demerouti 2007, 312). While there are many types of job resources, two that are closely related to the supervisor–employee relationship are supervisory coaching and performance feedback (Dahling et al. 2016; Saks and Gruman 2014); both are vital for employees to be competent at their job (Schaufeli and Taris 2014).

Supervisory coaching is defined as

an ongoing, face-to-face process of influencing behavior by which the manager (superior, supervisor) and employee (subordinate) collaborate to achieve increased job knowledge, improved skills in carrying out job responsibilities, a stronger and more positive working relationship, and opportunities for personal and professional growth of the employee. (Yoder 1995, 271)

When supervisory coaching is provided, employees are able to be more engaged with their work as they receive more guidance from their leader in achieving work goals (Kim 2014). Hence, under the motivational pathway of the JD-R theory, coaching stimulates extrinsic motivation. Coaching may also more directly foster learning and development by guiding employees to try new opportunities and by aiding them to reflect on their experiences. Employees can thus use coaching to develop their skills and self-regulate their motivation and behaviour (Strauss and Parker 2013), meeting the basic psychological need for autonomy and boosting intrinsic motivation. Even when facing high job demands, with the provision of supervisory coaching, employees perceive the demands as less daunting and remain engaged with their work (Schaufeli and Taris 2014). In comparison to organizational coaching, supervisory coaching has a greater influence on employee attitudes given the proximal distance and the daily interaction between the leader and the employees (Theeboom, Beersma, and van Vianen 2014).

Since performance feedback is defined as ‘information provided by an agent regarding aspects of one’s performance and understanding’ (Hattie and Timperley 2007, 81), performance feedback should also stimulate both extrinsic and intrinsic motivation. It allows leaders to communicate performance standards and expectations, thus reducing job ambiguity for employees (Bakker and Demerouti 2007) and increasing the understanding and clarity of work goals (Beenen, Pichler, and Levy 2017). In response, employees can adjust their performance to meet work goals or even alter the goals themselves which should increase extrinsic motivation. With regard to intrinsic motivation, performance feedback provides recognition of a job well done which can satisfy the basic psychological need for competence.

Together, these two leadership behaviours represent forms of performance management by leaders which characterize a high-quality leader–employee relationship that can be used to increase employee work engagement (Mone et al. 2011). As these two job resources require interactions and communications between the leader and the employee, they should meet the basic psychological need for relatedness which is a core component of intrinsic motivation. According to the motivational pathway of the JD-R theory (Bakker and Demerouti 2014), when job resources such as these are provided, and intrinsic and extrinsic motivation are fostered, work engagement should result.

Hypothesis 1: (a) Supervisory coaching and (b) performance feedback will be positively related to work engagement.

Job resources reduce turnover intention through work engagement

Turnover intention is a cognitive–affective state regarding the ‘conscious and deliberate wilfulness to leave the organization’ (Tett and Meyer 1993, 262). It has been argued that turnover intention is a powerful organizational outcome and reflects voluntary turnover (Lee and Bruvold 2003). Ample evidence in the past decade has demonstrated a positive correlation between work engagement and the intention to leave employment (Halbesleben and Wheeler 2008; Jones and Harter 2005; Schaufeli and Bakker 2004), with this topic remaining a contemporary interest (Gabel-Shemueli et al. 2015; Tims, Bakker, and Derks 2013). Since work engagement is ‘a positive fulfilling, work-related state of mind that is characterized by vigor, dedication and absorption’ (Schaufeli et al. 2002, 74), employees who are engaged with their work tend to have more positive emotions and think about their work more positively (Bakker and Schaufeli 2008). Fredrickson (2001) and Kirk-Brown and Van Dijk (2016) argued that these positive emotions bring employees psychologically closer to other employees within the organization and to the organization itself, resulting in their increased identification with the organization and their greater willingness to allocate extra time and resources to their co-workers and their organization. On the flip side, according to Bakker et al. (2003a), employee withdrawal – such as absenteeism and intention to leave the organization – may therefore be a form of coping strategy used to deal with low work engagement.

Hypothesis 2: Work engagement will be negatively related to turnover intention.

The human resource management and development literature has discussed the importance of human resource management in retaining employees (Alfes et al. 2013; Buck and Watson 2002). The reviewed literature to date has indicated that a lack of job resources and the presence of stressors may predict turnover intention. Job resources, such as supervisory coaching and performance feedback, are used to ‘signal’ to employees that they are valued and the organization cares for them (Wanberg, Welsh, and Hezlett 2003). We propose that leaders play an important role in helping employees to feel psychologically safe and to experience a sense of belonging, signs that the organization places its attention on their needs. When leaders provide good performance feedback, employees feel that they will have a stable long-term career in the organization (Suazo, Martinez, and Sandoval 2009). Likewise, the presence of job resources, such as supervisory coaching, reduces turnover intention (Spell, Eby, and Vandenberg 2014).

