Microsoft Word 2007 ichrie conference Proceedings Final-Final 06-06-07. doc

Structural Model Analysis 
Structural equation modeling was used to assess the proposed conceptual model, following the maximum 
likelihood estimation method. The overall model fit was satisfactory (
χ
2
= 41.830, 
df
= 23, 
p 
< .01, 
χ
2
/
df
= 1.819).
Other goodness-of-fit indices also revealed that the measurement model fit the data well (RMSEA = 0.052; CFI = 
0.994; NFI = 0.986). OSQ explained approximately 63% of the variance in CS. CS explained about 71% of RVI 
and 54% of WOMI, indicating the model could predict and explain customers’ BI fairly well.
 
Hypothesis 1 was tested. As hypothesized, the relationship between OSQ and CS was significant (
γ
11 
= .79, 
t = 16.65, p < .01), thus supporting Hypothesis 1. This result was consistent with the previous findings that service 
quality is a significant predictor of CS (Cronin and Taylor, 1992; Ting, 2004). That is, once customers perceive that 
a restaurant provides excellent service, their satisfaction level increases. In addition, the results of the Sobel test 
showed that OSQ significantly affected RVI and WOMI through CS (
β
OSQ-CS-RVI
= .748, t = 11.77, p < .01; 
β
OSQ-CS-
WOMI
= .745, t = 8.57, p < .01). Thus, it can be concluded that CS has a significant mediating role in the relationship 
between OSQ and RVI and in the relationship between OSQ and WOMI. A restaurant firm should enhance CS 
level as well as service quality to maximize the impact of service quality on increasing customers’ BI. 
Hypotheses 2 and 3 were tested to investigate the causal relationship from CS to RVI/WOMI. The 
regression paths from CS to RVI (
β
21
= .84, t = 18.75, p < .01) and to WOMI (
β
31 
= .73, t = 13.41, p < .01) were 
significant, supporting Hypotheses 2 and 3. These findings imply that increasing CS level contributes to enhancing 
customers’ likelihood of returning to the restaurant and their willingness to be engaged in WOM behaviors, such as 
recommending the restaurant and providing positive WOM to family, friends, or others. The results were consistent 
with previous findings that increasing CS level plays a significant role in forming customers’ positive BI (Oliver, 
1997). Additionally, the beta linking CS and RVI (
β
= .84) was relatively greater than the casual path linking CS 
and WOMI (
β
= .73). This indicated that the influence of CS on RVI is stronger than its influence on WOMI.