Normalized Weights of Destination Image Scale

Table 4.9: Standardized weights of destination image scale


Observed Variable Factor

Weight

Observed Variable Factor

Weight

AT4

<---

AT

.555

AMP2

<---

AMP

.733

AT3

<---

AT

.548

AMP1

<---

AMP

.692

AT2

<---

AT

.604

AC5

<---

AC

.533

AT1

<---

AT

.562

AC4

<---

AC

.661

PV3

<---

PV

.800

AC3

<---

AC

.733

PV2

<---

PV

.863

AC2

<---

AC

.616

PV1

<---

PV

.677

INF5

<---

INF

.693

AT5

<---

AT

.683

INF4

<---

INF

.759

AT6

<---

AT

.706

INF3

<---

INF

.709

AMP5

<---

AMP

.605

INF2

<---

INF

.786

AMP4

<---

AMP

.731

INF1

<---

INF

.653

Maybe you are interested!

Source: Author's analysis and synthesis from 2014 survey data


Table 4.10: Discriminant validity results between concepts of destination image scales


Relationship

Estimate

SE

CR

p

AT

<-->

AMP

0.319

0.039

11,949

0.000

PV

<-->

AMP

0.237

0.042

14,185

0.000

PV

<-->

AC

0.319

0.040

12,754

0.000

AT

<-->

PV

0.687

0.040

12,804

0.000

AMP

<-->

AC

0.666

0.041

13,056

0.000

AT

<-->

AC

0.573

0.031

8,472

0.000

AT

<-->

INF

0.486

0.034

9,660

0.000

AMP

<-->

INF

0.446

0.040

12,939

0.000

PV

<-->

INF

0.494

0.037

10,818

0.000

AC

<-->

INF

0.335

0.034

9,484

0.000

Source: Author's analysis and synthesis from 2014 survey data.


* Convergent value of the scale: Gerbring and Anderson [40] stated that the scale achieved convergent value when the standardized weights of the scale were greater than 0.5; and had a significant P-value of less than 0.05. According to the CFA coefficients of the component scales of the destination image presented in the table (Table 4.9 ) , all observed variables had weights > 0.5, so the scale achieved convergent value.

*Regarding the discriminant value between the research concepts (Table 4.10), it shows that the correlation coefficients along with the standard deviation of each pair of concepts are different from 1 (p< 0.05) at the 95% confidence level. Therefore, the research concepts (5 components) have discriminant value.

4.2.3.2 Loyalty scale (loyalty attitudes and behaviors)

The second data set for the loyalty variables was submitted to a CFA, using multivariate (ML) estimation: the two-factor (component) model fit the data quite well (Table 4.11). Figure 4.3 shows that the Chi-square (χ2) = 25.627, which is significant at p < 0.05, indicating that the model fits the data well. The degrees of freedom (χ2/df = 4.271) are lower than recommended (i.e. < 5.0; Bollen, 1989). The RMSEA value indicates that the two-component model had a tentatively acceptable fit (RMSEA = 0.091; Hu and Bentler, 1999). The SRMR (0.091) was ≤ 0.10; (Kline, 2005). The CFI was 0.984, which is considered acceptable (Kline, 2005). The GFI was 0.979 as indicative of an acceptable model, as was the TLI of 0.960, supporting the finding that the model fits the market data well.



Figure 4.3: CFA results of loyalty scale (standardized) Source: Analytical data surveyed by the author in 2014

Table 4.11: Standardized weights of the loyalty scale


Observation variable


Ingredient

Weight

A1

<---

ATL

.777

A2

<---

ATL

.733

A3

<---

ATL

.705

B1

<---

BHL

.892

B2

<---

BHL

.883

B3

<---

BHL

.777

Source: Author's analysis and synthesis from 2014 survey data.


Table 4.12: Discriminant validity results between loyalty concepts


Relationship

Estimate

SE

CR

p

ATL

<-->

BHL

.707

0.032

9.020

0.000

Source: Author's analysis and synthesis from 2014 survey data


The correlation coefficients between the concepts with the accompanying standard deviations (Table 4.12 ) show us that these coefficients are less than 1 (statistically significant), so the concepts: Attitude loyalty (ATL) (including 3 observed variables) and behavioral intention loyalty (BHL) (including 3 observed variables) have discriminant validity.

