Tourists’ Experiences with Smart Tourism Technology at Smart Destinations and Their Behavior Intentions

Table 4.
Correlation Matrix and Discriminant Assessment.
1
2
3
4
5
6
7
1. Accessibility
0.916
2. Informativeness
0.774
0.900
3. Interactivity
0.632
0.708
0.888
4. Personalization
0.619
0.694
0.796
0.892
5. Memorable experience
0.540
0.621
0.696
0.677
0.872
6. Satisfaction
0.549
0.610
0.717
0.700
0.797
0.889
7. Behavior Intention
0.471
0.563
0.543
0.550
0.600
0.633
0.927
Note. Goodness-of-fit statistics for the measurement model: 
χ
²=853.9, df 
=
186, p 
=
0.00, 
χ
²/df 
=
4.591, NFI 
=
0.955, CFI 
=
0.965 IFI 
=
0.965, TLI 
=
0.956, RMSEA 
=
0.060.

---

Jeong and Shin 
9
insignificant, leading to rejecting hypothesis 1a. Memorable 
experience exerted significant positive effects on both satis-
faction and behavioral intention and satisfaction had a sig-
nificant positive relationship with behavioral intention, 
supporting hypotheses 3 to 5, respectively. Squared multiple 
correlations indicated that the measurement items extracted 
70% variance of tourists’ memorable experience, 86% of 
their satisfaction, and 51% of their behavior intentions. All 
the 
R
2
of the endogenous constructs in the model exceeded 
the 10% benchmark recommended by Falk and Miller 
(1992).
To examine moderating effects of tourists’ perceptions of 
security/privacy when using STTs at the destination, this 
study conducted a series of group comparisons by dividing 
the data set into two groups: low (less concerned about secu-
rity and privacy) and high (highly concerned about security 
and privacy). The data grouping was based on the median of 
the security/privacy construct (N
low
= 511 vs. N
high
= 499) 
(DeCoster, Iselin, and Gallucci 2009; Ro 2012). A proposed 
model with free parameters was compared to a constrained 
model to evaluate whether the model fit and parameter esti-
mates changed significantly between the two models (Kline 
1998). The change was assessed by the ratio of the 
χ
2
differ-
ence to the difference in the degree of freedom between the 
low and high groups. The proposed relaxed model was con-
sidered superior in its explanatory power to the constraint 
model when it showed a smaller value of 
χ
² than 3.84 (
χ
²
0
.
05,1
=
3.84) for one less degree of freedom. Compared with the 
proposed model (
χ
² 
=
1412.38, df=392, p
<
.000), 
χ
2
differ-
ence tests resulted in significant differences in three paths 
from informativeness to memorable experience (
Δχ
² 
=
8.01), from interactivity to memorable experience (
Δχ
² 
=
4.73), and from personalization to memorable experience 
(
Δχ
² 
=
4.69) (see Table 6).
Additionally, to identify whether different sociodemo-
graphic groups (i.e., gender, residence, and STT familiarity) 
of respondents had different relationships for the path from 
STT attributes and their memorable experience, this study 
conducted multigroup moderation tests by generating the 
critical ratios for the differences in regression weights 
between the two groups (V. Singh and Sharma 2016). Only 
STT familiarity (N
low
= 530, N
high
= 480) showed significantly 
different perceptions in informativeness and personalization. 
Consequently, a group comparison of STT familiarity was 
conducted by dividing the data into two groups by using the 
median of the total number of STT used by the respondents 
(DeCoster, Iselin, and Gallucci 2009; Ro 2012): low (low 
familiarity with STTs) and high (high familiarity with STTs).