4.2. Measurement Model
To test the model fit between the research model and empirical data, a CFA model was conducted with AMOS 18 statistical software. According to Kline [
45], a good model fit is indicated when Chi-square/degree of freedom is less than 3, GFI (goodness of fit index), CFI (comparative fit index), and NFI (normed fit index) are all greater than 0.90, and RMSEA (root-mean-square error of approximation) is less than 0.08. The results of CFA in this study show that all criteria met these requirements (χ
2/d.f. = 83.33/38 = 2.19, GFI = 0.97, CFI = 0.98, NFI = 0.97, and RMSEA = 0.05). Thus, this study shows a good model fit between empirical data and research models.
Internal consistency reliability is often tested using Cronbach’s alpha. Reliability is acceptable if the value of Cronbach’s alpha is greater than 0.70 [
45,
46].
Table 4 shows that Cronbach’s alpha for environmental consciousness, perceived food quality, price sensitivity, and organic food purchase intention were 0.82, 0.75, 0.70, and 0.90, respectively. Therefore, the reliability of all variables is satisfactory in this study.
This study followed Kline’s [
45] suggestions and calculated composite reliability (CR), average variance extracted (AVE), and the square root of average variance extracted (√AVE) to test the validity of the measures.
Table 4 shows the results of these indicators. Kline [
45] noted that if CR is greater than 0.70 and AVE is greater than 0.50, then convergent validity is acceptable. Results in
Table 4 indicate that all CR and AVE values of all variables exceed the thresholds. Thus, convergent validity is satisfactory in this study. Furthermore, Kline [
45] suggested that when values of the square root of AVE of all variables are greater than all correlation coefficients, discriminant validity is acceptable. Results in
Table 3 show that values of √AVE on the main diagonal exceed all correlation coefficients between variables. Thus, the discriminant validity of the measures in this study is satisfactory.
Podsakoff et al. [
47] suggested that a self-reported survey should check the common method variance (CMV) problem due to the same respondent providing measures for independent and dependent variables simultaneously. Podsakoff et al. [
47] stated that the CMV problem may occur if the results of the principal component analysis (PCA) with an unrotated solution show a single factor solution or the first factor accounts for a majority of the variance. Results of PCA in this study indicate that four factors emerged, which accounted for 76.54% of the variance. The first factor accounted for 35.16% of the variance. To confirm this result, a one-factor model of CFA was conducted in this study. Results of this CFA model present that all criteria did not meet the requirements of model fit (χ
2/d.f. = 15.88, GFI = 0.79, CFI = 0.74, NFI = 0.73, and RMSEA = 0.17). This CMV problem may be ignored in this study.
4.3. Hypotheses Testing
This study used hierarchical regression approach to analyze data. Based on previous organic food literature [
4,
23,
24], some common variables were controlled in the analysis due to their potential impact on a dependent variable (i.e., gender, age, marital status, income, and purchase experience).
Table 5 presents the results of hierarchical regression analysis for testing the hypotheses in this study. As shown in Model 1a, purchase experience has a negative effect on organic food purchase intention (β = −0.13,
p < 0.01). All other control variables were not significantly related to organic food purchase intention. Fundamentally, consumers’ intention to purchase organic food does not depend largely on their personal characteristics (i.e., gender, age, income, marriage).
Furthermore, Model 2a in
Table 5 tests the direct effect of environmental consciousness on organic food purchase intention. Results show that environmental consciousness was positively related to organic food purchase intention (β = 0.54,
p < 0.001). This result indicates that environmental consciousness enhances consumers’ intention to purchase organic food. Thus, H1 was supported.
