Understanding consumer online shopping behaviour from the perspective of transaction costs

Model Fit 
Statistic 
Model Fit 
Statistic 
χ
2
45.844 
TLI 
.993 
df
21 
CFI 
.996 
P .001 
RMSEA 
.035 
GFI .989 
RMR 
.044 
* 
p 
< .001
REL1 REL2 REL3 REL4 REL5 RESP1 RESP2 RESP3 RESP4 PERS1 PERS2 PERS3 
REL1 
1.00 
REL2 
.784 
1.00 
REL3 
.681 
.805 
1.00 
REL4 .629 .723 .811 1.00
REL5 .596 .629 .667 .749 1.00
RESP1 .552 .539 .539 .506 .549 1.00
RESP2 .560 .549 .534 .503 .534 .754 1.00
RESP3 .534 .503 .508 .475 .516 .611 .714 1.00
RESP4 .639 .575 .530 .519 .540 .610 .674 .714 1.00
PERS1 .528 .538 .527 .524 .539 .489 .469 .449 .495 1.00
PERS2 .484 .468 .471 .460 .498 .489 .444 .450 .478 .744 1.00
PERS3 .497 .481 .486 .482 .519 .495 .448 .420 .455 .684 .797 1.00 

---

226
5.5.6 Preliminary Analysis – Reputation of Online Store 
Four items (REP1 to REP4) were used to measure the reputation of online store construct. 
The two-step preliminary data analysis process shown in Figure 5.1 was employed as a 
screening mechanism for data measured by these items (refer Table 5.7 for results). 
Evaluation of the correlation matrix through the KMO and Bartlett’s Test results in a high 
KMO statistic (.810) 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. 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 .90 and the 
variance explained was 76.41%. Cronbach’s alpha of .90 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 four items retained from Step One 
which resulted in all computed statistics indicating a good fit for the measurement model. For 
example, 
χ
2
/df (0.037) was less than 5, GFI (0.975), TLI (0.981) and CFI (0.972) were 
above .90, RMSEA (.002) was less than .08 and RMR (.003) was less than .05. Therefore, the 
CFA results indicated that this measurement model provided a good fit to the data.
Furthermore, as displayed in Table 5.7, the standardized factor loadings of the four items 
ranged from .72 to .92, all exceeding the desirable level of .70 (Hair
 et al.
2006). The AVE 
was .66, exceeding the recommended criterion of .50 (Hair
 et al.
2006). The CR was .887, 
exceeding the recommended benchmark of .70 (Nunnally 1978). As a result, items REP1 to 
REP4 were considered to provide a reliable and valid measurement scale for this construct, 
and all four items were retained for the overall measurement model. 

---

227