**4. Results**

*4.1. Measurement Model*

Table 2 is the descriptive statistics of respondents' answers to the Likert scale.

**Table 2.** Descriptive statistics.


Note: N = number; Min. = Minimum value; Max. = Maximum value; M = Mean value; SD = standard deviations; CII = consumer information innovation; CPI = consumer product innovation; PHV = perceived hedonic value; PNV = perceived novelty value; PSV = perceived social value; PUV = perceived usefulness value; WTB = willing to buy; 1 = completely disagree; 2 = disagree; 3 = partially disagree; 4 = neutral; 5 = partially agree; 6 = agree; 7 = completely agree.

Table 3 displays the findings of the investigation of reliability. It can be seen that Cronbach's α of the seven latent variables is between 0.841 and 0.896, which are all higher than the benchmark value of 0.70. The composite reliability values are between 0.904 and 0.933, which were higher than 0.70, indicating that the reliability of the questionnaire is good, and the collected data are reliable. All item factor loadings should be more than 0.70, as per the Fornell and Larcker criteria [119], and the AVE for each construct was greater than 0.50 [120]. As can be seen in Table 3, the data for this model all met the requirements, indicating strong convergent validity.

Discriminant validity was tested using the Fornell-Larcker criterion [119], which is a measure of the expected level of "difference" between items for different factors. To test the discriminant validity of the model, the AVE of each factor was compared with the correlation square. The value on the diagonal is the square root of the AVE, and the other values are the correlation coefficients between the factors, which are considered to have good discriminant validity when the AVE is greater than the correlation coefficient between the factor and the other factors. From Tables 4 and 5, it can be seen that the model has good discriminant validity.


**Table 3.** Reliability and convergent validity test of the measurement model.

Note: loadings = factor loading; Cronbach = Cronbach's alpha; CR = construct reliability; AVE = average variance extracted, a measure of convergence among observable variables reflecting a latent variable; R2 = coefficient of determination [121]. CII = consumer information innovation; CPI = consumer product innovation; PHV = perceived hedonic value; PNV = perceived novelty value; PSV = perceived social value; PUV = perceived usefulness value; WTB = willing to buy.

#### *4.2. Structural Model*

Evaluate the three research models proposed by the authors according to the research of Hair et al. [115,122]. The hypotheses are tested under these three different conceptual models, and the influences of independent and mediating variables on dependent variables are analyzed.


**Table 4.** Cross-loadings values for each block of indicators.

**Table 5.** Discriminant validity matrix (Fornell and Larcker criterion).


Note: Values (bold) on the diagonal represent the square root of the AVE, while the off-diagonals are correlations.

#### 4.2.1. Research Model A. (without Mediator)

Each path coefficient's statistical significance was evaluated using t-tests, and as noted before, bootstrapping (5000 sub-samples) was utilized to do so.

Table 6 shows the results of direct effects analysis show that the direct effects are significant. The results of the indirect effect analysis show that the indirect effect is significant.

As seen in Figure 3 and Table 6, CPI had a significant positive effect on PUV (t = 17.950, *p* < 0.001), PSV (t = 12.116, *p* < 0.001), PHV (t = 13.867, *p* < 0.001), and PNV (t = 1.690, *p* < 0.001). CII had a significant positive effect on PUV (t = 2.159, *p* < 0.05), PSV (t = 6.065, *p* < 0.001), PHV (t = 4.765, *p* < 0.001), and PNV (t = 4.655, *p* < 0.001) also had a significant positive effect, but it was significantly weaker than the CPI effect on them. PUV (t = 6.537, *p* < 0.001), PSV (t = 3.330, *p* < 0.01), PHV (t = 4.673, *p* < 0.001), and PNV (t = 5.938, *p* < 0.001) all had significant positive effects on WTB.

The value of the coefficient of determination *R*2, which also explains the variance of the regression model, is in the range of [0, 1], and the closer it is to 1, the more the independent variable can explain the variance of the dependent variable, and the larger the value, the better [123]. According to Hair, "0.25 is weak, 0.50 is moderate, and 0.75 is the model's substantial explanatory power" [115]. Table 6 shows that PUV, PSV, PHV, and PNV are moderate, and WTB is strong.


