• Keine Ergebnisse gefunden

Discussion and conclusions

Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr

6.1 Longitudinal relationships

6.1.11 Discussion and conclusions

This study of longitudinal relationships between leadership, job characteristics and employee wellbeing had four main aims. First, we aimed to examine whether leader-ship behaviour can have an effect on job characteristics and, second, on the wellbe-ing of employees. The analyses were conducted on three different time-lags with several constructs on leadership, job characteristics and wellbeing. Analyses were performed to find out whether leadership predicts change in job characteristics and wellbeing of employees (normal causation), and whether job characteristics and wellbeing themselves predict change in leadership behaviour (reversed causation). In these analyses we identified the best predictors in both directions. In other words, among the set of leadership constructs, we determined the best predictors for job characteristics and wellbeing, and among job characteristics and wellbeing, respec-tively, the best predictors for leadership ratings. Third, we examined whether leader-ship behaviour predicts employee wellbeing beyond job characteristics conceptual-ized as job demands and job resources. Fourth, multiple mediator analyses were conducted to investigate, whether leadership can have an effect on employee wellbe-ing through job characteristics.

The results showed, first, that leadership predicts changes in wellbeing over time and wellbeing predicts changes in leadership over time. However, the relationships from leadership to wellbeing were rather weak in magnitude. It seems that wellbeing con-structs explained a larger proportion of leadership behaviour than leadership behav-iours explained in wellbeing. Regarding the amount of significant relationships when controlling for the baseline of the outcome variable, there were more reversed cau-sality effects (from wellbeing to leadership) than normal caucau-sality effects (from lead-ership to wellbeing). This may be due to the high intercorrelations among the

leader-ship constructs. The highly correlated leaderleader-ship constructs provide only few predic-tors which can explain unique variance in the wellbeing outcomes, while there are several independent predictors among the less correlated wellbeing constructs.

Leadership behaviours are generally and across time lags more related to the posi-tive indications of wellbeing, like work engagement, good team climate and organiza-tional commitment, than to indicators of ill-being, like exhaustion, irritation, somatic stress or turnover intentions. Health-promoting leadership emerged as the best pre-dictor of employee wellbeing besides transformational leadership which also was an important leadership construct in this relation.

Second, concerning job characteristics, leadership predicts changes in job resources but not in job demands. Job insecurity as a job demand makes an exception in this respect. Hence it seems that leadership can have an enhancing effect on job re-sources. The same pattern on relationships between leadership and positive con-structs was observed in terms of job characteristics and wellbeing. Compared to the other leadership behaviours, health-promoting leadership was somewhat better pre-dicted by job characteristics, especially by job resources and in the shortest time lag, from T2 to T3, also by job demands.

Third, results concerning the role of job characteristics in the relationship of leader-ship and employee wellbeing indicate that leaderleader-ship behaviour has more to do with job resources than job demands. This applies also to leadership’s relation to employ-ee wellbeing. The relationships betwemploy-een leadership and wellbeing were attenuated more by introducing job resources to the models than by introducing job demands.

After adjustment for baseline wellbeing and job demands, leadership behaviour ex-plained from 0.4 % to 2.5 % of later wellbeing. Adjusting the same models for job demands concurrent with wellbeing indicators, the proportion of leadership declined slightly to explain 0.4-1.6 %. Possibly, the leaders participating in our study were not in the kind of a position that they would have an influence on job demands like the workload of employees. Also, the job demands asked in our study seem to be em-bedded in the work itself. For example, the job may or may not include dealing with problems of other people depending on whether the job locates in social sector or on service branch. Similarly, the nature of the job determines whether there is a lot to process cognitively.

With regard to job resources, leadership behaviour still significantly predicts some of the wellbeing constructs over time when adjusted for job resources. However, the explanation rate of solely leadership behaviour in these cases is less than 1 % inde-pendent of whether job resources included in the model were concurrent with leader-ship behaviour or concurrent with wellbeing indicators. When job resources were concurrent with wellbeing, instead of the baseline measurement, there were only two significant leadership-wellbeing relationships left. Health-promoting leadership was related to increasing organizational commitment from T1 to T2, and transformational leadership was related to increasing organizational commitment from T2 to T3. In both cases, the only job resource staying significant was meaning of work.