Table 4.13: Composite Reliability (CR) and Variance Extracted (AVE) Results



Concept

Observation variable


Alpha Reliability


CR


AVE

Destination appeal

6

0.843

0.782

0.476

Tourism infrastructure

5

0.924

0.867

0.523

Tourist atmosphere

4

0.822

0.785

0.479

Accessibility

4

0.738

0.733

0.410

Affordable

3

0.738

0.825

0.614

Attitude of loyalty

3

0.790

0.813

0.591

Loyalty behavior

3

0.872

0.872

0.697

Source: Author's analysis and synthesis from 2014 survey data.

Regarding the composite reliability (CR) and average variance extracted (AVE) (Table 4.13), the results show that the research concepts of the scale all have good composite reliability, but the variance extracted is not high, including the components Destination Attractiveness (AT) and Tourism Atmosphere (AMP) with AVE less than 0.5, however, this value can be temporarily accepted as 5, because the value is close to the requirement, and at the same time, it is a topical topic. However, the component "Accessibility" (AC) has a slightly low AVE, and is also not topical, so it can be considered to remove this component from the proposed model.

4.2.4 Results of SEM theoretical structural model analysis


Figure 4.4: SEM structural model

Source: Data analyzed by the author in 2014


Table 4.14: Summary of CFA standards


Standard specifications

Result

Conclude

Chi-square has P-value >0.5


Model fit to market data

GFI ≥ 0.90

0.887

TLI≥0.90

0.905

CFI≥0.90

0.917

RMSEA ≤ 0.08

0.056

Chi-square/df ≤ 5

2,235

Standardized factor weight ≥ 0.5

≥ 0.55

Satisfied

Composite reliability ≥ 0.60

≥ 0.7

Satisfied

Variance Extracted (AVE) ≥ 0.35

≥ 0.45

Satisfied


Source: Author's synthesis and analysis from 2014 survey data.


4.2.5 Testing the research model


Testing the discriminant validity between the concepts of destination image and destination loyalty.

The SEM results show that the relationships between the research concepts are all different from 1 (p< 0.05) at the 95% confidence level (Table 4.15). Thus, the concepts of destination image and destination loyalty have discriminant validity.

Table 4.15: Discriminant validity results between the concepts of the image and destination loyalty scales

Relationship

Estimate

SE

CR

p

ATL

<---

INF

0.499

0.040

12,587

0.000

ATL

<---

AT

0.072

0.046

20,257

0.000

ATL

<---

AMP

0.134

0.046

19,026

0.000

ATL

<---

PV

0.161

0.045

18,508

0.000

ATL

<---

AC

0.114

0.046

19,416

0.000

BHL

<---

INF

0.33

0.043

15,453

0.000

BHL

<---

AT

0.401

0.042

14,236

0.000

BHL

<---

AMP

0.034

0.046

21,043

0.000

BHL

<---

PV

0.122

0.046

19,259

0.000

BHL

<---

AC

0.015

0.046

21,447

0.000

Source: Author's analysis and synthesis from 2014 survey data

Testing research hypotheses


In the linear structure analysis (Table 4.16), the component “Accessibility (AC)” was eliminated from the official research model because the relationship between this concept and Attitude Loyalty (ATL) and Behavior Loyalty (BHL) was not statistically significant at the 90% confidence level. This means that the hypothesis H4A:Accessibility” has a positive impact on tourists’ attitude loyalty ); and the hypothesis H4B:Accessibility” has a positive impact on tourists’ behavior loyalty was rejected at the 90% significance level.

Furthermore, the relationship between the component “Destination Attractiveness (AT)” and “Attitudinal Loyalty (ATL)” is not statistically significant at the 90% confidence level. As well as the relationship between the component “Tourism Atmosphere (AMP)” and Tourist Loyalty Behavior is not statistically significant at the 90% confidence level. This means that the hypothesis H1A (“ Destination Attractiveness” has a positive impact on Tourist Attitude Loyalty ) and hypothesis H3B ( “Tourism Atmosphere” has a positive impact on Tourist Loyalty Behavior ) are rejected and will not be considered in the next formal model when analyzing the multi-group SEM structural linear model.

