Our Boss is a Good Boss! Cross-level Effects of Transformational Leadership on Work Engagement in Service Jobs

Introduction

In recent decades, the nature of work has changed. A transition has taken place from mechanical jobs to work in service segments. Employment changes involve changing emotional demands that require some kind of emotional labor from employees. Many occupations require significant emotional competences, such as coping with difficult social interactions with customers, students, or colleagues, emotional regulation at work, and so on. Occupations that require emotional work include face-to-face or voice-to-voice contact with others. In these circumstances, employees need to deliver positive emotional states to others, and employers can manage (to some degree) the emotional activities of their employees (Hochschild, 1983). This is a vital part of the work in service jobs.

In the past two decades, interest in affective and emotional psychological experiences at work (Brief & Weiss, 2002) has grown, and research has examined the role of emotions and their effects on job performance and wellbeing (Martins et al., 2010; O’Boyle et al., 2011), prosocial behavior (Nozaki, 2015), and motivation (Sherdell & Waugh, 2012).

In service work, feelings and emotional demands are critical due to their impact on interactions and social connections with customers, patients, pupils, and so on. As service jobs increase in number, emotional labor becomes a more common part of work. The effect of emotional demands on employees has been a pertinent subject in research on emotional labor. These consequences incorporate both financial and non-economic costs and rewards. In service job contexts, like the one in this study, employees generally anticipate showing positive feelings and suppressing negative feelings when interacting with customers (Diefendorff & Richard, 2003). However, they cannot find positive feelings in every situation, and sometimes feel uncomfortable but “have” to express positive emotions to others. This emotion-rule dissonance can be harmful to employee wellbeing (Morris & Feldman, 1996).

Because of the widespread movement toward a team-based organization in companies, managers frequently have to lead and energize individuals and teams (Hackman, 2002; Kozlowski & Bell, 2003). The role of managers and leaders has moved from managing individuals to managing teams, defined as two or more individuals who share common task goals, perform interrelated undertakings, and are mutually responsible for and in charge of collective goals. Most of the research on teams recognizes that knowledge about leading people can extend to the group level, and that there is a need for a genuine multilevel theory of team leadership. Moreover, research must consider top-down effects of contextual factors (such as leadership) on individuals’ functioning in teams, as well as bottom-up effects of individual factors on leadership and the organizational context.

This study is carried out with employees who provide services to customers, and focuses on emotional demands and the role of leadership in employees’ work engagement. There is a large body of literature on work environment factors related to employee wellbeing and stress (Danna & Griffin, 1999), and this study assumes that supervisors affect employees’ emotional experiences. Miner et al. (2005) discovered that employees rated 80% of their interactions with supervisors as positive and 20% as negative. In any case, negative interactions had effects on employee affect that were, all in all, five times stronger than those of positive interactions.

Work engagement is a motivational state involving work-related wellbeing, and it is composed of three dimensions, i.e., vigor, dedication, and absorption. Engaged workers spend high levels of energy at work, feel enthusiastic and strongly involved in their jobs, and are completely immersed in the work they do (Schaufeli et al., 2002). Some studies have shown the relationship between leadership and in-role performance and work engagement (Mehmood et al., 2016). We propose that leadership is a moderator variable of the effects of emotional demands on work engagement. Leadership will have an interaction effect on the relationship between emotional demands and work engagement. Thus, when emotional demands are high but there is positive leadership, the impact on work engagement will be less negative. In these types of situations, supervision and leadership roles are a vital resource for employers.

Theoretical Background

Emotional Demands and Work Engagement

The Job Demands-Resources (JD-R) model displays a balanced approach to explaining both negative and positive aspects of work. This model (JD-R) determines that employee wellbeing (work engagement) may result from particular working conditions, i.e., the specific combination of job demands and job resources (Bakker & Demerouti, 2013). The basic assumptions of the JD-R model have also been supported by cross-sectional, longitudinal, and diary studies using multilevel designs (Bakker & Demerouti, 2017).