Hypothesis 3: (a) Supervisory coaching and (b) performance feedback will be negatively related to turnover intention via their positive relationship with work engagement.

Influence of transformational and transactional leadership styles on supervisory coaching, performance feedback, and work engagement

Thus far, the literature has shown that transformational leadership is linked to job resources (Fernet et al. 2015; Gillet and Vandenberghe 2014). For example, Gillet and Vandenberghe (2014) found that transformational leadership increased employees’ performance feedback and decision-making autonomy. Consistent with this idea, we also propose that transformational leadership, in contrast to transactional leadership, would provide employees with valuable job resources in the form of supervisory coaching and performance feedback and, through these resources, would stimulate work engagement.

Transformational and transactional leadership styles exude different qualities (Judge and Piccolo 2004). Transformational leadership is characterized by four components: idealized influence, inspirational motivation, intellectual stimulation, and personal consideration. All of these components refer to leadership characteristics that focus attention on the needs of individuals in achieving organizational goals and objectives. Transactional leadership, on the other hand, is characterized by three components: contingent reward, active management by exception, and passive management by exception. All of these components refer to leadership characteristics that focus on punishment and rewards in achieving organizational goals and objectives.

These characteristics can be used to link leadership styles to specific leader behaviours (i.e. the provision of specific job resources). Arguably, transformational leaders will exhibit behaviours that attend to employees’ needs, such as supervisory coaching and performance feedback. This idea resonates with Lapointe and Vandenberghe’s (2017) suggestion that transformational leadership exhibits behaviours that provide job resources to employees, in contrast to transactional leadership.

Personal consideration, one of the four components of transformational leadership, is defined as ‘the degree to which the leader attends to each follower’s needs, acts as a mentor or coach to the follower, and listens to the follower’s concerns and needs’ (Judge and Piccolo 2004, 755). Personal consideration, in particular, contains elements of development and support provided by a leader (Rafferty and Griffin 2006). Given the holistic focus on employees that comes with personal consideration, leaders who adopt a transformational leadership style would have high awareness of individual employee strengths and capabilities. This in-depth understanding can underpin the provision of supervisory coaching and high-quality performance feedback to maximize employees’ strengths and improve weaknesses. Engaging in supervisory coaching is also a way in which concern for employees’ need for competency (Kovjanic et al. 2012), a hallmark of personal consideration, can be understood and directly expressed. Hence, transformational leadership should boost intrinsic motivation through the provision of employee-oriented resources such as supervisory coaching and performance feedback.

In addition, even though both transformational and transactional leaders influence employees to achieve expected goals, they use different methods to motivate employees. As transformational leaders use intellectual stimulation and inspirational motivation, this leadership style should be more likely to foster work engagement, as compared to transactional leadership. Intellectual stimulation is defined as ‘the degree to which the leader challenges assumptions, takes risks, and solicits followers’ ideas’, while inspirational motivation is defined as ‘the degree to which the leader articulates a vision that is appealing and inspiring to followers’ (Judge and Piccolo 2004, 755). Transformational leaders, in particular, place emphasis on ‘a new vision’ and have a ‘shared’ orientation with their employees (Wang, Waldman, and Zhang 2014) rather than influencing them in traditional ways. In order to align employees’ competencies to organizational objectives, transformational leaders focus on employees’ competency development and motivate employees to go beyond what they are expected to accomplish (Breevaart, Bakker, and Demerouti 2014). When employees contribute to organizational objectives, they become more motivated when working on tasks entrusted to them (Kim and Barak 2015). Past studies have demonstrated that the transformational leadership style, due to the behaviours exhibited by the leader, leads to a higher level of work engagement among employees. Using a diary study, Tims, Bakker, and Xanthopoulou (2011) found that transformational leaders increased work engagement on a daily basis as they provided motivational power and inspiration to their employees.

Transactional leadership, on the other hand, is characterized by the component ‘contingent reward’ with which a leader ‘clarifies expectations and establishes the rewards’ (Judge and Piccolo 2004, 755) through leader–follower transactions. Both active and passive management by exception – the other two aspects of transactional leadership – focus on the presence and correction of problems, with this mostly being to reach organizational objectives, rather than on employees’ development needs. Hence, while extrinsic motivation may be stimulated to some degree, intrinsic motivation is likely to suffer because the resources are not provided to meet basic psychological needs nor directly to simulate learning, growth, and development. Transactional leadership also imposes more control on employees and lacks motivational elements (Van Vugt et al. 2004). The focus of transactional leaders is primarily on employee performance with employees only rewarded when agreed-upon goals are met (Bass et al. 2003). While transformational leaders provide praise, support, and encouragement, transactional leaders utilize incentives and punishment (Breevaart, Bakker, and Demerouti 2014). Such leadership behaviours reduce employees’ motivational level (Thomas 2009), thus impeding their enjoyment of their work and their effective performance.

