**9. Data Analysis**

**9.8 Model and Hypotheses Testing**

contribution to the model. The smaller the value of significance, the greater the contribution that that predictor has (Field, 2009). The standardized beta tells how much the independent variable changes in terms of number of standard deviation (SD) if the independent variable changes of one SD. This value is particular useful because it assesses how much each predictor contributes to the model.

Finally, the models are tested to judge whether the hypotheses are confirmed or rejected.

*Table 9.7 Model Summary 1 *

**Model Summary**

Model R R Square Adjusted R

Square Std. Error of

the Estimate F Sig. F Durbin-Watson

1 .843^{a} .711 .707 .62033 170.116 .000 2.165

Table 9.8 gives us the estimates of b-values that can be used to write the equation of the model:

CE = - 0.210 + 0.535 (HED) + 0.435 (UTI)

Hp1 states that “the utilitarian value of the mobile AR technology has a significant and positive
*effect on the customer experience”. The test found the correlation between UTI and CE significant (t *

= 6.348; p < 0.001) with a standardized coefficient (ß) of 0.428. Thus, Hp1 is supported because the impact of the utilitarian dimension on the customer experience was significant and positive.

Hp3 postulates that “the hedonic value of the mobile AR technology has a significant and positive
*effect on the customer experience”. From the table it appears that the correlation between HED and *
CE is significant (t = 5.856; p < 0.001) and that there is a positive impact with the value of Beta (ß )
of 0.466. Thus, also Hp3 is supported.

*Table 9.8 : Regression Analysis Coefficients Model 1 *
Unstandardized

Coefficients Standardized Coefficients

t Sig.

B Std. Error Beta Tolerance VIF

1 (Constant) -.210 .306 -.688 .493

**HED ** .535 .084 .466 6.384 .000 .392 2.551

**UTI ** .435 .074 .428 5.856 .000 .392 2.551

9.8.2 Independent variable: UTI; Dependent variable: HED

To test Hp2, a simple regression analysis was run on a model constructed with the utilitarian dimension of the AR technology as the independent variable and the hedonic dimension of AR as the dependent variable. HED and UTI highly and significantly (p<0.001) correlate; However, the Person correlation coefficients (r = 0.780) was below the suggested multicollinearity threshold of 0.9 (table 9.9).

*Table 9.9 : Correlations Model 2 *

**Correlations**

**HED ** **UTI **

Pearson Correlation **HED ** 1.000

**UTI ** .780 1.000

By looking at the value of Adjusted R² it is possible to see that UTI accounts for 60.5% of the variation in HED (table 9.10). Furthermore, the shrinkage between the Adjusted R²and the R² was only 0.003, thus the cross-validity of model is confirmed. The F-ratio had a value of 215.651 significant for p<0.001, which confirms that our model explains our set of data better than the simple mean of the dependent variable. Moreover, the Durbin-Watson value was confirmed to be close to 2.

*Table 9.10 : Model Summary 2 *

**Model Summary**

Model R R Square

Adjusted R Square

Std. Error of

the Estimate F Sig. F Durbin-Watson

2 .780^{a} .608 . 605 .62728 215.651 .000 2.263

Table 9.11 gives us the estimates of b-value that can be used to write the equation of the model:

HED = - 2.021 + 0.690 (UTI)

Hp2 claims that “The utilitarian value of the mobile AR technology has a significant and positive
*effect on the hedonic dimension”. The standardized coefficient (ß) had a positive and value of 0.780 *
and t-test had a significant value (p < 0.001) of 14.685. Thus, we can support the hypothesis that UTI
has a positive and significant effect on HED.

*Table 9.11: Regression Analysis Coefficients Model 2 *
Unstandardized Coefficients

Standardized Coefficients

t Sig.

B Std. Error Beta

2 (Constant) 2.021 .258 7.843 .000

**UTI ** .690 .047 .780 14.685 .000

9.8.3 Independent variables: INT, INF, VIV; Dependent variables: UTI

To test Hp4, Hp6 and Hp8, we constructed a model in which the AR characteristics of vividness, interactivity and informativeness are the independent variables and the utilitarian dimension of the AR technology is the dependent variable. A multiple regression analysis was run to test the significances of the effects. The independent variables have significant (p<0.001) and moderate correlations, only in two cases they had high correlations (r = 0.767 and r = 0.705). Field (2009) suggested that the value of 0.9 or more is a clear signal of multicollinearity. A second test to excluded this issue is to look at the value of VIF and Tolerance Statistics, which were both in the acceptable range (Table 9.12).