Emotional demands can be defined as job aspects that require sustained emotional effort due to social connections with customers (De Jonge et al., 2008). Emotional demands have to do with emotionally charged interactions at work (e.g., customer/colleague misbehavior) (Heuven et al., 2006) that are sometimes viewed as an important source of job stress (Van Woerkom et al., 2016). Some of the effects, such as burnout and job (di)satisfaction, have been studied extensively. Employees who manage high emotional demands will probably display more mental and emotional distress (Castillo-Gualda et al., 2018; Wang et al., 2013), which places them at higher risk of mental health disorders and reduced performance (Johannessen et al., 2013). Only a couple of studies have researched their potential positive effect on wellbeing (Bakker et al., 2007; Xanthopoulou et al., 2013). These studies discovered that emotional demands can be beneficial, especially when job resources are accessible at work. Ashforth and Humphrey (1993) suggested some time ago that emotional labor may be related to greater job satisfaction because interaction regulation makes interactions less surprising. Concurring with this idea, Coté and Morgan (2002) discovered that employees who had more positive feelings when sincerely engaging in emotional interactions with others felt more satisfied at work. However, results from studies on work engagement also reveal some inconsistences. For example, Bakker et al. (2007) found that dealing with pupil misbehavior is negatively associated with work engagement, and that when emotional demands are high, the positive relationship between resources and work engagement becomes significantly weaker (Bakker & Sanz-Vergel, 2013). However, Van den Tooren and Rutte (2016) found no relationship between emotional labor and work engagement in airline crew members. These inconsistent findings may be explained by other variables that moderate the relationship between emotionally demanding conditions and work engagement, such as social job resources (e.g., leadership).

As mentioned above, work engagement is a positive and fulfilling work-related state of mind characterized by factors such as vigor, dedication, and absorption (Schaufeli & Bakker, 2010). Vigor refers to high levels of energy and mental resilience at work, willingness to try hard in one’s work, and persistence when facing challenges and difficulties. Dedication refers to feelings of significance, energy, motivation, and pride in one’s work. The third dimension of engagement is called absorption, which is characterized by being completely focused on and joyfully engrossed in one’s work, where time passes quickly and it is hard to detach oneself from work. In this way, engaged employees have a lot of vitality, feel vigorous and strong, are energetic while working, and are exceptionally focused on their work.

High levels of work engagement have positive effects on people, such as better mental health (Upadyaya et al., 2016). Other evidence additionally links engagement to better work performance (Ferreira et al., 2019) and other gainful outcomes, such as proactive behavior and organizational commitment (Vecina et al., 2012).

Drivers of work engagement can be viewed from the perspective of occupational stress models. The hypothetical underlying foundations of our study lie in the Job Demand-Resources (JD-R) model (Bakker & Demerouti, 2017), which expects people’s work environments to differ, dividing their work characteristics into two categories: job demands and job resources. The fundamental proposal of the JD-R model highlights the relations among job demands and job resources, wellbeing, and performance. Based on the premises of the JD-R Model, we can hypothesize that job demands (i.e., emotional demands) and job (social) resources (i.e., effect of the leader’s role in the workplace) are likely to impact employee wellbeing (i.e., work engagement).

A few studies have shown a positive relationship between resources and wellbeing in the form of work engagement (see Burke & Page, 2017). Llorens et al. (2006) revealed a positive relationship between job autonomy and work engagement. Other studies found that work engagement relied on personal resources (e.g., self-efficacy and optimism) and emotionally demanding conditions at work (Xanthopoulou et al., 2013). Furthermore, this effect was mediated by other personal resources, such as efficacy beliefs. In Finnish teachers, work engagement was positively associated with several job resources, e.g., job control, social support from the supervisor, and positive workplace climate (Hakanen et al., 2006). Research showed that job demands and resources operate as antecedents or drivers of work engagement. On the one hand, in service contexts, such as public administrations, as in this study, employees are usually expected to display positive emotions and suppress the expression of negative emotions in interactions with customers (Diefendorff & Richard, 2003). Emotional demands require energy expenditure, and job strain will probably occur. On the other hand, job resources exist and may be related to work engagement. In this study, we consider that leadership has a crucial role in organizations and, therefore, it is an important social resource that positively affects work engagement.

