Understanding consumer online shopping behaviour from the perspective of transaction costs

5.8 TESTING FOR MEDIATION 
H6, posited that customer satisfaction partially mediates the effect of consumer TCs on
 
loyalty was tested in this section. The mediation testing was undertaken in terms of the 
guideline suggested by Baron and Kenny’s (1986). To conduct such a test, three conditions 
must be met. First of all, it is important to determine if the independent variable (consumer 
TCs) significantly affects the dependent variable (customer loyalty). Secondly, the 
independent variable (consumer TCs) must significantly impact the mediator (customer 
satisfaction). The third condition is that the mediator (customer satisfaction) must 
significantly impact the dependent variable (customer loyalty).
To test these above three conditions and further examine the mediating role of customer 
satisfaction between consumer TCs and customer loyalty, Model 1 and Model 2 were 
developed. As shown in Table 5.24, consumer TCs negatively and significantly influenced 
customer loyalty (Model 1, 
β
= -.724, t = -32.552) and customer satisfaction (Model 2, 
β
= -
.749, t = -35.036), which also positively influenced customer loyalty (Model 2, 
β
= 0.647, t = 
24.610), thus the three conditions were achieved. Comparing Model 1 and 2, it was found 
that the negative effect of consumer TCs on customer loyalty in Model 1 (
β
= -.724) had 
simply been reduced (
β
= -.240) in Model 2, but it was still significant (t = -9.110) in Model 2. 
The reason for the reduction in the path coefficient is due to the mediation variable (customer 

---

269
satisfaction) explaining some of the variance in the dependent variable (customer loyalty) that 
was previously explained by consumer TCs, but it was more appropriate to be explained 
through the mediator (customer satisfaction). Therefore, customer satisfaction partially 
mediates the relationship between consumer TCs and loyalty, providing support for H6.