Hypothesis 4: Transformational leadership (but not transactional leadership) will be positively associated with (a) supervisory coaching and (b) performance feedback and, in turn, with work engagement.

Hypothesis 5: Transformational leadership (but not transactional leadership) will be positively associated with work engagement.

Especially in Asian countries, such as Malaysia, wherein work relationships – usually close-knitted – are regarded as an important part of the workplace (Brewer and Chen 2007), transformational leadership has been suggested as a facilitator in employee development through the provision of supervisory coaching and feedback that helps employees to improve their motivation through engagement at work (Schaufeli and Bakker 2004). Given the developmental focus of transformational – but not transactional – leadership, we predict that transformational leadership would be associated with greater provision of supervisory coaching and performance feedback. Taking all these arguments together, we propose the following hypothesis:

Hypothesis 6: (a) Supervisory coaching and (b) performance feedback mediate the relationship between transformational leadership and work engagement (but not between transactional leadership and work engagement).

Overall, the proposed research framework is illustrated in Figure 1.

Figure 1. Hypotheses and research model.

Method

Participants

The current study employed a cross-sectional multilevel design with 500 employees (average age = 31.11 years; standard deviation [SD] = 8.47) in 65 workgroups from various private organizations in the capital city of Malaysia. All organizations that participated were from the services industry and all participants were administrative assistants whose main tasks involved ensuring that the administrative aspects of the organization ran smoothly. The mean length of working experience was 4.65 years (SD = 5.05). There were more females (N = 283, 56.6%) than males and most were Chinese (70%), followed by Malay (18.4%) and Indian (1.6%). Most participants were single (N = 297, 59.4%), followed by those who were married (N = 190, 38%), while the remainder were divorced, separated, or widowed (N = 13, 2.6%).

One department was selected per organization based on the organization’s recommendation, and the departments were treated as workgroups. The sample size is adequate according to scholars’ recommendations. For example, in simulation studies, Mathieu et al. (2012) reported that as few as 25 workgroups at the upper level (UL) are considered reasonable, while other studies, such as that of Dollard, Tuckey, and Dormann (2012), used 23 workgroups and 139 individuals. Organizations were approached via email using the snowballing sampling method, with the help of research assistants. Interested organizations nominated a workgroup who may be interested in participating in the study. Participant information sheets and questionnaires were then distributed to employees within the workgroup. One week following distribution, we returned to each organization to collect the completed (and incomplete) anonymous questionnaires which had been placed in sealed envelopes. The number of participants per workgroup ranged from 5 to 26.

Instruments

The instruments were chosen based on their established validity, reliability, and, most importantly, their suitability within the Asian context (e.g. Diaz-Saenz 2011; Fong and Ng 2011; Iyer 2016). The English language of the original scales was retained as most Malaysians have high-level proficiency in English (Thirusanku and Yunus 2014). The reliability of the scales is shown in Table 1.

Table 1. Means, standard deviations, reliability, and Pearson’s bivariate correlations.

Variables	Mean	SD	α	No. Items	1	2	3	4	5	F	ICC(I)
1. Transformational leadership	4.82	0.90	0.91	23	1					2.348***	0.1472
2. Transactional leadership	4.81	1.05	0.86	5	.81***	1				2.138***	0.1295
3. Supervisory coaching	3.34	0.89	0.89	5	.50***	.49***	1			2.068***	0.1291
4. Performance feedback	3.36	0.88	0.85	3	.39***	.35***	.56***	1		1.611**	0.0782
5. Work engagement	3.82	1.09	0.91	9	.44***	.43***	.21***	.23***	1	1.643**	0.2266
6. Turnover intention	2.80	1.41	0.90	3	−.25***	−.25***	−.13**	−.13**	−.26***	1.906***	0.1163

SD = standard deviation; ICC = intraclass correlation coefficient; N (individual) = 500; **p < .05; ***p < .001.

Transformational leadership and transactional leadership were measured using 28 items from the Transformational Leadership Inventory (Podsakoff et al. 1990), with the term ‘my leader’ rephrased as ‘my group leader’. Of the 28 items, 23 measured transformational leadership (e.g. ‘My group leader has stimulated the group to rethink the way we do things’) while 5 items measured transactional leadership (e.g. ‘My group leader gives the group special recognition when our work is very good’). The scale ranged from ‘1’ (strongly disagree) to ‘7’ (strongly agree).

