Results of Construction and Evaluation of the Research Scale

The relationship between the specific components is as follows: Latent independent variables (AT attitude; SN subjective norm; PBC perceived behavioral control; Destination image of access to resources and resources TCTN; Destination image of quality and reputation CLDT; Destination image of overall HATT); Latent dependent variables (SAT satisfaction; Intention to revisit INT). The results of the linear structural measurement model for each latent variable are determined by the corresponding observed variables as shown below (see Figure 4.3).

Figure 4.3. SEM Linear Model of Destination Revisit Intention

Source: Official survey results of 443 tourists in 3 provinces, April 2021

The results of testing the SEM structural model of the research model of factors affecting the intention to return to the destination showed that Chi-square = 824.497; df = 412; P = 0.000; Chi-square/df = 2.001; GFI = 0.896; CFI = 0.939; RMSEA = 0.048.

This result shows that the theoretical model fits the actual data well.

Table 4.16. Regression Weights‌

Relationship

Unstandardized regression coefficients

Standard error

Critical value

Price

value (P)

Conclusion Hypothesis

discussion

SAT

<---

TCTN

SAT

<---

CLDT

SAT

<---

HATT

Maybe you are interested!

(Estimate)	(SE)	(CR)
,487	,071	6,897	*** H 5 Acceptance
,481	,086	5,588	*** H 6 Acceptance
-,032	,063	-,506	,613 H 7 Reject
,089	,052	1,728	,084 H 1 Reject
,209	,059	3,570	*** H 3 Acceptance
,320	,044	7,284	*** H 4 Acceptance
,257	,061	4,228	*** H 2 Acceptance

accept accept

INT <--- AT

INT <--- PBC

INT <--- SAT INT <--- SN

receive receive

Note: AMOS symbol *** is Sig equal to 0.000

Source: Official survey results of 443 tourists in 3 provinces, April 2021

The estimation results show that, using the 95% confidence standard, we see that the Sig of HATT affecting SAT is 0.613 > 0.05 and the Sig of AT affecting INT is 0.084 > 0.05. Therefore, the results show that the HATT factor has no impact on SAT and the AT factor has no impact on INT. Most of the remaining factors have Sig < 0.05, so these relationships are all significant. In conclusion, there are 2 factors affecting SAT including TCTN and CLDT; there are 3 factors affecting INT including PBC, SN and SAT. Of the 7 research hypotheses, the author rejects 2 hypotheses: H 1 and H 7 ; and accepts the remaining hypotheses (H 2, H 3, H 4, H 5, and H 6 ). (see Table 4.16).

Next, the author relies on the regression coefficient Estimate (standardized) to assess the impact of the independent variables on the dependent variable. In the 2 variables affecting SAT, the order of the variables affecting from strong to decreasing is as follows: TCTN, CLDT. The variables affecting INT are also in order of strong to decreasing specifically: SAT, SN and PBC. (see Table 4.17).

Table 4.17. Standardized regression estimation results (SRWeights)‌

Relationship Standardized Regression Coefficients (Estimate)

SAT <--- TCTN ,372

SAT <--- CLDT

,309
SAT <--- HEAT	-,024
INT <--- AT	,086
INT <--- PBC	,181
INT <--- SAT	,352
INT <--- SN	,216

Source: Official survey results of 443 tourists in 3 provinces, April 2021

To evaluate the significance of the research model results, the author uses the R-square value, which ranges from 0 to 1 (the higher the value, the more accurate the model's prediction). Accepting the R-square value is also relatively difficult, because it depends on the complexity of the model and the research context. Normally, the quality of the research model is determined by the coefficient of determination R-square with values of 0.26; 0.13 and 0.02 (corresponding to large, medium and small levels). (according to Cohen, 1988; Hanh et al., 2019; Nam, 2020).

The calculation results show that the R-square values for the dependent variables are all higher than the threshold of 0.26. Thus, it can be concluded that the level of explanation of the research variables in the model is good. This result also shows that the research model of factors affecting tourists' intention to return to the destination is statistically significant, of good quality and appropriate (see Table 4.18).