*Table 9.12 : Correlations Model 3 *

**Correlations**

**UTI ** **INT ** **INF ** **VIV **

Pearson Correlation **UTI ** 1.000

**INT ** .767 1.000

**INF ** .705 .692 1.000

**VIV ** .597 .549 .482 1.000

In table 9.13, the value of the Adjusted R² said that the three technology characteristics (VIV, INT, INF) accounts for 66.9% of the variation in the utilitarian dimension of the AR technology.

Moreover, by looking at its difference with the R² it can be claimed that there is cross-validity for the model since the shrinkage was only 0.007. The F-ratio had a value of 95.367 significant for p<0.001, which proves that our model explains our set of data better than the simple mean of the dependent variable. Furthermore, the Durbin-Watson had a value close to 2 which confirms that errors in the regression are independent.

*Table 9.13 : Model Summary 3 *

**Model Summary**

R Square

Adjusted R Square

Std. Error of the

Estimate F Sig. F Durbin-Watson

.676 .669 .64917 95.367 .000 2.101

Table 9.14 gives us the estimates of Beta values that can be used to write the equation of the model:

UTI = - 0.303 + 0.470 (INT) + 0.330 (INF) + 0.253 (VIV)

Hp4 asserts that “Vividness has a significant and positive impact on the utilitarian value of AR”.

This hypothesis is supported by the regression analysis on the proposed model because the t-test was significant (t = 3.538; p < 0.01) and the standardized beta had a coefficient which is positive (ß = 0.209).

Hp6 claims that “Interactivity has a significant and positive impact on the utilitarian value of
*AR”. Since the T-test was found significant (t = 6.274; p < 0.001) and the Standardized coefficient *
(ß) had a positive value (0.449), it can be inferred that Hp6 is supported.

Hp8 postulates that “Informativeness has a significant and positive impact on the utilitarian
*value of AR”. The test has been found significant (t = 4.297; p < 0.001) and value of the Standardized *
coefficient is positive (ß = 0.294), hence also Hp8 is supported.

*Table 9.14 : Regression Analysis Coefficients Model 3 *
Unstandardized

Coefficients

Standardized Coefficients

t Sig.

B Std. Error Beta Tolerance VIF

3 (Constant) -.303 .365 -.831 .408

**INT ** .470 .075 .449 6.274 .000 .461 2.169

**INF ** .330 .077 .294 4.297 .000 .506 1.976

**VIV ** 253 .072 .209 3.538 .001 .679 1.473

**- **

9.8.4 Independent Variables: INT and VID; Dependent Variable: HED

To test Hp5 and Hp7 we constructed a model in which the AR characteristics of vividness and interactivity are the independent variables and the hedonic dimension of the AR technology is the dependent variable. A multiple regression analysis was run. The two independent variables had moderate correlations, all significant (p<0.001). The values of VIF and tolerance statistics were in the acceptable range (Table 9.15). Thus, multicollinearity can be excluded.

*Table 9.15 : Correlations Model 4 *

**Correlations **

**HED ** **INT ** **VIV **

Pearson Correlation **HED ** 1.000

**INT ** .667 1.000

**VIV ** .591 .549 1.000

In table 9.16 the value of the Adjusted R² proved that VIV and INT account for 51.1% of the variation in HED. We can infer that there is cross-validity for the model since the shrinkage was only 0.007. The F-ratio had a value of 74.053 significant for p<0.001, which confirms that our model explains our set of data better than the simple mean of the dependent variable. With a value of 2.054, the Durbin-Watson test proved that the assumption regarding the measurement error is valid.

*Table 9.16 : Model Summary 4 *

**Model Summary**

R Square Adjusted R

Square Std. Error of the Estimate F Sig. F Durbin-Watson

.518 .511 .69839 74.053 .000 2.054

Table 9.17 gives us the estimates of b-value that can be used to write the equation of the model:

HED = 1.388 + 0.454 (INT) + 0.345 (VIV)

Hp5 states that “Vividness has a significant and positive impact on the hedonic value of AR”. The correlation between Vividness and the Hedonic dimension was significant (t = 4.558; p < 0.001) and it had a positive sign (ß = 0.322). Therefore, hypothesis 5 is supported.

Hp7 proposes that “Interactivity has a significant and positive impact on the hedonic value of
*AR”. The T-test was significant (t = 6.934; p < 0.001) and the standardized coefficient has a positive *
value (ß = 0.490), thus Hp7 is supported.

*Table 9.17 : Regression Analysis Coefficients Model 4 *
Unstandardized

Coefficients

Standardized Coefficients

t Sig.

B Std. Error Beta Tolerance VIF

4 (Constant) 1.388 .375 3.705 .000

**HED ** .454 .065 .490 6.934 .000 .699 1.431

**UTI ** .345 .076 .322 4.558 .000 .699 1.431