Transformational Leadership

Transformational leadership is performed by leaders who transform values, needs, aspirations, desires, and priorities of followers, and persuade them to perform beyond expectations (Bass & Avolio, 1994). Bass (1985) further described transformational leadership as a process that emerges from four elements: “charismatic leadership or idealized influence” (i.e., leaders focus on their followers’ individual needs in order to achieve general wellbeing); “inspirational motivation” (i.e., leaders are able to inspire their followers by introducing an appealing perspective on things to come); “intellectual stimulation” (i.e., leaders challenge their followers to look at issues from various perspectives and create new ideas); and “individual consideration” (i.e., leaders pay personal attention to the particular needs and capacities of their followers). Transformational leaders influence others because they are able to motivate and inspire them to achieve organizational goals in an efficacious way (Bass & Avolio, 1994; Yukl, 2002). Furthermore, transformational leadership behaviors may make job resources more available to the followers, who will probably feel supported by their leader and experience greater autonomy in doing their tasks when s/he takes their needs into account. Finally, transformational leaders delegate tasks based on followers’ needs and competences (intellectual stimulation), which means that they provide each follower with challenging but reasonable tasks, thus helping their followers to develop and create.

Research has found that transformational leaders influence organizational outcomes such as organizational commitment, organizational citizenship behavior, job satisfaction, and in-role performance (Kovjanic et al., 2012; Kovjanic et al., 2013). Associations between transformational leadership and work-related attitudes and behaviors, such as job satisfaction, have been well established (Bass et al., 2003; Kovjanic et al., 2013). Basically, both empirical and meta-analytic studies suggest that followers with transformational leaders show more job involvement, job satisfaction, and motivation in the workplace, as well as greater organizational trust and commitment (see Christian et al., 2011).

Many studies have shown the relationship between leadership and wellbeing (Nielsen & Daniels, 2012). Arnold (2017) published a review paper focusing on this relationship. Leaders also have a strong influence on employee happiness by creating less psychological distress and other negative outcomes and improving general psychological wellbeing and positive states of mind (Syrek et al., 2013). In addition, a diary study showed the positive relationship between transformational leadership and positive affective states, and the negative relationship between this leadership style and negative affective states (Lanaj et al., 1916). Wu and Wand (2015) also showed the effect of leadership on proactivity through positive affect, and Kranabetter and Niessen (2017) indicated that when managers are role models for health, employees benefit more from the transformational leadership style. Less clear, however, are the psychological mechanisms that explain why leaders can foster positive health and wellbeing outcomes in their followers (Tuckey et al., 2012).

Through cross-sectional and longitudinal designs, research has shown that transformational leaders positively influence work environment (Nielsen et al., 2008; Tuckey et al., 2012). The present study contributes to the literature by providing information about the effect of leadership on followers’ perceptions of job demands (i.e., emotional demands). Leadership can affect workers’ perceptions of emotional demands and, therefore, the effects of these demands on work engagement.

The present study contributes to the literature by examining the relationship between transformational leadership and followers’ perceptions of job demands, using a multilevel model to explain the effectiveness of transformational leadership in dealing with job demands (i.e., emotional demands) and their effects on work engagement.

Within this theoretical framework, we hypothesize:

Hypothesis 1: Individual level effect: emotional demands are negatively related to work engagement.

Hypothesis 2: Cross-level effect: leadership (level 2) is positively related to work engagement (level 1).

Hypothesis 3: Cross-level moderation effect: leadership (level 2) moderates the relationship between emotional demands and work engagement (level 1).

In summary (Figure 1), the aim of this study is to determine how emotional demands and transformational leadership are related to work engagement in a multilevel analytical context (i.e., individual and group levels). In general, we propose two levels of analysis: individual-level analysis (emotional demands and work engagement) and group-level analysis (i.e., transformational leadership). Moreover, another important aspect of our multilevel model consists of cross-level influences between group- and individual-level variables (i.e., leadership and work engagement).

Method

Sample and Procedure

Convenience sampling was used to obtain a sample of 1,079 employees from four Spanish municipal public administrations. These employees were nested in 124 work teams. Participants’ jobs were similar because the organizations were all public administrations. The sample consisted of 59.6% females. Group size ranged from 5 to 15 people. For data collection, we previously contacted stakeholders in each organization, presenting main objectives and study aims and requirements. Then, we explained that participation was voluntary, that only aggregated data would be reported, and that all identifying information would be removed in order to guarantee anonymity. Groups were natural work teams that usually worked together and were supervised by a leader. Employees were considered members of a team if they had the same supervisor and set of standards and guidelines in working to achieve common goals and their tasks were interdependent. Data collection was performed using an online questionnaire. Each group was identified by a code, and all the people in the group were identified with the same code.

