AMOS’s bootstrap estimation function which indicated that all parameters are
robust under the conditions of multivariate non-normality (see Appendix IV).
Figure 7-7 Estimates for SEM 2:1
The structural coefficient ML→DSB (H 2 ) remains at much the same magnitude as
in Model 1 (0.400 compared with 0.396 previously). Reflecting the inclusion of
Work Meaning as a mediating factor, the ML→JP path (H 1 ) has, however, changed
considerably from 0.414 in Model 1 to 0.258 in Model 2. This is because some of
the effect of ML on JP is now being ‘transmitted’ by the mediating variable ME.
Testing for mediator effects
It is necessary now to statistically assess the extent to which the influence of
Motivational Leadership (ML) on Job Performance (JP) is mediated by Work
Meaning (ME). Accordingly, the procedures outlined in Hair et al. (pp. 866-870)
are followed to assess whether ME is acting as a full or partial mediator.
Firstly, a chi square difference test was undertaken to compare the difference in
global model fit between a constrained SE model (Figure 7 - 8 ) which did not
contain the ML→JP path (that is, only the indirect ML→ME→JP path sequence was
specified) and the hypothesised model SEM 2:1 – the constrained model and its
estimates are illustrated in Figure 7 - 8.