|
n
⫽
200, 64.3 per cent) and aged between 26-35 years. Also, the
majority of the participants (
n
⫽
193, 62.1 per cent) were traveling for the first time to Korea.
Finally, most of the participants stayed in Korea for four to six days (
n
⫽
223, 71.7 per cent).
Measures
The survey instruments were extracted from the literature review. The destination image
scale consists of two dimensions: seven items for cognitive image (
Prayag and Ryan, 2012
)
Table I
Sample demographic characteristics
Characteristics
Frequency
(%)
Gender
Male
111
35.7
Female
200
64.3
Age (years)
16-25
67
21.5
26-35
156
50.1
35-45
57
18.3
⬎
45
31
10.1
Income (RMB)
⬍
5,000
137
44.1
5,001-10,000
86
27.6
⬎
10,000
88
28.3
Past experience
First-time visit
193
62.1
Repeated visit
118
37.9
Length of stay (days)
1-3
30
9.6
4-6
223
71.7
⬎
6
58
18.7
PAGE 226
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
VOL. 10 NO. 2 2016
Downloaded by 80.82.77.83 At 02:42 07 May 2017 (PT)
and four items for affective image (
Pike and Ryan, 2004
). Because the two dimensions of
destination image were separately examined in the previous studies (
San Martín and del
Bosque, 2008
), the scales of cognitive and affective images were adapted from different
studies. Moreover, these scales were used and examined by many studies which revealed
good reliability and validity. Thus, the use of these scales was considered appropriate in
this study. Tourist satisfaction (three items) and loyalty (three items) were taken from
Lee
et al.
’s (2011)
study. All scale items were evaluated using a five-point Likert scale ranging
from 1 (
strongly disagree
) to 5 (
strongly agree
). These scales were selected because they
reported adequate reliability and validity in the previous studies. Moreover, some items of
the scale were slightly modified to make it suitable in the context of Chinese tourists visiting
Korea.
As a next step, the original English version of the questionnaire was translated into a
Chinese version based on
Brislin’s (1970)
guidelines. First, two experts individually
translated the original items to Chinese. After comparing and adjusting any discrepancy
between these two drafts, a final draft of the Chinese version was completed. Next, the
other two experts independently back-translated the Chinese version into two English
versions, and then, the researchers compared the two back-translated English versions
with the original English questionnaire to ensure that the meaning was conceptually
equivalent.
In addition, 20 Chinese travelers were invited to pilot test the Chinese version of the
questionnaire. They evaluated the clarity and appropriateness of all items to ensure the
content validity of the Chinese version questionnaire. Consequently, only minor changes
were made in the final questionnaire.
Data analysis
The current study used the SPSS 18.0 statistical package for a descriptive analysis of the
sample structure. AMOS 18.0 was used for structural equation modeling (SEM). Maximum
likelihood estimation was evaluated in the SEM analysis software. Data analysis proceeded
in a two-stage analysis based on
Anderson and Gerbing’s (1988)
study. First, we measured
the reliability and validity of the observed and latent variables to estimate the obvious levels
of these variables using a confirmatory factor analysis (CFA). Second, SEM analysis was
used to test the hypotheses in the proposed conceptual model. With the structural model
analysis, this study uses the standard factor loading and
t
-value of the path coefficient to
determine the path strengths and significance levels of the latent variables.
Results
Reliability and validity of measurement
The CFA results revealed that the measurement model fulfilled the criteria suggested by
Hu
and Bentler (1999)
and yielded an acceptable level of model fit,
2
(113)
⫽
293.659,
confirmatory fit index (CFI)
⫽
0.951, Tucker–Lewis index (TLI)
⫽
0.941 and root mean
square error of approximation (RMSEA)
⫽
0.072.
Next, the reliability and validity of the scales were evaluated. First, the reliability of the
measures was investigated using Cronbach’s
␣
and composite reliability (CR) analysis. As
shown in
Table II
, all the values of Cronbach’s
␣
exceeded 0.80, which exceeded the
recommended level (0.70) proposed by
Nunnally (1978)
. The CR values ranged from 0.876
to 0.912, exceeding the criterion (0.70) suggested by
Fornell and Larcker (1981)
. Second,
the validity of measures was examined using convergent validity and discriminant validity.
Convergent validity was investigated by the values of factor loading and the average
variance extracted (AVE). All the factor loadings of the construct indicators were higher
than 0.50, and all the AVE values were greater than the suggested value (0.50) (
Hair
et al.
,
2010
). Discriminant validity was also evaluated, and the results revealed that the values of
AVE exceed both maximum shared squared variance and average share squared
VOL. 10 NO. 2 2016
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
PAGE 227
Downloaded by 80.82.77.83 At 02:42 07 May 2017 (PT)
variance, and the squared AVEs were greater than the correlation estimates in the model
(
Tables II
and
III
).
Hypothesis testing
According to the SEM results, the conceptual model revealed an adequate level of fit,
2
(113)
⫽
293.659, CFI
⫽
0.951, TLI
⫽
0.941 and RMSEA
⫽
0.072 (
Hu and Bentler, 1999
).
As illustrated in
Figure 1
, the path from cognitive image to affective image was significant
(

⫽
0.43,
t
⫽
6.81); therefore,
H1
was supported. The path from cognitive image to
satisfaction was significant (

⫽
0.59,
t
⫽
8.72), supporting
H2
. The path from affective
image to satisfaction was significant (

⫽
0.20,
t
⫽
3.59), and thus,
H3
was supported. The
path from cognitive image to tourist loyalty was not significant (

⫽
0.09,
t
⫽
1.55), while the
path from affective image to tourist loyalty was significant (

⫽
0.14,
t
⫽
2.87); therefore,
H4
was rejected, and
H5
was supported. Finally, the path from satisfaction to tourist loyalty was
significant (

⫽
0.63,
t
⫽
2.86), supporting
H6
.
Exploration of mediating effect
Furthermore, the exploration of mediating effects was examined based on the paths in the
conceptual model. Although a typical three-step procedure suggested by
Baron and
Kenny (1986)
provides useful insights for examining a mediating effect, the recent research
on mediation reported that three steps are not necessary and provided a more practical
Do'stlaringiz bilan baham: |
