variable but with due consideration for the substantive implications this may have
for the model.
Considering the four indicator variables in the initially specified model that have
been flagged for attention:
ML4 has a moderately low loading of 0.602 and two standardised residuals
2.58 (at 3.295 and 3.264)
ML5 has moderately low loading of 0.606 and one standardised residual >2.58
(at 3.264)
JP3 has a moderately low loading of 0.556 and one standardised residual
2.58 (at 2.605)
JP4 has a lower loading (0.494), one standardised residual >1.96 (at 2.359)
and two standardised residuals >2.58 (at 2.605 and 3.308)
o JP4’s very large standardised residual covariance (3.308) is shared with
DSB4 flagging DSB4 as a candidate for removal
Considering the substantive implications of removing these variables (this is, how
their removal affects the construct’s content validity, in other words, the meaning
of the construct). Both ML4 and ML5 were included somewhat speculatively (see
Section 5.2) and their removal from the construct will not affect the core
substantive content of the ML construct. ML4 relates to the recognition/feedback
dimension of Motivational Leadership and ML5 describes the leader putting the
group interests before their own. Removing these would leave three ‘core’ items
(ML1, 2 and 3) related to transformational leaders’ ‘vision→goal→effort’
behaviour. Removing ML4 and ML5, then, results in a very clear vision- / goal-
related ML construct that very closely reflects the central goal-setting, goal-
clarification and goal-encouragement dimensions of Motivational Leadership as
described by Avolio and Bass (2004a: 96) (see Section 2.4.3).
Secondly, considering JP3 and JP4. JP3 is related, substantively, to JP1 – both
relate to intensity of work (JP1 = Try to work harder; JP3 = Find that you have
done more than you expected to do). Therefore, if JP3 was removed from the
model, the substantive content of work intensity remains in the construct.
JP4, on the other hand, has an important substantive role to play in the wider
model. Considering the relationship between the Job Performance (JP) construct
and the Discretionary Service Behaviour (DSB) construct: DSB was designed (by
Blancero and Johnson 1997 and 2001) specifically for application in a services
context and all of its items explicitly relate extra effort to the customer service