people could be managed in such a way that organisational and personal
objectives could be met was promoted with the underlying rationale that by
humanising the workplace, employees could become more satisfied and more
productive. McGregor’s (1966) Theory X / Theory Y work described (a)
‘traditional’ (Theory X) leaders who believe employees are self-serving and
require inducements to achieve organisational goals and (b) more employee-
oriented (Theory Y) leaders who believe that, provided with a suitable work
environment, employees can be self-actualising and self-motivating.
Following this reasoning, and based on the premise that leadership is not the sole
motivating factor in the workplace, this research sought to measure the
contribution of work values to work motivation.
Work values (WV) were found to have a significant and positive effect on
employees’ positive attitudes (EPA) (WV→EPA = 0.334) and a smaller (fully
mediated by EPA) indirect effect on job performance (JP) in Model 4
(WV→EPA→JP = 0.091). These effects of work values were, however, negated
when employee empowerment was added to the model, suggesting that while
work values do have an influence on outcomes such as employees’ positive
attitudes and job performance, these effects are considerably diminished when
set alongside the more influential employee empowerment factor.
To further investigate the role of work values in influencing individual and
organisational outcomes, a multi-group analysis was conducted to investigate the
question of whether or not employees’ work orientations moderate the
relationships between constructs in structural models 2 and 5b.
The first step was to create a dichotomous categorical variable with which to
specify the groups. Following the guidelines for creating summated scales in Hair
et al. (2006: 135-139), the three Work Values items (WV 1, 4 and 5) that formed
the Work Values construct in Model 4 were used to create the summated scale.
In this way, the rules of thumb described by Hair et al. (2006: 139) (that the
variables should form a unidimensional construct with good reliability and which
demonstrates convergent, discriminant and nomological validity) are adhered to.
The distribution of the summated variable reflected those of the individual
variables insofar as the new variable was negatively skewed. The distribution is
illustrated in Figure 7 - 26.