Supervisory coaching was measured using five items from Le Blanc’s (1994) scale. The scale ranged from ‘1’ (never) to ‘5’ (very often). An example of one item is as follows: ‘My supervisor uses his/her influence to help me solve problems at work’. Its validity has been supported (e.g. Schaufeli and Bakker 2004; Xanthopoulou et al. 2007, 2009).

Performance feedback was measured using three items from Bakker et al. (2003b). The scale ranged from ‘1’ (never) to ‘5’ (very often). An example of one item is as follows: ‘I receive sufficient information about the results of my work’.

Work engagement was measured using nine items from the Utrecht Work Engagement Scale (UWES-9) (Schaufeli, Bakker, and Salanova 2006) which consists of vigour, dedication, and absorption. The scale ranged from ‘1’ (never) to ‘5’ (always). An example of one item is as follows: ‘I am enthusiastic about my work’. The UWES-9 measure was selected (a) to match the theoretical framing of the study in the JD-R model (b) because it has been utilized in studies conducted in both individualistic (e.g. Bakker et al. 2011) and collectivistic (e.g. Lee, Idris, and Delfabbro 2016) cultures and (c) in recognition that it has been applied in many different types of occupations (e.g. Tuckey, Bakker, and Dollard 2012). We note that questions regarding the theoretical and empirical distinction between the constructs of burnout and engagement have been raised and substantiated with meta-analytic data (Cole et al. 2012). Against this backdrop, we adopt a work engagement conceptual framing and measurement approach in this study given that engagement is closely related to job resources in JD-R theory (as compared to burnout, which is closely related to job demands; Bakker and Schaufeli 2008).

Turnover intention was measured using three items from O’Driscoll and Beehr (1994). The scale ranged from ‘1’ (never) to ‘5’ (all the time). An example of one item is as follows: ‘Thoughts about quitting this job cross my mind’.

Analysis strategy

Prior to undertaking multilevel analyses using hierarchical linear modelling (HLM) software, we ran analyses to see if demographic variables would influence our studied variables. We found that only marital status had a negative influence on work engagement and a positive influence on turnover intention. This can be explained as employees who were married had to place more attention on both family and work (Allen et al. 2014). Thus, we controlled for marital status in all our HLM analyses.

The UL variables (i.e. transformational leadership and transactional leadership) were analysed to ascertain if they possessed group-level properties, and whether they could be aggregated as group-level variables. Group-level properties are present if a shared perception exists between group members and the UL variables have influence on individual-level variables (Kozlowski 2012). Overall, the r(WG)(J) (index of agreement) values for transformational leadership and transactional leadership were .93 and .95, respectively, indicating a high level of within-group agreement (LeBreton and Senter 2008). The intraclass correlation coefficient (ICC[I]) value for transformational leadership was .15, while it was .13 for transactional leadership, indicating that 15% and 13% of the variance, respectively, in both leadership constructs was due to group factors. Bliese (2000) suggested ICC(I) values should be between .05 and .20. The F_(III) values were found to be significant (transformational leadership = 2.35, p < .001; transactional leadership = 2.14, p < .001), indicating justification of the aggregation of these variables.

To test our hypotheses, HLM software (Bryk and Raudenbush 1992) was used: leadership was treated as a group-level construct that has top-down influence on employees, with this being the lower construct (Snijders and Bosker 2012). Three types of analyses were conducted comprising lower level (LL) direct effects, cross-level direct effects, and mediation effects. LL direct effects and cross-level direct effects were tested using Mathieu and Taylor’s (2007) recommendation. First, we ran the analysis for LL direct effects (i.e. regressing the LL outcomes’ variable on LL independent variables), followed by conducting a cross-level direct effects analysis (i.e. regressing LL variables on transformational leadership and transactional leadership).

An example of a cross-level HLM equation is as follows:

Level 1 Model: Work engagement = β₀ + β (supervisory coaching) + β (performance feedback) + β (marital status) + r

Level 2 Model: β₀j = G₀₀ + G₀₁ (transformational leadership) + G₀₁ (transactional leadership) + u_0j

For LL direct effects (Hypotheses 1, 2, and 3), the LL dependent variable was regressed on the independent variables. For example, in Hypothesis 5, turnover intention was regressed on the variables: work engagement and marital status (see Model 1).