Measures

Emotional demands. We assessed emotional demands from the RED.es Questionnaire (Resources, Emotions, and Demands; Salanova et al., 2007), referring to emotional demands at work. This scale includes four items (e.g., “My job requires me to deal with difficult and special people”).

Transformational leadership. We measured transformational leadership using four items from the RED.es Questionnaire (Salanova et al., 2007), based on Bass and Avolio’s (1990) scale, (e.g., “My supervisor is capable of organizing and allocating responsibilities”).

Work engagement. We measured work engagement using the Spanish version of the UWES (Utrecht Work Engagement Scale; Schaufeli et al., 2006), with three dimensions: vigor (three items; e.g., “At my work, I feel bursting with energy”), dedication (three items; e.g., “I am enthusiastic about my job”), and absorption (three items; e.g., “I feel happy when I am working intensely”). Scores ranged from 0 (never) to 6 (always) on the three response scales.

Data Analyses

Different data analyses were performed. First, internal consistencies were calculated (Cronbach’s alpha), as well as intercorrelations and descriptive data for the study variables using SPSS 21.0. Second, because the leadership variable is measured at work-unit level, we aggregated individual perceptions to group level, and checked the agreement among individual perceptions. Thus, in order to justify the use of aggregated scores for the study variables, we calculated interrater agreement on these measures with the rwg(j) index (James et al., 1993). Then, we examined the intraclass correlations, ICC (1) and ICC (2), at work-unit level. ICC (1) estimates the proportion of variance between participants that could be explained by differences in group membership, and ICC (2) estimates the reliability of aggregated scores at group level (James, 1982). Analyses of variance (ANOVAs) were also performed in order to test whether there was any statistically significant between-group discrimination for measures used in the study.

Finally, our statistical analysis considers a macro-micro multilevel context (Snijders & Bosker, 1999). In a macro-micro multilevel situation, a dependent variable measured at lower level (i.e., individual) can be explained by variables measured at a higher level (i.e., group level). In this study, a dependent variable (i.e., work engagement) measured at lower level (level 1) is assumed to be affected by an antecedent and contextual variable (i.e., emotional demands), also measured at lower level, and by explanatory variables (i.e., transformational leadership) measured at higher level (level 2).

Our data were hierarchically structured, so that 1,079 individual cases (level 1) were nested in 124 work- units (level 2). We analyzed data through hierarchical linear modeling (HLM) (Hofmann & Gavin, 1998; Hox, 1995) using LISREL software. This method is suitable for the analysis of data in a nested structure, constructing a separate model at each level in data structure (Bryk & Raudenbush, 2002). Thus, we can simultaneously make inferences about the effects of variations in independent variables at the individual level (i.e., emotional demands) and at the group level (i.e., transformational leadership) on dependent variables (i.e., work engagement), as well as about the cross-level moderating effect of the independent variable on the dependent variable. We also decided to center predictor scores relative to the mean of the entire sample – grand-mean centering, as Hoffman and Gavin (1998) recommend. Finally, we carried out an analysis focused on the interaction effect of leadership x emotional demands on engagement.

Results

Descriptive Analyses

Table 1 shows means, standard deviations, internal consistencies (Cronbach’s alpha), and intercorrelations. All the scales showed acceptable internal consistency. As expected, all the variables were significantly related to work engagement (see Table 1).

Aggregation Analyses

To statistically demonstrate within-team agreement and between-team differences, we conducted several analyses. First, we tested within-group interrater reliability by computing rwg (James et al., 1993). For the team variable (transformational leadership), results of the rwg(j) index revealed strong agreement among team members. The rwg(j) value for group transformational leadership was .72. Traditionally, an rwg of .60 provides sufficient evidence to warrant aggregation (Glick, 1985).