**Table 6.** Hypothesized relationships for all effects.

Note: Path significance: \*\*\* *p* < 0.001; \*\* *p* < 0.01; \* *p* < 0.05. The levels of significance for the *f* <sup>2</sup> statistic are as follows: <sup>a</sup> > 0.02 (little effect), <sup>b</sup> > 0.15 (moderate effect), and <sup>c</sup> > 0.35 (large effect) [115].

In addition to the *R*<sup>2</sup> values, the size effect *f* 2, is used [115]. There are three different effect values: small (greater than 0.02), medium (greater than 0.15), and large (greater than 0.35) [115]. As seen from Table 6, the *f* <sup>2</sup> value of H1a is equal to 0.421, reflecting the strong influence of consumer product innovation on perceived useful value. The effect of consumer information innovation on perceived useful value is negligible (*f* <sup>2</sup> = 0.008).

In Table 6 and Figure 3, the structural model's validity was evaluated using *R*<sup>2</sup> (a measure of predictive accuracy). The *R*<sup>2</sup> of WTB was 0.797%, indicating that the latent variables explained 79.7% of the purchase intention, which was greater than 50% and demonstrates that the model's assumptions are reasonable. The model fit well [115].

**Figure 3.** Findings of structural model analysis. (Note: \*\*\* denote 0.1% significance levels; \*\* denote 1% significance levels; \* denote 5% significance levels.).

From Table 7, it can be concluded that Q<sup>2</sup> > 0 indicates that the structural model is valid [124].


**Table 7.** Cv-communality (Q2 for measurement blocks).

Note: Q<sup>2</sup> = predictive relevance.

According to Wetzels et al. [124], the goodness of Match (GoF) values greater than 0.1 indicate small, 0.25 and 0.36 indicate medium, and GoF values greater than 0.36 indicate largely. The GoF value can be used to examine the model fit, and when the GoF value is higher than 0.36, the model has good applicability [124]. According to GoF's calculation

formula: GoF = communality <sup>×</sup> <sup>R</sup><sup>2</sup> [125]. GoF <sup>=</sup> <sup>√</sup>0.519 <sup>×</sup> 0.669 <sup>=</sup> 0.589. This model's GoF value was calculated to be 0.589, indicating a decent fit.

4.2.2. Research Model B. (Perceived Value as a Mediator for CPI and WTB; Perceived Value as a Mediator for CII and WTB)

Each path coefficient's statistical significance was evaluated using t-tests, and as noted before, bootstrapping (5000 sub-samples) was utilized to do so. The results are shown in Figure 4.

**Figure 4.** Findings of structural model A analysis. (Note: \*\*\* denotes 0.1% significance levels; \* denotes 5% significance levels.).

Table 7 shows the hypothetical relationships for all effects of study model B.

Analysis results show that the total effect, as well as the CPI of WTB total allow effect, are remarkable. The direct effects of CII on PHV, PNV, PSV, and PUV were significant. The direct effect of PHV, PNV, PSV, and PUV on WTB was significant. The direct effect of CPI on WTB is significant. The indirect effect analysis shows that CPI→PSV→WTB had a significant effect, the indirect effect of CPI→PNV→WTB was significant, the indirect effect of CPI→PUV→WTB was significant, and the indirect effect of CPI→PHV→WTB is significant. Therefore, according to the literature of James, L. and Brett, J. [80], the author will further study whether PHV, PNV, PSV, and PUV have partial mediating effects on CPI and WTB.

Table 8 analyzes the mediating effect of PUV on CPI and WTB, the mediating effect of PSV on CPI and WTB, the mediating effect of PHV on CPI and WTB, and the mediating effect of PNV on CPI and WTB. According to the literature published by Hair et al. [114], it can be known that VAF represents the percentage of indirect effect and total effect. As a rule of thumb, VAF values are divided into three levels: VAF < 20% indicates no mediation effect, 20%≤ VAF ≤ 80% indicates partial mediation effect, and VAF > 80% indicates complete mediation effect. It can be inferred from the value of VAF that PUV partially mediates CPI and WTB. PSV, PHV, and PNV have no mediating effect on CPI and WTB.


