case, while the data for Model 1 were found to be univariate normal (Section
7.2.3), a multivariate C.R. (critical ratio) value of 10.9 for CFA 1:3 is indicative of
departure from multivariate normality. Accordingly, model CFA 1:3 was re-
estimated using AMOS’s bootstrap function. This analysis found that all of the
parameters were robust under the conditions of multivariate non-normality (see
Appendix IV for details of both the C.R. estimate and the subsequent
bootstrapped estimates).
With the measurement model now exhibiting adequate parameter estimates and
model fit statistics, it is possible now to move on to Step 2 and estimate the
structural regression model in order to test the research hypotheses.
7.3 Model 1, Step 2: the structural model
Having satisfactorily established the measurement model, the model can be
converted to its structural regression form. For Step 2, the indicator variables will
no longer be illustrated in the figures, as the main focus of attention is on the
structural parameters (the causal paths between constructs). With respect to
how the individual items load onto the constructs, the structural model is specified
in the same way as CFA 1:3, that is, with eight indicators in total; three loading
on ML, three on JP and two on DSB.
In contrast to Step 1, where the measurement model was specified with bi-
directional non-causal covariance arrows connecting each latent factor to all of the
others, the structural model (Step 2) replaces these connecting arrows with uni-
directional connections indicating causal relations. These uni-directional
connections represent the theory-driven hypotheses that underpin the
hypothesised model. These causal effects are illustrated in Figure 7 - 4 and
articulated by hypotheses H 1 and H 2 as follows:
(H 1 ) as employees experience greater levels of Motivational Leadership they
will also experience greater levels of Job Performance
(H 2 ) as employees experience greater levels of Motivational Leadership they
will also report greater levels of Discretionary Service Behaviour