An example of a LL HLM equation is as follows:

Turnover intention = β₀ + β (work engagement) + β (marital status) + r

Finally, to test the mediation hypotheses, Shapiro–Wilk tests were conducted with all variables found to be normally distributed (p > .05). We then followed the testing steps as recommended by Baron and Kenny (1986). First, we found a significant relationship between X → Y (supervisory coaching/performance feedback → turnover intention) (Model 5). Second, a significant relationship was found between X → M (supervisory coaching/performance feedback → work engagement) (Model 6). Third, we found a significant relationship between M → Y, in the presence of X (supervisory coaching/performance feedback +work engagement → turnover intention) (Model 4). As indicated in the second step, if the relationship from X to Y remains significant with the inclusion of M, then it is partial mediation. If the addition of M produces an insignificant relationship from X to Y, it is considered to be full mediation. To confirm the mediation pathway relationship, we used the Monte Carlo test (Selig and Preacher 2008) which has been suggested as being more applicable for multilevel analyses. We tested the mediation pathway by using estimates of Path a (X → M) and Path b (M → Y). The mediation effect is confirmed if the values of LL and UL variables do not contain zero (0) (MacKinnon, Lockwood, and Williams 2004). The Monte Carlo test was conducted using a 95% confidence interval (CI) and with 20,000 repetitions.

Results

Table 1 presents the descriptive analysis and correlations between all measures at the individual level. The results from the HLM analysis are shown in Tables 2 and 3. A summary of the findings is presented in Figure 2.

Table 2. Hierarchical Linear Modeling (HLM) analysis of lower level outcomes.

Effect	Turnover intention	Turnover intention	Turnover intention	Turnover intention	Turnover intention	Work engagement
Model	1	2	3	4	5	6
Lower Level Effects
Supervisory coaching		−.05(05)		−.03(.06)	−.12(.05)*	.13(.06)*
Performance feedback			−.05(.05)	−.04(.06)	−-.16(.05)**	.21(.06)***
Work engagement	−-.24(.05)***	−-.22(.05)***	−-.22(.05)***	−-.22(.05)**
Marital status	−.12(.05)**	−-.12(.05)**	−-.12(.05)**	−-.11(.05)**	−-.05(.05)	.05(.05)

The first value is the unstandardized parameter estimate; the value in parentheses is the standard error (SE); N (individual) = 500; N (workgroup) = 65; *p < .05. **p < .01; ***p < .001.

Table 3. Hierarchical Linear Modeling (HLM) analyses of lower level outcomes and cross-level effect of transformational leadership and transactional leadership on lower level outcomes.

Effect	Work engagement	Work engagement	Supervisory coaching	Performance feedback
Model	7	8	9	10
Lower Level Effects
Supervisory coaching	.13(.06)*
Performance feedback	.21(.06)***
Marital status	.05(.05)	.08(.05)	.06(.05)	.11(.05)
Cross-Level Effects
Transformational leadership	.22(.10)*	.22(10)*	.23(.07)**	.36(.11)**
Transactional leadership	.04(.10)	.04(.10)	.13(.10)	−-.11(.10)

The first value is the unstandardized parameter estimate; the value in parentheses is the standard error (SE); N (individual) = 500; N (workgroup) = 65; *p < .05; **p < .01; ***p < .001. We controlled for marital status in the analysis.

Figure 2. Final model.

Hypothesis 1 predicted that supervisory coaching and performance feedback would show positive relationships with work engagement. As indicated in Table 2, Model 6, our analysis suggests that supervisory coaching (β = .13, p < .05) and performance feedback (β = .21, p < .05) have positive significant relationships with work engagement. Hence, Hypothesis 1 was supported. Hypothesis 2 predicted that work engagement would negatively relate to turnover intention, which was also supported by the data (β = −.24, p < .001) (see Table 2, Model 1).

Hypothesis 3 predicted that work engagement would mediate the negative relationship between both job resources and turnover intention. To evaluate the mediation testing, we used the parameter estimate value for Table 2, Model 5 as the value for the direct effects of supervisory coaching/performance feedback → work engagement (β = .13, standard error [SE] = .06/β = .21, SE = .06), and the parameter estimate from Table 3, Model 4 for work engagement → turnover intention with supervisory coaching/performance feedback in the model (β = −.22, SE = .05). Consistent with Hypothesis 3, our analysis confirmed the mediation effect from supervisory coaching to turnover intention via work engagement (95% CI, LL = −.0610, UL = −.0027), and the mediation effect from performance feedback to turnover intention via work engagement (95% CI, LL = −.0840, UL = −.0171).

Hypothesis 4, predicting that transformational leadership but not transactional leadership, would be positively related to supervisory coaching and performance feedback was also supported. As indicated in Table 3, our results showed that transformational leadership had a significant cross-level effect on supervisory coaching (γ = .23, p < .05; Model 9) and performance feedback (γ = .36, p < .05; Model 10), while transactional leadership was not associated with supervisory coaching (γ = .13, not significant [ns]; Model 9) nor performance feedback (γ = −.11, ns; Model 10). Likewise, consistent with Hypothesis 5 (see Table 3, Model 8), our result suggests that transformational leadership had a significant cross-level effect on work engagement (γ = .22, p < .05), while transactional leadership was not associated with work engagement (γ = .04, ns).

