Procedures
In order to test the study’s hypothesis, a procedure was needed that could tap workers’
turnover intentions, and advantageously combine the observed variables into a
construct that represented these states (Joreskog and Sorbom, 1988; Shumaker and
Lomax, 1996). Structural equation modeling (SEM) is especially useful for this analysis
because latent variables of interest in this study cannot be directly measured, but can
be alternatively examined by using other measurable, or manifest, variables (Joreskog,
1993; Shumaker and Lomax, 1996). In this study, two latent constructs were measured:
worker creative environment perceptions and intent to turnover. These factors were
evaluated using the attendant scales discussed in the measures section. SEM is also
important for measurement bias reduction reasons. While it is often common practice
to combine multiple indicants (such as questions) of a single theoretical construct (such