i.e. an attitudinal outcome of motivational leadership. The research also
measures the Job Performance (JP) construct as a behavioural outcome of
motivational leadership on employees’ behaviour. The total effect of motivational
leadership on job performance is calculated as:
the direct effect, ML→JP, plus
the indirect, ML→EPA * EPA→JB.
The importance of the variation in the ML→EPA between Models 4 and 5b is
diminished when considering the total effects on motivational leadership on job
performance (ML→JPTOT). In Model 4 ML→JPTOT = 0.380 and in Model 5b
ML→JPTOT = 0. 315 ; this is a far smaller reduction than that seen in the direct
ML→EPA parameter estimate. Accounting for the effect of multicollinearity in
Model 5b on the magnitude of ML→JPTOT, we might expect that a more accurate
estimate for ML→JPTOT will lie somewhere between 0.315 and 0.380. Regardless
of whether the adjusted estimate lies towards the bottom or the top of the range,
it will be interpreted (according to Cohen’s 1988 categorisations) as a medium-
strength effect.
Model 5b represents the culmination of the model building exercise. Following
the development and estimation of Model 5b, the Social Support construct was
introduced in Model 6 and the Service Quality construct in Model 7. However,
neither Model 6 nor 7 could be developed into a structural model. In the case of
Model 6, the Social Support construct did not covary at a statistically significant
level with the Job Performance construct (p = 0.227) and therefore the model
was not suitable for further development as a structural model. In the case of
Model 7, the Service Quality construct did not covary at a statistically significant
level with any other construct.
For Model 6, the substantive implications of this outcome is the finding that Social
Support (peer support) is not likely to be a significant predictor of job
performance.
For Model 7, it appears that the exploratory method (see Section 5.4) for
measuring service quality based on employee perceptions of frequency of
maintaining customer satisfaction during service failure events simply did not
work. Indeed, feedback from one hotel manager in a participating hotel indicated
that some respondents did not fully understand the wording for the Service
Quality set of indicator variables (although, unfortunately, no such feedback was