Table 4.16: Test of the relationship of the research model (n=396)


Relationship

E(β)

SE

CR (t)

P

ATL

<---

INF

.501

.086

5.801

***

ATL

<---

AT

.079

.097

.809

.418

ATL

<---

AMP

.131

.057

2,300

.021

ATL

<---

PV

.141

.052

2,692

.007

ATL

<---

AC

.141

.113

1,244

.213

BHL

<---

INF

.329

.081

4,071

***

BHL

<---

AT

.437

.104

4,217

***

BHL

<---

AMP

.033

.055

.596

.551

BHL

<---

PV

.106

.051

2,086

.037

BHL

<---

AC

.018

.110

.167

.868

Source: Author's analysis and synthesis from 2014 survey data

Table 4.17: Results of testing research hypotheses


Contents of the hypotheses

Test results

H1A: “Destination attractiveness” has a positive impact


to tourist loyalty


Not accepted at 90% significance level

H2A: “Tourism infrastructure” has a positive impact


to tourist loyalty

Accept at level


99% meaning

H3A: “Tourism atmosphere” has a positive impact


to tourist loyalty

Accept at level


95% significance

H4A: “Accessibility” has a positive impact on


tourist loyalty


Not accepted at 90% significance level

H5A: “Affordability” has a positive impact on


tourist loyalty

Accept at level


95% significance

H1B: “Destination attractiveness” has a positive impact


to tourist loyalty behavior

Accept at level


99% meaning

H2B: “Tourism infrastructure” has a positive impact


to tourist loyalty behavior


Accept at 99% significance level

H3B: “Tourism atmosphere” has a positive impact


to tourist loyalty behavior


Not accepted at 90% significance level

H4B: “Accessibility” has a positive impact


to the loyal behavior of tourists


Not accepted at 90% significance level

H5B: “Affordability” has a positive impact on


tourist loyalty behavior

Accept at level


95% significance

Source: Author's synthesis from analysis of 2014 survey data

However, the analysis of the linear structural model (SEM) with estimation through the ML method, gives the results of the relationships between “Tourism infrastructure (INF)”; “Tourism atmosphere (AMP)” and “Affordability (PV)” respectively have a positive impact on “Tourist loyalty attitude” at a significance level of over 95% with statistical value (t) > 1.96. Therefore, it can be affirmed that hypotheses H1A, H3A and H4A are accepted. Next, the relationship between the components “Tourism infrastructure (INF)”; Destination attractiveness (AT)”; “Affordability (PV)” has a positive impact on “Tourist loyalty behavior” at a significance level of over 96.3% with statistical value

(t) > 1.96. Therefore, hypotheses H1B, H2B and H4B are accepted.


In summary, the results show that the two components “Tourism Infrastructure (INF) and Affordability (PV)” have positive impacts on both tourists’ attitude and loyalty behavior, especially the component “Destination Attractiveness (AT)” has a significant positive impact on loyalty behavior but has no impact on tourists’ attitude loyalty. On the contrary, the component “Tourism Atmosphere (AMP)” has a significant impact on tourists’ attitude loyalty, but has no impact on tourists’ loyalty behavior. The summary of the hypothesis testing results is shown in Table 4.17.

The official SEM structural model after eliminating hypotheses H1A, H4A, H3B, H4B is shown in Figure 4.5.

Table 4.18: Summary of CFA standards


Standard specifications

Result

Conclude

Chi-square has P-value >0.5


Model fit to market data

GFI ≥ 0.90

0.894

TLI≥0.90

908

CFI≥0.90

920

RMSEA ≤ 0.08

0.60

Chi-square/df ≤ 5

2,442

Standardized factor weight ≥ 0.5

≥ 0.55

Satisfied

Composite reliability ≥ 0.60

≥ 0.80

Satisfied

Variance Extracted (AVE) ≥ 0.35

≥ 0.476

Satisfied

Source: Author's synthesis from analysis of 2014 survey data