Hypothesis 6 predicted that supervisory coaching and performance feedback would mediate the relationship between transformational leadership and work engagement. In testing the hypothesis, the conditions stated by Baron and Kenny (1986) were fulfilled. First, we found a direct effect only from X → Y (transformational leadership → work engagement). We then analysed the mediation effect using the path from transformational leadership → supervisory coaching/performance feedback → work engagement by using the Monte Carlo test. Specifically, we used the parameter estimate from Table 3, Model 9 as the value for the direct effect from transformational leadership to supervisory coaching (γ = .23, SE = .07), and the parameter estimate for Table 3, Model 7 (supervisory coaching → work engagement; β = .13, SE = .06) with transformational leadership and transactional leadership in the model. The Monte Carlo bootstrapping indicated that transformational leadership has a significant effect on work engagement through supervisory coaching (95% CI, LL = .0023, UL = .0687).

We repeated the same procedure to see the effect of transformational leadership on work engagement through performance feedback. Thus, we used the parameter estimate from Table 3, Model 10 as the value for the direct effect from transformational leadership to performance feedback (γ = .36, SE = .11), and the parameter estimate from Table 3, Model 7 (performance feedback → work engagement) with transformational leadership and transactional leadership in the model (β = .21, SE = .06). Again, the Monte Carlo bootstrapping supported the mediation process (95% CI, LL = .0222, UL = .1480). Thus, Hypothesis 6 is supported.

Discussion

Previous research linking leadership to employee outcomes has focused on how leaders influence employees’ cognitive and affective states (e.g. Chen et al. 2011; Henker, Sonnentag, and Unger 2015). Our contribution lies in highlighting what leaders actually do to strengthen employee work engagement and reduce turnover intention and, in so doing, to identify which leadership style – transaction or transformational – is more effective and why. Specifically, our study investigated (1) how supervisory coaching and performance feedback are important job resources that leaders can provide within the employee learning and development context, in relation to work engagement and turnover intention; and (2) the role of leadership styles enacted through leadership behaviour. Overall, we found that transformational leadership (in contrast to transactional leadership) translates into development-focused leadership behaviours – supervisory coaching and performance feedback – that act as resources in employee development and, in return, benefit the organization by retaining human capital and promoting work engagement.

Theoretical contributions

While it is commonly known that supervisory coaching and performance feedback are viewed as job resources within the JD-R theory (Bakker and Demerouti 2017), through our research, the importance of providing job resources is added to the role of leadership in the context of employee development. To be specific, when compared with transactional leadership, transformational leadership is more likely to provide resources, with (at least) two specific behaviours directed to employees: supervisory coaching and performance feedback. These two behaviours function as job resources: through the provision of support and feedback, leaders convey information that can directly help employees to meet goals, thereby increasing extrinsic motivation. In addition, supervisory coaching particularly, but also performance feedback, directly encourages ongoing personal development, a form of intrinsic motivation. Specifically, the study shows that transformational (and not transactional) leaders are likely to ‘grow’ their employees through supervisory coaching and performance feedback, and that these resources, in turn, foster greater work engagement and less turnover intention.

Our results support the notion that transformational leaders are more employee-oriented and thus more concerned about their employees’ careers and development (Tansky and Cohen 2001). This is achieved by maximizing their role and addressing employee development needs through positive leadership behaviours which serve as job resources to employees, with this concern demonstrated via communication in the form of coaching and feedback. As transformational leadership emphasizes employee progress and development, this style of leadership tends to involve greater communication about and sensitivity to employees’ needs (Stone, Russell, and Patterson 2004) and motivates employees to thrive at work (Paterson, Luthans, and Jeung 2014). When employees perceive that leaders care about their work performance and personal development, they in turn will feel more motivated to work and become more committed to their workplace. This finding is consistent with previous studies which have shown that transformational (and not transactional) leadership influences positive employee outcomes, such as organizational citizenship behaviour (Guay and Choi 2015) and work engagement (Breevaart, Bakker, and Demerouti 2014; Tims, Bakker, and Xanthopoulou 2011), from both Western and Eastern perspectives (Song et al. 2012; Tims, Bakker, and Xanthopoulou 2011).