Next, we compared the variability between and within a sample of teams by computing intraclass correlation coefficients. ICC (1) and ICC (2) values for the group transformational leadership variable were .15 and .62, respectively. Conventionally, values greater than .12 for ICC (1) and .60 for ICC (2) provide sufficient evidence to warrant aggregation (Bliese, 2000; James, 1982; LeBreton & Senter, 2008).

Multi-level Analyses and Testing of Hypotheses

Although problems with common method variance due to data self-reported from one source may be overstated, there are potential concerns that common method variance might influence the results. Using AMOS 23.0, Harman’s single factor test (Podsakoff et al., 2003) was carried out for the study variables in order to test for common method variance bias. The results revealed a poor fit of the one-factor model (RMSEA =.203, IFI = .665, NFI = .658, TLI = .390, CFI = .657). Additionally, a model with three latent factors (emotional demands, transformational leadership, and engagement) was tested. The three latent factor model provided a better fit to the data (RMSEA = .060, IFI = .978, NFI = .966, TLI = .951, CFI =.979). Therefore, we found no evidence that common method variance had a significant impact on the results.

Next, data were analyzed via hierarchical linear modelling (HLM). Table 2 summarizes the results of the effects of emotional demands and transformational leadership on work engagement.

In this table, Model 0 refers to the Null Model. Model 1 proposes a relationship at individual level. This model predicted a negative relationship between emotional demands and engagement (H1). The relationship was positive and significant (β = .18, p < .001). Overall, these results do not support Hypothesis 1.

Hypothesis 2 proposed a cross-level relationship and predicted that transformational leadership would have a positive influence on work engagement. As Model 2 shows, this relationship was positive and significant (β = .30, p < .001) and, thus, supported Hypothesis 2. To test Hypothesis 3, we entered the cross-level moderation (Model 3), which predicted that transformational leadership would moderate the relationship between emotional demands and work engagement. In the analysis of engagement, the cross-level transformational leadership x emotional demands interaction was significant and positive (β = .10, p < .01). These results show the moderator effect of transformational leadership on the relationship between emotional demands and work engagement. Through subsequent tests, we have calculated the effect size (f²) in the three hypotheses (f² = .06, f² = .07, and f² =.11 respectively). In all cases the value has been low-moderate (Chen et al., 2010).

Finally, we carried out an analysis focused on the interaction effect of leadership x emotional demands on work engagement, in order to examine the differential effect on engagement of high scores on emotional demands (emotional overload) and low scores on emotional demands and leadership. Values of the moderator variable were chosen at 1SD above and below the mean (Cohen & Cohen, 1983; Jaccard et al., 1990). This result shows that workers who have more work engagement also have high leadership perceptions. In the case of low emotional demands, the level of engagement is similar at both leadership levels, high and low, but the level of engagement is different when there are high emotional demands. Workers with high levels of emotional demands and high levels of leadership have high levels of engagement. The level of engagement is low in workers with high levels of emotional demands and low levels of leadership. Leadership buffers the relationship between emotional demands and work engagement in such a way that these effects become strongly negative for employees with emotional overload. This significant interaction effect is graphically represented in Figure 2.

Discussion

This study contributes to the current knowledge about leadership’s positive effects in various ways. First, we demonstrated the positive effect of leadership on work engagement using a multilevel model with two levels of analysis (i.e., individual and group). However, cross-sectional and longitudinal designs have been used (Nielsen et al., 2008; Tuckey et al., 2012) to demonstrate that a transformational leader positively influences work environment. The results confirm that perception of transformational leadership by team members is positively related to their levels of work engagement. Thus, the higher the perception of our boss as a positive leader, the higher our levels of engagement at work. Previous studies have shown the relationship between transformational leadership and followers’ wellbeing (Arnold, 2017). However, these studies were cross-sectional and did not consider collective and multilevel aspects in this relationship. Our study contributes by moving the research a step forward and considering leadership as a multilevel phenomenon.

In addition, we extended the JD-R Model (Bakker & Demerouti, 2017) to other levels in organizations, analyzing team members’ shared perception of their transformational leader (level 2) and the psychological mechanism of emotional demands (i.e. emotional overload) to explain the effects of leadership on work engagement (level 1). Therefore, transformational leadership plays a role as a buffer or stimulator of work engagement through the interaction effect between emotional demands and engagement. Transformational leadership has been related to job resources, which in turn may be related to lower strain levels in followers (Schaufeli & Bakker, 2004). To summarize our findings, the hypothesized relationships among transformational leadership, emotional demands, and work engagement are supported by highly significant positive multilevel relationships.