This study followed the procedure proposed by Baron and Kenny [
48] to test the mediation of perceived food quality in the link between environmental consciousness and organic food purchase intention. First, the independent variable must be significantly related to the dependent variable in the first model. Second, the independent variable must be significantly related to the mediating variable in the second model. Third, when regressing the independent and mediating variables on the dependent variable in the third model, the mediating variable must be significantly related to the dependent variable. If the effect of the independent variable on the dependent variable is significantly reduced, then partial mediation is supported. However, if the independent variable is insignificantly related to the dependent variable, then full mediation is supported. Accordingly, results from Models 2a–4a in
Table 5 show that environmental consciousness was positively related to organic food purchase intention in Model 2a (β = 0.54,
p < 0.001) and perceived food quality in Model 3a (β = 0.36,
p < 0.001). Furthermore, perceived food quality was positively related to organic food purchase intention in Model 4a (β = 0.32,
p < 0.001). Environmental consciousness was also positively related to organic food purchase intention in Model 4a (β = 0.42,
p < 0.001). However, the effect of environmental consciousness on organic food purchase intention was significantly reduced in Model 4a (β = 0.42) as compared with that in Model 2a (β = 0.54). This result indicates that perceived food quality partially mediates the relationship between environmental consciousness and organic food purchase intention. This study also followed Preacher, Rucker, and Hayes [
49] to conduct a bootstrap analysis with 1000 bootstrap samples to confirm the mediating effect of perceived food quality in the relationship between environmental consciousness and organic food purchase intention. Results show that the indirect effect of environmental consciousness on organic food purchase intention through perceived food quality was statistically significant (environmental consciousness perceived food quality organic food purchase intention: β = 0.073,
p < 0.001, 95% CI = 0.031–0.140). This result provides further evidence for the mediating role of perceived food quality in the relationship between environmental consciousness and organic food purchase intention. Thus, H2 was supported.
This study adopted a multiple regression analysis method proposed by Baron and Kenny [
48] to test the moderating role of price sensitivity on the relationship between perceived food quality and organic food purchase intention. Results are shown from Model 1b to 3b in
Table 5. First, control variables were included in Model 1b. Second, the main effects of perceived food quality and price sensitivity on organic food purchase intention were tested in Model 2b. Third, the interaction effect between perceived food quality and price sensitivity on organic food purchase intention was tested in Model 3b. As indicated in Model 3b in
Table 5 and
Figure 2, the interaction effect between perceived food quality and price sensitivity was significantly and negatively related to organic food purchase intention (β = −0.11,
p < 0.01). This result implies that price sensitivity weakens the relationship between perceived food quality and organic food purchase intention. Ultimately, the effect of perceived food quality on organic food purchase intention was stronger when price sensitivity is low and weaker when price sensitivity is high. Thus, H3 was supported.
This study also followed the procedure proposed by Edwards and Lambert [
50] to conduct a bootstrap analysis for testing the moderated mediation hypothesis.
Table 6 shows that the results of the second stage analysis reveal the effect of perceived food quality on organic food purchase intention for low- and high-price sensitivity groups. Results indicate that the influence of perceived food quality on organic food purchase intention varied significantly between the two groups (∆β = 0.071,
p < 0.001). That is, the influence of perceived food quality on organic food purchase intention was stronger for low-price sensitivity group (β = 0.325,
p < 0.001) than that for the high-price sensitivity group (β = 0.254,
p < 0.001). This result provides further support for the moderating effect of price sensitivity on the link between perceived food quality and organic food purchase intention. Thus, H3 was further confirmed.
Furthermore, results in
Table 6 also exhibit that the indirect effect of environmental consciousness on organic food purchase intention through perceived food quality is stronger when price sensitivity is low (β = 0.102,
p < 0.001) and weaker when price sensitivity is high (β = 0.075,
p < 0.001). The indirect effect of environmental consciousness on organic food purchase intention varied significantly between low and high levels of price sensitivity (∆β = 0.027,
p < 0.01). Therefore, H4 was supported.
In sum, this study used analysis methods proposed by Baron and Kenny [
48] and confirmed by analysis procedures proposed by Preacher et al. [
49] and Edwards and Lambert [
50] to test the mediation and moderation. The results of these methods provide evidence to support H2, H3, and H4. That is, the mediating roles of perceived food quality and price sensitivity were supported in this study.