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

Table 2 
Invariance Tests of Hypothesized Paths 
Fit of the Model with the Path 
Test of Invariance 
Paths 
Baseline Model 
(Freely Estimated) 
Nested Model 
(Constrained to be Equal) 
Chi-square Difference Test 
OSQ
→
CS 
χ
2 
(51) = 92.927 
χ
2 
(52) = 95.032 
Δχ
2 
(1) = 2.105, p > .05 
CS
→
RVI 
χ
2 
(51) = 92.927 
χ
2 
(52) = 97.695 
Δχ
2 
(1) = 4.778, p < .05 
Gender 
CS
→
WOMI 
χ
2 
(51) = 92.927 
χ
2 
(52) = 95.968 
Δχ
2 
(1) = 3.051, p > .05 
OSQ
→
CS 
χ
2 
(51) = 62.587 
χ
2 
(52) = 68.037 
Δχ
2 
(1) = 5.450, p < .05 
CS
→
RVI 
χ
2 
(51) = 62.587 
χ
2 
(52) = 63.645 
Δχ
2 
(1) = 1.058, p > .05 
Age 
CS
→
WOMI 
χ
2 
(51) = 62.587 
χ
2 
(52) = 67.823 
Δχ
2 
(1) = 5.236, p < .05 
Note. OSQ: Overall Service Quality, CS: Customer Satisfaction, RVI: Repeat Visit Intention, WOMI: Word-of-mouth Intention. 
Figure 1 
Results for Male and Female Groups––Standardized Path Estimates (t-values in Parentheses) 
 
 
 
 
 
 
 
 
 
 
 
 
 
Note. OSQ: Overall Service Quality, CS: Customer Satisfaction, RVI: Repeat Visit Intention, WOM: Word-of-mouth Intention. Male group on 
top (
n
=155), female group below (
n
=146); *
p
< .01 
Figure 2 
Results for High Age and Low Age Groups––Standardized Path Estimates (t-values in Parentheses) 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
OSQ
WOMI 
CS
RVI 
.84* 
.76* 
.87* 
.82* 
.67* 
.80* 
OSQ
WOMI 
CS
RVI 
.72* 
.81* 
.79* 
.89* 
.85* 
.58* (6.21)

---

 
2007 Annual International CHRIE Conference & Exposition 
397
Note. OSQ: Overall Service Quality, CS: Customer Satisfaction, RVI: Repeat Visit Intention, WOMI: Word-of-mouth Intention. High age group 
on top (
n
=150), Low age group below (
n
=148); *
p
< .01 
Invariance tests for key paths.
 
The findings for the invariance tests of key paths are presented in Table 2.
The standardized path estimates with t-values for gender and age groups are shown in Figures 1 and 2, respectively.
First, invariance in the path coefficient from CS to RVI was assessed by constraining the CS-RVI path to be equal 
across gender groups. A significant chi-square difference was found (
Δχ
2 
= 4.778, 
Δ
df
= 1, p < .5), revealing gender 
to have a significant moderating role in the relationship between CS and RVI. As shown in Figure 1, the coefficient 
values in the CS-RVI relationship in both gender groups were significant (male: 
β
21
= .82, t = 12.13, p < .01; female: 
β
21
= .87, t = 14.64, p < .01). As expected, the coefficient value of female was higher than the value of male, thus 
supporting Hypothesis 4. This finding implies that once female customers are fully satisfied, they are more likely to 
return to the restaurant than male customers. Next, the equality of the parameter estimate from CS to WOMI across 
male and female groups was tested. As shown in Table 2, an insignificant chi-square difference was detected (
Δχ
2 
= 
3.051, 
Δ
df
= 1, p > .5). Thus, Hypothesis 5 was not supported.
For the CS-RVI relationship across high/low age groups, the coefficient values were significant (high 
group: 
β
21
= .89, t = 17.26, p < .01; low group: 
β
21
= .79, t = 10.82, p < .01). This demonstrates that the customers’ 
(both high and low age groups) decision to revisit a specific restaurant is strongly determined by their satisfaction.
However, the results of the invariance test revealed that the relationship between CS and RVI for high and low age 
groups was not significantly different. As shown in Table 2, there was no significant chi-square difference across 
two groups (
Δχ
2 
= 1.058, 
Δ
df
= 1, p > .5). Thus, Hypothesis 6 was not supported. This result indicated that age, as a 
moderator variable, does not have a significant effect on the linkage between CS and RVI. The analysis of a 
moderator effect of age on the link between CS and WOMI revealed that the relationship was stronger for the high-
age group (
β
31
= .85, t = 11.11, p < .01) than for the low-age group (
β
31
= .58, t = 6.21, p < .01). A significant chi-
square difference was found (
Δχ
2 
= 5.236, 
Δ
df
= 1, p < .5). Hence, Hypothesis 7 was supported. 
Further, the invariance in the effect of OSQ on CS across male and female groups was assessed. The path 
from OSQ to CS was constrained to be equal for the two groups. The chi-square test revealed that the chi-square 
difference is 2.105 with one degree of freedom, which was not significant at the .05 level. Thus, Hypothesis 8 was 
not supported. This finding indicated that gender does not have a moderating effect on the relationship between 
OSQ and CS. Finally, the invariance in the effect of OSQ on CS across high and low age groups was assessed. The 
OSQ-CS path was constrained to be equal across age groups. A significant chi-square difference was found at the 
.05 level––5.450 with one degree of freedom. The estimates for path coefficients were compared between groups.
For the path from OSQ to CS, the estimate was higher in the high-age group (
γ
11 
= .81, t = 11.85, p < .01) than in the 
low-age group (
γ
11 
= .72, t = 9.55, p < .01) (see Figure 2). Thus, Hypothesis 9 was supported.