Moreover, we found a counter-intuitive result regarding the relationship between emotional demands and work engagement. Research has shown that emotional demands are sometimes negatively related to wellbeing and sometimes not. For example, research has shown that employees with emotional overload display more mental and emotional distress and worse performance (Castillo-Gualda et al., 2018; Johannessen et al., 2013; Wang, et al., 2013), whereas other findings have shown a potential positive effect of emotional demands on wellbeing (Bakker et al., 2007; Xanthopoulou et al., 2013). This latter result can be explained by the perception of emotional demands as challenging and motivating demands. Our study takes a step forward by showing a motivational mechanism that can explain the relationship between emotional demands and work engagement, i.e., positive leadership. Nevertheless, when we consider high/low levels of emotional demands, the relationship with work engagement changes. High levels of emotional demands (i.e., emotional overload) showed a negative relationship with work engagement, and low emotional demands showed a positive relationship with work engagement. Our findings suggest that leadership moderates the effect of emotional demands on work engagement. Leadership has an interaction effect between emotional demands and engagement. In the case of low emotional demands, the level of engagement is similar for workers who perceive high and low leadership. But when there is an emotional overload, only workers who perceive strong transformational leadership achieve high levels of engagement. The perception of shared leadership by team members directly affects their levels of work engagement, and indirectly through the interaction effect on the relationship between emotional demands and work engagement. When employees feel emotionally overloaded, their leaders can mitigate this negative effect on their work engagement levels. Our study tries to address Tuckey et al.’s (2012) challenge to discover the psychological mechanisms that explain why leaders can foster positive health and wellbeing outcomes in their followers, in our case, work engagement.

Practical Contributions

This study expands the knowledge through a multilevel model highlighting the effectiveness of transformational leadership and its effects on service work engagement. The results suggest that shared perceptions of work teams about their leader can facilitate workers’ engagement when they face high emotional demands due to their work. In these situations, the role of the leader is an important social resource for employees. A practical proposal would be to provide training in practical skills to cultivate transformational leadership dimensions in supervisors. This would help employees to deal with the demanding aspects of an effective customer-service relationship, and would contribute to empowering workers’ engagement by providing them with social job resources. The idea of enriching work environments in order to effectively perform jobs involving emotional demands is consistent with previous research (Bakker & Demerouti, 2013, 2017; De Jonge et al., 2008). It is necessary for organizations to create environments that promote work engagement by fostering transformational leaders in a way that leads to a steady growth and better performance.

Limitations and Future Research

One potential limitation of this study is that the data at both individual and work team levels were obtained through self-report measures. However, in terms of the data obtained at individual level, we understand that it is not a relevant problem because findings are consistent with the theory, and common method variance is likely to attenuate rather than increase the effects of the interaction (Evans, 1985). Regarding group level, we considered aggregated perceptions of transformational leadership among team members. Therefore, when using these data added to work teams, we were able to increase validity of scores, considering that they were “intersubjective” perceptions shared among team members, and not individual subjective scores.

We used a convenience sample consisting of employees and work teams belonging to the same sector (four town councils). Although they all belonged to the Spanish Public Administration, their work environments could be different, which would indicate a tendency toward a heterogeneous sample. In future studies, the sample should be more heterogeneous and belong to different economic sectors. Moreover, it should be noted that, although transformational leadership was assessed only at work team level, a future challenge would be to investigate whether individual perceptions of leadership have the same influence on positive results as aggregated perceptions at work-team level.

Another limitation of the present study is its cross-sectional design. Our relational model implies causal mechanisms, but relationships in this study cannot be interpreted in a causal direction. Future research should carry out longitudinal studies. In terms of a process described by Schaufeli et al. (2009) as a “positive gain spiral”, it is also possible that employees’ work engagement increases or facilitates transformational leadership behavior in a reciprocal way over time.

The last limitation is related to the effect size obtained in our study whose values are low-moderate. However, despite this limitation, we can consider the results are significant and support an interesting multilevel line of study.