new construct. Theoretical support for the new relationships (ML→EA and EA→JP)
is provided by the values→attitudes→behaviour (V→A→B) hierarchy described by
Homer and Kahle (1988: 638-639) and described in Section 4.2 above.
Theoretical integration of the new construct and hypotheses will be re-visited (i)
when (and if) a valid construct is revealed and (ii) in the light of the construct’s
relationships with the other variables in the model.
As with previous models, Motivational Leadership is initially specified with 5 items
and Job Performance and Discretionary Service Behaviour with 4 items each. The
first step is to use AMOS’s model-based imputation procedure to replace the
missing values. Following this, the initial model is estimated.
Construct Item Standardised factor loading estimat es
ML EA JP DSBMotivational
LeadershipML1 .893 (^)
ML2 (^) .935 (^)
ML3 (^) .896 (^)
ML4 .618 (^)
ML5 .619 (^)
Employee Attitudes
ME7 .264 (^)
ME6 .799 (^)
ME5 .592 (^)
ME3 .827 (^)
ME1 .847 (^)
AC4 .821 (^)
AC2 .782 (^)
AC1 .788 (^)
JS6 .825 (^)
JS5 .686 (^)
JS4 .604 (^)
JS3 .450 (^)
JS2 .554 (^)
JS1 .747 (^)
Job Performance
JP1 .852 (^)
JP2 .856 (^)
JP3 .554 (^)
JP4 .494 (^)
Discretionary
Service Behaviour
DSB1 .801
DSB2 .773
DSB3 .673
DS B4^ .696
Average variance
extracted (AVE)
0.500 0.521 0.557 0.585
Construct
reliability (CR) 0.750^ 0.935^ 0.826^ 0.849^
Model fit statistics
χ^2 = 751.420; d.f. = 318; sig = 0.000
RMSEA = 0.080 (0.088; 0.073; pclose = 0.000)
CFI = 0.876
SRMR = 0.0754
CN (0.05) = 102