H 22 (EPA→JP) = 0.279 = medium;
H 4 (ME→JP) = 0.258 = medium;
H 21 (ML→EPA) = 0.213 = small to medium; and
EM→EPA→JP = 0.174 = small to medium.
The final path in the list above was not expressed as a hypothesis owing to it
being a fully-mediated path (i.e. there is no direct line connecting EM with JP).
The indirect effect of EM on JP (= 0.174) is, however, described in Table 7 - 30
along with the other standardised total effects for Model 5b. This mediated path
represents the indirect effect of EM on JP that has been ‘transmitted’ through EPA
and is calculated as the product of the two direct effects 0.626 * 0.279 = 0.174
and classed as a small to medium effect.
7.15 Multi-group analyses............................................................
SEM analyses can also be used to investigate and assess the ways in which effect
sizes (relationships between variables) are moderated by categorical independent
variables such as demographic or employment characteristic variables. Where
effect sizes are significantly moderated and the difference in effect sizes between
groups is deemed to be considerable, there may be implications with regard to
the ways in which organisations manage different groups of employees.
Moderator effects such as these can be assessed using multi-group SEM analyses
and this section describes how multi-group analyses are employed to:
(i) highlight differences in effects sizes across demographic grouping (gender, age
etc). Any such variations in effect sizes can then be considered in the light of
survey non-response to assess how non-response may be influencing the
research findings;
(ii) assess the moderator effects of respondents’ degree of supervisor contact and
perceptions of training and information provision; and
(iii) examine to effect of employee work orientations in moderating inter-factor
relationships by using Work Values as a grouping variable.