Understanding consumer online shopping behaviour from the perspective of transaction costs

Table 5.31 Comparison of Search Products and Experience Products Model 
Paths Search 
Products 
n=455 
Experience Products 
n=507 
Comparison of Models 
H1a Not 
significant Significant 
Different 
H1b Significant 
Significant 
Same 
H1c Significant Significant 
Same 
H2a Not 
significant Significant 
Different 
H2c Significant Significant 
Same 
H2d Significant Not 
significant Different 
H3a Significant Significant 
Same 
H3b Not 
significant 
Significant 
Different 
H4a Significant Significant 
Same 
H4b Significant 
Significant 
Same 
H4c Significant Significant 
Same 
H5 Significant Significant Same 

---

284
The comparison of the two models showed the significances of the corresponding paths were 
the same for all paths except for four paths. For example, the path from perceived Internet 
expertise to consumer TCs across the two models both showed significant path coefficients. 
Also, the effect of perceived convenience on consumer TCs in both models was significant. 
On the other hand, with regard to the differences in significant paths, the effects of Internet 
access availability (H1a), product quality concern (H2a ), and privacy and security concerns
(H3b) on consumer TCs were not significant within the search products model, whereas they 
were significant within the experience products model. In addition, the comparisons made in 
Table 5.31 also indicated that the relationship between reputation of online store and TCs was 
significant within search products model, while it became non-significant within the 
experience products model. To summarize, H9a was partially supported in regards to the 
differences in the significance of the four paths across two models, which included the paths 
from Internet access availability, product quality concerns privacy and security concerns and 
reputation of online store to consumer TCs. For the rest of the hypotheses concerning the 
significance of the effects of antecedents on consumer TCs in the structural model, they were 
still the same across two models.
As for the different effects of consumer TCs on behavioural consequences across product 
categories (H9b), the comparisons in Table 5.31 showed that there were all significant 
between search products and experience products, therefore indicating that there are no 
differences in the significance of the paths from consumer TCs to behavioural consequences 
(i.e., online purchase behaviour and customer loyalty) across two models. H9b was rejected 
in regards to the significances of the corresponding paths. 

---

285
While H9a was partially supported and H9b was rejected by the results in Table 5.30, the 
individual search products model and experience products model were further compared to 
determine if there were differences in the 
strengths
of the paths within the structural model. 
Analysis was conducted to examine differences in paths across the two models using 
t
-tests 
procedure which was recommended by Keil et al. (2000). The following formula, as 
suggested by Keil et al. (2000) was used to calculate the t- statistics shown in Table 5.31. 
The pooled estimate for the variance was firstly calculated as such: 
Sp=
∗ . .
1
∗ . .
2
Where m = sample size 1 
Where n = sample size 2 
Then: t value =
∗ 
Where p1= Highest path coefficient 
Where p2 = lowers path coefficient
The results of the analysis, conducted via one-tailed t-tests, are shown in Table 5.32. Tests 
performing on the corresponding path beta weights of both models that resulted in t-values 
greater than 1.64 were deemed to be significantly different. 
Table 5.32 indicated that t-tests conducted on all paths H1a to H5 all showed significant 
differences in the strengths, supporting H9a and H9b from the perspective of strength of the 
paths. For search products model, path coefficients were significantly stronger for H1c 
(online buying frequency 

TCs), H2c (e-service quality 

TCs), H2d (reputation of online 
store 

TCs) and H5 (customer satisfaction 

loyalty). The stronger effect of reputation of 
online store on TCs (H2d) in search products model was actually found to be statistically 

---

286
supported in the preceding analysis whereby online store reputation was found to be 
significant in the search products model (refer Table 5.29) but not significant in the 
experience products model (refer Table 5.30). On the other hand, for the experience products 
model, H1a (Internet access availability 

TCs), H1b (perceived Internet expertise 

TCs), 
H2a (product quality concern 

TCs), H3a (perceived convenience 

TCs), H3b (privacy 
and security concerns 

TCs), H4a (TCs 

online purchase behaviour), H4b (TCs 

customer loyalty) and H4c (TCs 

customer satisfaction) were significantly stronger than the 
corresponding paths in the search products model. The stronger effects of H1b, H1d and H1e 
in experience products model are consistent with the results predicted and discussed in the 
previous analysis whereby these three hypotheses were proved to be significant in the 
experience products model (refer Table 5.30) but not significant in the search products model 
(refer Table 5.29). Thus, path coefficients exhibit differences in their strength within the 
individual models across product categories, supporting H9a and H9b in terms of the 
magnitude of the aforementioned relationships.