Understanding consumer online shopping behaviour from the perspective of transaction costs

Table 5.8 Preliminary Data Analysis- Perceived Convenience (CONV) 
Step One - Exploratory 
Indicators Wording 
EFA 
 
Loadings 
CONV1 
Save time by shopping on the Internet 
.77 
CONV2 
Shop online where I can reduce my efforts 
.84 
CONV3 
Convenient to search product information 
.87 
CONV4 
Shop on the Internet at home 
.81 
CONV5 
Buy things on the Internet at any time 
.77 
CONV6 
Assess online shopping experience as convenience 
.79 
Reliability 
0.89 
KMO 
.838 
Variance Explained 
65.60 
Bartlett’s 
.000 
 
Step Two – Confirmatory 
CONV1 
CONV2 
CONV3 CONV4 CONV5 CONV6 
CONV1 
1.00 
CONV2 
.725 1.00
CONV3 
.624 .773 1.00
CONV4 
.484 .529 .665 1.00
CONV5 
.440 .457 .554 .693 1.00
CONV6 
.483 .561 .587 .575 .646 1.00 
Indicators CFA 
 
Stand. Loadings 
Stand. 
Error 
Critical 
Ratio
AVE CR 
CONV1 .68 
-- 
-- 
.59 
.878 
CONV2 .85 
.045 
27.97
*
CONV3 .91 
.059 
22.36
*
CONV4 .73 
.058 
19.78
*
CONV5 n/a 
n/a 
n/a 
CONV6 .65 
.052 
18.68
*
Model Fit 
Statistic 
Model Fit 
Statistic 
χ
2
1.565 TLI .998 
df
1 
CFI 
.998 
P .211 RMSEA 
.024 
GFI .999 
RMR .010 
* 
p 
< .001
 
5.5.8 Preliminary Analysis – Privacy and Security Concerns 
Five items (PSC1 to PSC5) were used to measure the privacy and security concerns construct. 
All of the items were included in the two-step preliminary data analysis process shown in 
Figure 5.1, the results of which are displayed in Table 5.9. Evaluation of the correlation 
matrix through the KMO and Bartlett’s Test results in a high KMO statistic (.865) and a 
significant probability level (p< .001) for the Bartlett’s test. These results indicate that 
sufficient correlations were found within the correlation matrix for factor analysis to proceed. 

---

230
In addition, bivariate correlations were inspected and all coefficients fell within the 
acceptable range for factor analysis of .30 and .90. EFA was then conducted which produced 
a single factor structure with strong item loading ranging from .85 to .93 and the variance 
explained was 79%. Cronbach’s alpha of .93 was computed indicating good reliability of the 
scale. At this point, as all the items met the criteria of Step One, they were retained for CFA 
analysis in Step Two. 
In Step Two (refer Figure 5.1), CFA was conducted on the five items retained from Step One. 
Statistics indicating a good fit of the measurement model included 
χ
2
/df (1.417) which was 
less than 5, GFI
(.968
), TLI (.975) and CFI (.971), which were above .90. RMSEA (.026) was 
less than .08 and RMR (.005) was less than .05. In addition, as shown in Table 5.9, the 
standardized factor loadings of the five items ranged from .73 to .96, all exceeding the 
preferable criterion of .70 (Hair
 et al.
2006). The AVE was .71, exceeding the recommended 
criterion of .50 (Hair
 et al.
2006). The CR was .922, exceeding the recommended benchmark 
of .70 (Nunnally 1978). Thus, items PSC1 to PSC5 were considered to provide a reliable and 
valid measurement scale for the construct, privacy and security concerns. All five items were 
retained for constructing the overall measurement model. 

---

231