Understanding UX Better
Results
Data assessment
To test the potential risk of common method variance (CMV), Harman single factor analysis was performed. The results showed that there was no single factor that accounted for more than 50% of the total variance, so common method variance was not serious problem in present study.
Reliability and validity test
Firstly, we measured Cronbach's α coefficient for assessing internal consistency. All α values exceeded 0.75, which means high reliability levels. Secondly, a confirmatory factor analysis was conducted to ensure convergent validity of the constructs. As shown in Table 2. The composite reliability (CR) values of all constructs exceeded 0.65; the average variance extracted (AVE) values exceeded 0.5 and the standard factor loadings exceeded 0.65. These results indicate good convergent validity. Finally, we measured the square root of AVE values of all variables and correlation coefficients between variables to test the discriminant validity. The square root of the AVE values of each variable (bold values in Table 3) is higher than the correlation coefficients of other variables, which means good discriminant validity.
Table 2. Confirmatory factor analysis results.
Table 3. Square root and correlation coefficient of mean extracted variance.
Structural model test
We evaluated the theoretical model and hypotheses by using AMOS 21.0, the results show a good model fitness: χ2/df = 1.574, AGFI = 0.873, CFI = 0.959, GFI = 0.898, NFI = 0.896, RMSEA = 0.046. All the fitting indices of the model reached the recommended value, which indicates a good fitness. As shown in Figure 2. The results prove that functional experience has a positive impact on platform dependency (β = 0.567，p < 0.001), which positively influenced platform identity (β = 0.340, p < 0.001), intention to continuously obtain tourism information (β = 0.223, p < 0.05), and intention to continuously provide tourism information (β = 0.193, p < 0.05).Social experience has a significant impact on platform identity (β = 0.601, p < 0.001)，which positively influenced intention to continuously obtain tourism information (β = 0.590, p < 0.001), and intention to continuously provide tourism information (β = 0.588, p < 0.001). Hypothesis 3 was not supported.
Figure 2. Structural model results. ***p < 0.001, *p < 0.05.
