leadership and motivation in hospitality

The re-estimated model (SEM 5:2) is illustrated along with the parameter
estimates and model fit statistics in Figure 7 - 20.

Figure 7-20 Estimates and model fit for SEM 5:2

The model fit is now satisfactory, however, the WV→EPA path remains non-
statistically significant (p = 0.255;  = 0.092). This situation is somewhat
unexpected because the previous model iteration (Model 4, without Employee
Empowerment) estimated the WV→EPA path at p <0.001 and  = 0.334.

An examination of the correlations (r) between the independent variables
provides a possible explanation for this phenomenon. Specifically, the moderately
high correlations^17 (EM  WV, r = 0.578 and EM  ML, r = 0.574) between the
Employee Empowerment construct and the other two independent variables
(Work Values and Motivational Leadership) draws attention to a potential problem
with multicollinearity between independent variables. Cohen et al. (2003)
describe the difficulty that this situation generates (in a general multiple
regression context) as follows:

...as one of the independent variables, Xi becomes increasingly

correlated with the set of other IVs [Independent Variables] in the

regression equation, Xi will have less and less unique information that

17
Kline (2005: 56-57) describes high multicollinearity in SEM analysis as being
indicated by inter-factor correlations greater than approximately r = 0.85