• Keine Ergebnisse gefunden

Does leadership behaviour predict job characteristics across time?

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

6.1 Longitudinal relationships

6.1.2 Does leadership behaviour predict job characteristics across time?

Testing normal causality

We started by exploring in the first model (M1) the statistically best predictors of each job characteristic at T2 sought from leadership behaviours at T1 (see figure 6.1 ar-rows 1-2) using stepwise linear regression. After that, each baseline job characteris-tic at T1 was controlled for in the second model (M2) to see whether leadership be-haviour explained the change in job characteristics between T1 and T2. Only signifi-cant predictors from the first model (M1) were entered as predictors in the second model (M2), in addition to the baseline level of each job characteristic. Predictors which did not correlate with the dependent variable were removed only from those models where the beta-coefficient turned out to be significant and in an unexpected direction. We repeated this procedure in examining the relationships between T2 and T3 and T1 and T3. All examined normal causality relationships across varying time lags are shown in tables 6.1-6.6, separately for job demands and job resources.

Tab. 6.1 Leadership behaviour at T1 predicting job demands at T2 (time lag 15 months)

Workload

Note. Transformational leadership was not included as a predictor in the job insecurity mod-el, as it did not correlate with job insecurity (r = -.08). Contrary to the positive beta coefficient (marked in red) in the multiple regression model, authentic leadership had a negative zero-order correlation with job insecurity (r = -.13).

Tab. 6.2 Leadership behaviour at T2 predicting job demands at T3 (time lag 8 months)

Note. Contrary to the positive beta coefficient (marked in red) in the multiple regression mod-el, authentic leadership had a negative zero-order correlation with job insecurity (r = -.10).

Tab. 6.3 Leadership behaviour at T1 predicting job demands at T3 (time lag 22 months)

Workload

Note. None of the leadership constructs correlated with workload (correlation with transfor-mational .03, authentic .04, fair -.06, health-promoting -.07, abusive .10). Contrary to the pos-itive beta coefficient (marked in red) in the multiple regression model, authentic leadership had a negative zero-order correlation with job insecurity (r = -.12).

Tab. 6.4 Leadership behaviour at T1 predicting job resources at T2 (time lag 15 months)

Role clarity

Note. Abusive leadership was not included as a predictor in the role clarity model, as it did not correlate with role clarity (r = -.04). Contrary to the negative beta coefficient (marked in red) in the multiple regression model, transformational leadership had a positive zero-order correlation with job autonomy (r = .17).

Tab. 6.5 Leadership behaviour at T2 predicting job resources at T3

Note. Abusive leadership was not included as a predictor in the models as it did not correlate with role clarity, autonomy or meaning of work (r = -.06, -.05 and -.06, respectively). Contrary to the negative beta coefficient (marked in red) in the autonomy model, transformational leadership had a positive zero-order correlation with job autonomy (r = .21) and so did fair leadership (r = .16).

Tab. 6.6 Leadership behaviour at T1 predicting job resources at T3 (time lag 22 months)

Role clarity

Note. Abusive leadership as a predictor was removed from the models for autonomy and meaning of work, as it did not correlate with autonomy or meaning of work (-.04 and -.09, respectively).Contrary to the negative beta coefficient (marked in red) in the job autonomy model, transformational leadership had a positive zero-order correlation with job autonomy (r

= .12) and so did fair leadership (r =.16).

From these tables 6.1-6.6 we can see that the proportion of the variance in job de-mands explained by leadership behaviour was small (range 1-12 %). Job insecurity seemed to be best and workload least explained by leadership behaviour. The ex-planation rates were slightly higher for job resources (range 7-23 %), of which job autonomy seemed to have the highest and role clarity the lowest explanation rate.

When each baseline measure was controlled for the explanation rates rose to 39-62

% for job demands and to 39-54 % for job resources. It seemed that perceived job insecurity was slightly less stable across time (β = .58-.67) than the other job mands (β = .68-.79) which may partly explain why it was the best explained job de-mand. The same concerned to some extent also job autonomy (β = .56-.66) com-pared to other job resources (β = .62-.74). There was a tendency that the rank-order stability of job demands as well as job resources decreased along with the lengthen-ing of the time lag between the measurements.

A more detailed examination revealed the following significant relationships for job demands. Of the leadership behaviours examined, only health-promoting leadership predicted one job demand, that is, job insecurity across different time lags (from 8 months to 22 months) when the baseline level of job insecurity was controlled for.

Thus, it seemed that health-promoting leadership had the power to explain the change (a decrease) in job insecurity over time. Of the other leadership behaviours, abusive leadership was significantly and positively related to workload and cognitive and emotional demands across all time intervals examined, but when the baseline levels of these job demands were controlled for, the relations were not any more sig-nificant. Therefore, it seemed that abusive leadership did not explain changes in these job demands across time.

When looking at job resources, again health-promoting leadership seemed to be the most important of the leadership behaviours. It explained both job autonomy and meaning of work when their baseline levels were controlled for. Health-promoting leadership predicted an increase in job autonomy across all time intervals examined, and it seemed to explain an increase in perceived meaningfulness of one’s job on the long-term, that is across 15 and 22 months. In addition, transformational leadership had the unique power to explain an increase in role clarity across 15 and 22 months, when baseline role clarity was taken into account.

Of the examined leadership behaviours, authentic leadership and fair leadership be-haviour did not belong to the best predictors in relation to employees’ job demands and resources. However, it is good to keep in mind that all the leadership behaviours were in the analyses at the same time (M1-model), that is, their unique significance was compared with each other and the best predictors were selected on the basis of this comparison. Thus, it is possible when each leadership behaviour is examined separately, that also those having non-significant role when examined together with other leadership behaviours may turn out to be significant predictors.