While transactional leadership provides rewards when agreed-upon goals are met, this leadership style emphasizes task-oriented goals, resulting in a lack of attention given to employees’ needs (Tracey and Hinkin 1998). In addition, the punishment aspect when agreed-upon goals are not met also implies a style that is bureaucratic, conforming, non-inspiring, and not motivating (Emery and Barker 2007). Although transactional leadership is not deemed to be a negative leadership style, especially with the provision of appropriate leadership behaviours through active management by exception, transformational leadership is related to a more positive work attitude among employees. In addition, the insignificant finding may only be applied on service work employees but may work differently in other work types (Pieterse et al. 2010).

In addition, the study has shown that supervisory coaching and performance feedback are two important leadership behaviours. Supervisory coaching is a form of leadership behaviour that allows leaders to develop their employees through active encouragement and support (Redshaw 2000). Performance feedback, on the other hand, allows employees to know what needs to be improved and reduces job ambiguity through the high quality of communication (Bakker and Demerouti 2007). Consistent with our prediction, we found that supervisory coaching and performance feedback play an important role in employee work engagement and turnover intention. As expected, we found that, in comparison to transactional leadership, the transformational leadership style is associated with a higher level of the two leadership behaviours – supervisory coaching and performance feedback – which, in turn, mediate the relationship with work engagement. Overall, our findings are consistent with the key principle of the integrative model of leadership in which leadership styles are exhibited through specific leadership behaviours (Behrendt, Matz, and Göritz 2017). Thus, the findings help to answer the question of how leadership styles contribute to employee work attitudes.

As working conditions are created through management initiatives (Johns 2010), leaders are viewed as providers of job resources to employees (Sirmon, Hitt, and Ireland 2007). The provision of job resources to employees can be explained through the lens of the norm of reciprocity where employees feel valued and that the leader and the organization have invested in them. In return, they are more engaged at work and less likely to have turnover intentions due to the psychological contract they have with the leader (Bhatnagar 2014; Tuzun and Kalemci 2012).

Strengths, limitations, and future directions

This study is novel in that it investigates the behavioural aspects of two leadership styles, namely, transformational and transactional leadership. Most previous studies have not investigated the influence of specific leadership behaviours on employee attitudes. Moreover, we demonstrate the importance of having specific job resources, in this case, within the HRD context. This is important as each job resource is different in nature, purpose, and functionality (Bakker et al. 2011). The finding illustrates that leadership effectiveness within the HRD context consists of both task-oriented and relationship-oriented behaviours which are consistent with the concepts of these leadership styles.

In addition, the current study uses a multilevel perspective, taking into consideration the view that social contexts (including leadership styles) affect employee attitudes. In this way, our study provides a relational perspective between the leader and the employees, at both group and individual levels (cf. Xing et al. 2016). Leadership styles are viewed as an UL construct as leaders command a top-down influence on employees (Dionne et al. 2004; Wang et al. 2013). Obtaining responses from individuals, rather than from groups, may result in bias and an incorrect reflection of the leadership style (e.g. Gordon et al. 2014; Park, Kim, and Song 2015). The multilevel perspective allows multiple participants within the same group to rate their leader’s leadership style, thus making the ratings more objective and presenting a more accurate reflection of the leadership style (Kristof-Brown et al. 2014).

Our study was conducted in Malaysia which, being a collectivist culture, is characterized by emphasis on the quality of relationships but also with high power distance and subordination of the individual. Findings from the study further support the role of transformational leadership on employee development not only from the Western perspective but also from the Eastern perspective. To summarize, transformational leadership, rather than transactional leadership, is able to develop employees’ mental complexity through necessary guidance and feedback across cultural contexts (Crane and Hartwell 2018; Dong et al. 2017).

Despite these strengths, the current findings offer only partial support for the core processes of the JD-R theory as we did not examine job demands, which are related to job resources but which play a different role according to JD-R theory (Bakker and Demerouti 2014). First, although job demands are mostly negative and the opposite of job resources, certain types of job demands may be conducive to work engagement. Specifically, whereas relationships between job resources and work engagement tend to be positive, meta-analysis showed that relationships hindrance demands tend to undermine engagement whereas challenge demands tend to promote engagement (Crawford, LePine, and Rich 2010). In addition, job resources may buffer the negative impacts of job demands (Bakker, Demerouti, and Euwema 2005), especially those job resources that are viewed as important in coping with job demands (Xanthopoulou et al. 2007), with this being highly related to turnover intention (van Woerkom, Bakker, and Nishii 2016). Third, leaders are able to reduce employees’ job demands (Van Yperen and Hagedoorn 2003) as well as provide additional resources. Hence, it is important to study the role of job demands and job resources together (cf. Tuckey, Bakker, and Dollard 2012). Comparing leadership styles with an overall model of job resources and job demands would provide a better perspective on how these job characteristics work and interact to affect employees’ work outcomes. Noting concerns about the distinction between burnout and engagement (Cole et al. 2012), in future research of this kind it may be useful to adopt Kahn’s (1990) conceptualisation, in which engagement reflects harnessing of the full self – physically, cognitively, and emotionally – into the work role.