Table 4.18. R-square (Squared Multiple Correlations) Value Information Table

Observed variable Estimated coefficient R- Predictive level

square (Estimate)

SAT ,317 High

Predictive meaning

Significant

INT

,313

High

Significant

Source: Official survey results of 443 tourists in 3 provinces, April 2021

4.2.6 Bootstrap analysis

Bootstrap testing was conducted to re-examine the model when analyzing SEM linear structural model data. The bootstrap analysis results showed that the absolute value of CRa was very small compared to 1.96, which means that P-value > 5%, rejecting H a and accepting H 0 , it can be said that this bias is very small, not statistically significant at the 95% confidence level. Therefore, we can confirm that the estimates in the original model are reliable. Researchers believe that this is also the expected result when analyzing the SEM linear structural model. (see Table 4.19).

Table 4.19. Bootstrap estimation results with AMOS (N= 1,500)‌

SE- SE relationship

Mean Bias SE- Bias

SAT <--- TCTN ,053 ,001 ,372 -,001 ,001 -1

SAT <--- CLDT

,055	,001	,309	,000	,001 0
SAT <--- HEAT	,048	,001	-,022	,001	,001 1
INT <--- AT	,057	,001	,085	-,001	,001 -1
INT <--- PBC	,052	,001	,183	,002	,001 2
INT <--- SAT	,051	,001	,349	-,003	,001 -3
INT <--- SN	,049	,001	,216	,000	,001 0

Source: Official survey results of 443 tourists in 3 provinces, April 2021

4.2.7 Multigroup analysis

Multi-group structural analysis aims to evaluate whether the research model is different between different subjects or not, more specifically, the author will test two variable and invariant models. Based on the primary data set collected from 443 Vietnamese tourists, the thesis has identified some basic information about the demographic characteristics of the surveyed subjects, such as: gender, age, expertise, occupation, survey area, trip purpose... With such a diverse range of information, to evaluate the difference in the impact relationships in the SEM model whether the above-mentioned information characteristics are different or not, it is necessary to select which important characteristics can be used to analyze the multi-group structure. According to the author's assessment, the information characteristics about the survey areas (Ca Mau, Bac Lieu and Soc Trang) can be an important characteristic for the author to conduct structural analysis.

Multi-group analysis for variable and invariant models of survey site characteristics to assess whether the differences between the surveyed tourist groups in Ca Mau, the tourist groups in Bac Lieu and the tourist groups in Soc Trang affect the components in the research model differently or not.

The results of testing the variable model and the invariant model according to the characteristics of the survey area determined by the corresponding observed variables show that the theoretical model fits the actual data (see Appendix 22). The test of model fit and the results of the Chi-Square test show that the characteristics of the survey area of tourists do not differ between MHBB and MHKB (P-value = 0.6 > 0.05), so MHBB is selected. Thus, we can confirm that the relationships in this research model are not affected by the differences according to the survey area between tourists in Ca Mau, tourists in Bac Lieu and tourists in Soc Trang. (see Table 4.20).

Table 4.20. Test results of differences by tourist gender

MHKB MHBB Difference Criteria (MHBB –

MHKB)

Chi-square

df Chi-square

df Options

Survey area 1788,500 1236 1800,434 1250 11,934 14 MHBB

Source: Official survey results of 443 tourists in 3 provinces, April 2021

The results of multigroup analysis showed that the relationships in this research model were not affected by differences in survey locations between tourists in different locations.

4.3 SUMMARY OF RESEARCH RESULTS

The results of testing the theoretical model also show that the data fits well with the survey information. The results of analyzing the linear structural model of the theoretical model with 7 initial hypotheses, 5 hypotheses are accepted and 2 hypotheses are not accepted. Specifically, the results of testing the hypotheses are as follows:

4.3.1 Results of building and evaluating the research scale

Based on the proposed research model, the author continued to conduct qualitative research, and the results formed a scale to study factors affecting tourists' intention to return to tourist destinations in 3 provinces of Ca Mau, Bac Lieu and Soc Trang (Vietnam). The scale consists of 35 observed variables, corresponding to 8 components.

This part (factor or attribute) was used to conduct preliminary quantitative research with 118 domestic tourist samples. The analysis results through assessing the reliability of the scale and the scale value showed that 32/35 observed variables met the requirements for the official quantitative research step (see Table 4.21).

