Understanding consumer online shopping behaviour from the perspective of transaction costs

Table 5.15 Preliminary Data Analysis- Perceived Enjoyment (ENJ) 
Step One - Exploratory 
Wording EFA 
 
Loadings 
ENJ1 
ENJ2
ENJ3
ENJ4 
ENJ5 
My favourite leisure activities 
.80 
ENJ1 1.00 
When shopping online, I feel playful 
.89 
ENJ2 .722 
1.00 
Assess online shopping as enjoyment .90 
ENJ3 .623 
.769 
1.00 
Enjoy 
shopping 
on 
the 
Internet 
.90 
ENJ4 .589 .714 .784 1.00
Online shopping brings me happiness .85 
ENJ5 .567 
.646 
.689 
.782 
1.00 
Reliability 
0.92 
KMO 
.857 
Variance Explained 
75.12 
Bartlett’s 
.000 
Step Two – Confirmatory 
Indicators CFA 
 
Stand. Loadings 
Stand. 
Error 
Critical 
Ratio
AVE CR 
ENJ1 .68 
-- 
-- 
.66 
.907 
ENJ2 .84 
.042 
28.61
*
ENJ3 .91 
.053 
24.41
*
ENJ4 .86 
.054 
23.52
*
ENJ5 .76 
.050 
22.04
*
Model Fit 
Statistic 
Model Fit 
Statistic 
χ
2
2.375 
TLI 
.998 
df
2 
CFI 
.997 
P .305 
RMSEA 
.014 
GFI .996 
RMR 
.008 
* 
p 
< .001
5.6 OVERALL MEASUREMENT MODEL 
The results of the scale by scale CFAs were used as the basis for constructing the overall 
measurement model. Specifically, the items retained from scale by scale CFAs (refer Step 
Two in Figure 5.1) in relation to nine antecedent variables, consumer TCs, customer 
satisfaction, customer loyalty, risk-bearing propensity and perceived enjoyment were used to 
construct the overall measurement model. The overall measurement model was established 
by performing a CFA via AMOS 21.0 using the maximum likelihood method (Byrne 2009). 
According to the two-stage model-building approach proposed by Anderson and Gerbing 
(1988), the researcher first tested the model fit, convergent validity and discriminant validity 

---

245
of the measurement model as well as the common method variance in this section, then 
examined the research hypotheses and structural model framework in the next section. The 
statistical results of the overall measurement model are presented in Table 5.16, which 
includes the standardized factor loadings, their associated standard errors and critical ratio, 
the AVE for each construct, CR, and the key overall model fit statistics. Next, the results of 
testing the overall measurement model were reported.