Finally, we ran a model in which the empowerment scale was modeled as a second-order factor composed of 4
subscales (meaning, competence, autonomy, and impact). As shown in Table 1, this model fit the data well. As a whole, these
results suggest that the scales used in sample 1 were distinct from one another and that collapsing the empowerment items into
one scale was appropriate.
We estimated a similar series of models for sample 2. We began with a 1-factor model (see Table 1 ) which did not fit the data
well. We followed this with a 4-factor model in which each variable was represented by a different factor: LMX, empowerment,
oCB, and job performance. Again, while model fit improved, the overall fit was only moderate. These results were not surprising