Table 4.21. Summary of scales used in formal quantitative research

Stt Thang

measure

Scale name

Number of variables (quantitative)

Number of variables (achieved)

request)

Note

preliminary)

AT	Attitude	4	4
2	SN	Subjective standard	5	5

3 PBC Behavior Control

5 4 PBC5 variable type

awareness

HADD on access

4 TCTN resources and 4 4

resources

Quality improvement

5 CLDT quantity and quality 5 5

language

6 HATT HADD about total 4 4

body

7 SAT Satisfaction 4 3 Type of variable SAT1

8 INT Intention to return

4 3 Variable type INT2

destination

Total 35 32 Adjust

order of variables

* The scale with 32 observed variables was used in the official quantitative study. Source: Preliminary survey results of 118 tourists in 3 provinces, February 2021

4.3.2 Results of testing research hypotheses

4.3.2.1 Results of testing hypothesis H 1

With hypothesis H 1: Attitude has a positive influence on tourists' intention to return to the destination. The author concludes: the hypothesis is not accepted (rejected).

The estimation results show that, using the 95% confidence standard, we see that the Sig of AT impact on INT is 0.084 > 0.05, then the hypothesis that AT attitude has a positive impact on intention to return to the destination INT will not be accepted (rejected). We can conclude that

The relationship between tourist attitudes and their intention to revisit the destination was not statistically significant (standardized regression coefficient was 0.086).

4.3.2.2 Results of testing hypothesis H 2

Similar to hypothesis H 2 : Subjective norm has a positive influence on tourists' intention to revisit the destination. Conclusion: accept the hypothesis.

The estimation results show that, using the 95% confidence standard, we see that the Sig of SN affecting INT is 0.000 < 0.05, so the hypothesis that subjective norm SN has a positive effect on tourists' intention to return to the destination will be accepted.

With a standardized regression coefficient of 0.216, when tourists evaluate the factor “subjective norm for a destination” increasing by 1 point, the impact of this factor on the factor “intention to return” increases by 0.216 points. We can conclude that tourists’ subjective norm has a positive impact on their intention to return to the destination.

4.3.2.3 Results of testing hypothesis H 3

With hypothesis H 3 : Perceived behavioral control has a positive influence on tourists' intention to revisit the destination. Conclusion: accept the hypothesis.

The estimation results show that, using the 95% confidence standard, we see that the Sig of PBC affecting INT is 0.000 < 0.05, then the hypothesis that perceived behavioral control has a positive effect on tourists' intention to return to the destination will be accepted.

With a standardized regression coefficient of 0.181, when tourists rate the factor “perceived behavioral control over a destination” increasing by 1 point, the impact of this factor on the factor “intention to return” increases by 0.181 points. We can conclude that tourists’ perceived behavioral control has a positive impact on their intention to return to a destination.

4.3.2.4 Results of testing hypothesis H 4

With hypothesis H 4 : Satisfaction has a positive influence on tourists' intention to return to the destination. Conclusion: accept the hypothesis.

The estimation results show that, using the 95% confidence standard, we see that the Sig of SAT affecting INT is 0.000 < 0.05, so the hypothesis that satisfaction has a positive effect on tourists' intention to return to the destination will be accepted.

With a standardized regression coefficient of 0.352, when tourists rate the factor “satisfaction with a destination” increasing by 1 point, the impact of this factor on the factor “intention to return” increases by 0.352 points. We can conclude that tourists’ satisfaction has a positive impact on their intention to return to the destination.

4.3.2.5 Results of testing hypothesis H 5

With hypothesis H 5 : Destination image in terms of accessibility to resources and resources has a positive impact on tourist satisfaction. Conclusion: accept the hypothesis.

The estimation results show that, using the 95% confidence standard, we see that the Sig of TCTN affecting SAT is 0.000 < 0.05, then the hypothesis that destination image on accessibility to resources and resources has a positive impact on tourist satisfaction will be accepted.

With a standardized regression coefficient of 0.372, when tourists evaluate the factor "destination image in terms of access to resources and resources" increasing by 1 point, the impact of this factor on the factor "satisfaction" increases by 0.372 points. We can conclude that destination image in terms of access to resources and resources has a positive impact on their satisfaction.

4.3.2.6 Results of testing hypothesis H 6

With hypothesis H 6 : Destination image in terms of quality and reputation has a positive influence on tourist satisfaction. Conclusion: accept the hypothesis.

The estimation results show that, using the 95% confidence standard, we see that the Sig of CLDT affecting SAT is 0.000 < 0.05, then the hypothesis that destination image in terms of quality and reputation has a positive impact on tourist satisfaction will be accepted.

With the standardized regression coefficient of 0.309, when tourists evaluate the factor

If “destination image of quality and reputation” increases by 1 point, the impact of this factor on the factor “satisfaction” increases by 0.309 points. We can conclude that the image

Comment