Fourth, utilizing three measurement points, health-promoting leadership was exam-ined in five multiple mediation models via job characteristics in relation to work en-gagement, organizational commitment, occupational self-efficacy, team climate and

depressive symptoms. Besides being an independent predictor for organizational commitment, meaning of work was also identified as an important mediator between health-promoting leadership and organizational commitment. Meaning of work stayed significant in all the mediator models in which it was included, whereas the signifi-cance of the other proposed mediators varied according to the wellbeing construct. In addition to enhancing organizational commitment through increasing the meaning of work, health-promoting leadership also increases work engagement through increas-ing meanincreas-ing of work and decreasincreas-ing job insecurity. Health-promotincreas-ing leadership can also influence occupational self-efficacy through increasing cognitive demands, de-creasing job insecurity, and inde-creasing autonomy and meaning of work. Further on, good team climate can be promoted by increasing cognitive demands, role clarity and meaning of work. Finally, depressive symptoms can be reduced by decreasing job insecurity and increasing autonomy and meaning of work.

Generally, meaning of work seems to be an important job resource which leaders may be able to influence, while it also captures unique predicting power in relation to most of the wellbeing outcomes. The experience of doing meaningful work is likely to stem from a large variety of sources. Based on the results of this investigation, lead-ership behaviour – employees’ perceptions of it – can been seen as one of those sources.

Finally, despite the longitudinal setting and relatively large sample of our study, a number of important limitations need to be considered. Two major limitations, both of which are not unique to our study but common to this type of research, can be dis-cerned. First, the most prominent limitation concerns the fact that leadership behav-iour in this study equals to employee perceptions of leadership behavbehav-iour. In other words, leadership behaviour is based on self-reports of individual employees. These self-reports may be relevant in their own right, and it can be claimed that it is anyway the subjective evaluation that matters. Yet, there are several possible confounding effects in self-reports. Hence it remains unknown, to which degree leadership is in the eyes of the beholder, and to which degree perceptions of leadership behaviour are shared among employees rating the same leader. This latter question will be part of future analyses of the data using multilevel frameworks. Further on, we are not aware to which degree a relationship between a leader and a subordinate should be viewed as dyadic relationships genuinely differing from one individual to another.

Research on perception of transformational leadership has shown that rater person-ality (e.g., BONO, HOOPER, & YOON, 2012) and rater affect towards the leader (lik-ing) (BROWN & KEEPING, 2005) play a role in leadership ratings. Individuals high in extraversion, agreeableness, openness and conscientiousness report more trans-formational leadership behaviours than individuals low on these personality traits (BONO et al., 2012; FELFE & SCHYNS, 2010). With this regard, perception of trans-formational leadership may be positively biased. In contrast, the relationship between rater neuroticism and transformational leadership is found less consistently (BONO et al., 2012). These results led us to rethink the pattern found in our study. Leadership behaviour was more strongly related to job resources than job demands, and similar-ly, leadership behaviour was more related to positive aspects of wellbeing than to the negative side, ill-being. The possibility of the biasing effect of perception through pos-itive lenses should be considered when making recommendations for leaders and organizations. As FELFE and SCHYNS (2010) caution, over-attributing

transforma-tional leadership may have negative consequences and lead leaders to be confront-ed with unrealistic expectations that they cannot fulfil.

The second main limitation concerns the issue of causality. Longitudinal research has the advantage of enabling prediction and examination of change, but especially in the context of self-reports, true relationships between causes and consequences cannot be proved. In fact, concerning psychological variables with respect to which the temporal order of the variables is unknown, it has been argued that the time of measurement should not be confused with the time of actual occurrence (KEL-LOWAY & FRANCIS, 2013). In addition, the third variable effects remain a competing explanation even in the case of longitudinal studies. Thus in this study it remains un-known to which degree leadership behaviours actually produce wellbeing or change in job characteristics, and what is the role of wellbeing or job characteristics in the perception of leadership. The results of this study lend support to both interpreta-tions.

All the analyses on longitudinal relationships between study variables were conduct-ed on the individual level, despite the fact that team members of the same team gave leadership ratings on the same leader, and also likely share similar levels in de-mands and resources. In the next chapter we will deal with this nested structure of the data, and present some findings based on multilevel-modelling.