**Table 8.** Hypothesized relationships for all effects (research model B).

Note: Path significance: \*\*\* *p* < 0.001; \*\* *p* < 0.01; \* *p* < 0.05. The levels of significance for the *f* <sup>2</sup> statistic are as follows: <sup>a</sup> > 0.02 (little effect), <sup>b</sup> > 0.15 (moderate effect), and <sup>c</sup> > 0.35 (large effect) [115].

Table 9 shows that the results of the total effect analysis showed that the total effect of CII on WTB was significant. The direct effect analysis showed that CII significantly impacted PHV, PNV, and PSV. The direct effect of PHV, PNV, and PSV on WTB was significant. The direct effect of CII on WTB was insignificant. The indirect effect analysis showed that CII→PHV→WTB had a significant indirect effect. The indirect effect of CII→PNV→WTB was significant. The indirect effect of CII→PSV→WTB was significant. Therefore, according to James, L. and Brett, J.'s research [126], the equation is tested by the coefficient c to distinguish between full and partial mediation. If the indirect effects are significant, but the coefficient c is not significant, it belongs to perfect mediation. PHV, PNV, and PSV were judged to mediate CII and WTB fully. The results are shown in Table 10.


**Table 9.** Mediation Tests for Parallel-Sequential Multiple Mediator Models (CPI→WTB).

Note: Path significance: \*\*\* *p* < 0.001; \* *p* < 0.05.

**Table 10.** Mediation Tests for Parallel-Sequential Multiple Mediator Models (CII→WTB).



Note: Path significance: \*\*\* *p* < 0.001; \*\* *p* < 0.01; \* *p* < 0.05.

#### **5. Discussion**

This study explores the characteristics of consumers who are willing to purchase innovative car seats. Through the study of model A, the hypothesis of this study has been confirmed. Direct effect analysis shows that both consumer product and information innovation have significant effects on perceived value. Truong, Y. Previously found that consumers' innovation ability would positively influence perceived value [109]. The results of this study show that perceived value (perceived product's usefulness, social value, hedonic value, and novelty value) has a significant positive impact on consumers' purchase intention. The previous article has a similar conclusion that the intention to use will be positively affected by perceived usefulness [66,67]. Hedonic value positively impacts consumers' behavioral intentions [74]. Jaleel, A. et al. found that perceived social influence and social value significantly impact usage intention [73]. A recent study by Adapa et al. established a positive correlation between perceived novelty and use intent [77].

In the total effect analysis of this study, consumer product and information innovation positively impact consumers' willingness to buy innovative car seats. This result is consistent with the opinions of articles in other industries. Lee, K. et al. studied whether product innovation significantly impacts the intention to buy smartphone products. The research results show that product innovation significantly positively impacts the intention to buy mobile phone products [127].

By studying model B, the following results are obtained. Perceived product usefulness plays a partial mediating role between consumer product innovation and purchase intention. Perceived social value, hedonic value, and novelty value have no mediating effect on consumer product innovation and purchase intention but only an indirect effect. The impact of consumer product innovation on purchase intent is still much explored in other industries. Saputra M. et al. studied the mediating role of green customer value. The research results show that green customer value has been proven to partially mediate between green product innovation and purchase intention [60].

Model B confirmed the following results. A product's perceived social value, hedonic value, and novelty value fully mediate between consumer information innovation and purchase intention. Researchers are investigating the public acceptance of self-driving cars. Research results show that perceived value fully mediates between consumer innovation and public acceptance of innovative products [128]. Hong et al. [85] proved that hedonic value and utilitarian value play an intermediary role in smartwatch consumers' innovation and use intention.

This study also found that the perceived usefulness of products had no mediating or indirect effect between consumers' information innovation and purchase intention. In the previous study, Abdurrahman C. and Umut A. explored the adoption of smart home devices [129]. The results show an insignificant relationship between innovation and perceived usefulness in specific domains.

According to the findings above, CPI has a significantly better effect on perceived value than CII [80,130], which means that consumers with a high CPI are more likely to perceive the value of an intelligent safety seat and may be more willing to pay for it than those with a high CII.