Lastly, the causal direction remains unclear due to the study’s cross-sectional design. Future research should examine longitudinal effects to see whether any changes occur over time due to the influences of transformational or transactional leadership styles on employees, via supervisory support and performance feedback behaviours. This step is important, especially in conditions where several mediators are used in the research model (Maxwell and Cole 2007). With the current study and its two mediating processes, collecting at least three times the amount of data would be ideal in order to see the effect from independent variables (Time 1 [T1]) → mediators (Time 2 [T2]) → dependent variables (Time 3 [T3]).

Practical implications

As more leaders are involved in employee development (Alfes et al. 2013), organizations should pay more attention to managing employees through leaders, ensuring that leaders can fulfil both supervisory and leadership roles. Effective leadership, such as transformational leadership, may be able to attend to these needs through supervisory coaching and performance feedback. This is particularly useful as leaders often contribute to the organizational design process; thus, they could ensure that employees’ motivational aspects are addressed (Frost, Osterloh, and Weibel 2010). Therefore, the resultant design would allow employees’ behavioural outcomes to be aligned with organizational objectives. This approach allows for a better understanding of the behaviours exhibited when explaining leadership styles that can be more concretely applied in practice. This is particularly important as employees in Asia’s presenteeism culture suffer from higher risks of ill health, with job resources an important element in buffering those risks (Lu, Cooper, and Lin 2013). More importantly, internal resources, such as knowledge and human capital, could then be preserved and maintained within the organization (Foss and Knudsen 1996).

As considerable resource investment is required to retain employees (Sonnentag, Binnewies, and Mojza 2010) and as job resources, such as supervisory coaching and performance feedback, are useful in increasing work engagement and reducing turnover intention, leaders may provide sufficient guidance and feedback to ensure that employees are skilful and knowledgeable. The significant influences of leadership styles on employee outcomes have been revealed in previous studies (e.g. Mullen, Kelloway, and Teed 2017). With the application of this knowledge based on the general characteristics of these leadership styles, practitioners faced difficulty in implementing the findings (Anderson and Sun 2017). The findings of our study are more practicable, explaining the behavioural aspects (i.e. supervisory coaching and performance feedback) of leadership with regard to their influence on increasing important employee attitudes (i.e. work engagement and turnover intention).

In addition to focusing on the specific behaviours studied here, and given the evidence that leadership styles can be cultivated (Bass and Avolio 1990), organizations could provide training for leaders to assist them to adopt a transformational leadership style when dealing with employees, together with improving their capacity to provide high-quality supervisory coaching and feedback. This may be particularly important in cultures and organizations wherein transactional approaches are routinely used: we saw in this study that, even alongside high levels of transactional leadership (as evidenced by the high correlation), transformational leadership has benefits.

As Asian countries focus on guan-xi through which the workplace not only focuses on work aspects but also on human relationships (Xing et al. 2016; Zhai, Lindorff, and Cooper 2013), organizations may want to invest in enhancing the relationship between leader and employees in developing human resources and retaining knowledge as a competitive organizational advantage by making sure that leaders possess transformational leadership style (rather than transactional leadership style) that is conducive in influencing employees’ behaviours positively (Nyberg and Ployhart 2013; Reilly et al. 2014). While Malaysia is viewed as a country with high power distance, it is also considered to be highly collectivistic. It is suggested that, despite the formal workplace hierarchy, supervisors can play a supportive role and provide sufficient feedback as part of assisting employee learning and development, with employees engaged with their work and finally, having less intention to leave the organization.

Conclusion

The current study supports the importance of supervisory coaching and performance feedback in stimulating work engagement and reducing turnover intention. In the context of the leader’s role in HRD, the study supports the applicability of transformational leadership (in contrast to transactional leadership) as an antecedent for supervisory coaching and performance feedback. These findings highlight the behavioural components (i.e. supervisory coaching and performance feedback) which function as job resources created through transformational leadership that enable this leadership style to more effectively carry out some HRD functions. Future studies could integrate job demands and job resources within the JD-R theory to gain an understanding of the role of effective leadership behaviours in creating challenging and supportive working conditions that can benefit employees and organizations alike and expand the conceptualisation and measurement of engagement to include the expression of the full self at work. The use of a longitudinal approach in investigating the effects of effective leadership on employee attitudes would also help to unpack the relationships involved.

Declarations

The data has not been published elsewhere. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.