You are on page 1of 1

some studies but other researchers (Doney and Cannon, 1997; Kumar et al.

, 1995b;
Mavondo and Rodrigo, 2001) have ignored its inclusion in the measure. Therefore,
deletion of the two items and using four items measuring trust would not affect the
content and face validity of the measurement.

Table 5.6
Summary of Initial Findings (CFA): Trust
Quest. Item wording Initial Final
Items Standardised Standardised C.R.
Loadings Loadings (t)
Trust1 Our major supplier has generally been
honest
.85 .87 16.19
Trust2 Our major supplier is truthful .91 17.27
.90
Trust3 Promises made by our major supplier
are reliable
.79 .79 13.91
Trust4 Our major supplier is open in dealing
business with us
.81 .78 13.80
Trust5 Our major supplier has a high degree
of integrity
.74
Trust6 Transactions with our major supplier
do not need close supervision
.43

Achieved Fit Indices


Md._Abu_Saleh_Thesis 2006
CMIN/DF RMSEA IFI TLI CFI
(χ2/df)
3.376
Initial .101 .974 .956 .974
(30.38/9)
Final 3.135 .096 .993 .978 .993
(6.27/2)
Composite Construct Reliability .91

The composite construct reliability for this four-item measure is .91 which is well
above the acceptable level as indicated in the literature (Hair et al., 1995). This
indicated that the retained four items are considered reliable as well as valid for this
construct measure.

5.2.2.2 Communication: Initial Findings

Communication was measured by five items. The initial analysis of the inter-item
correlation matrix revealed that communication ‘item 5’ was relatively poorly
correlated with all other items in the scale. All five items were subjected to a CFA.
The initial CFA results are exhibited in Table 5.7 which indicated that the model was

110

You might also like