*4.2. Measurement Model Results*

To evaluate the hypotheses, PLS structural equation modelling was used. Multiple parameters were considered in order to assess the validity and reliability of the measurements. Outer loadings, convergen<sup>t</sup> validity, composite reliability, and discriminant validity were required in order to evaluate the measurement model (Tables 4 and 5). As suggested by Hair et al. [87], the outer loadings should exceed 0.4. The smallest outer loading value was 0.556 (SocDis6; Table 4). In the testing reliability, the threshold value of Cronbach's alpha, rho-A, and composite reliability for a given construct is 0.7. All constructs had reliabilities of more than 0.70 (Table 4). Convergent validity was measured by the average variance extracted (AVE), the threshold value of which was 0.5 [88].



Notes: values in boldface are outer loadings, whereas others are cross loadings.


**Table 5.** Assessment of reliability and validity of constructs.

Note: The diagonals (in bold) represent the square root of the AVE.

Discriminant validity was examined using three criteria. Firstly, following Fornell and Larcker [88], the square root of the AVEs of each construct needed be greater than the correlation estimate among the constructs (Table 5). Secondly, the outer loading values on the respective constructs needed to be more significant than their cross-loadings on the other constructs (Table 4). Thirdly, the heterotrait–monotrait (HTMT) ratio and confidence interval needed to be less than 0.85 and 1, respectively [89]. The square root of AVE exceeded the intercorrelations of the constructs in the model (Table 5). This result suggests that the model had a sufficient discriminant validity [90]. The HTMT ratios and corresponding confidence intervals for each pair were less than 0.85 and 1, respectively (Table 6). Thus, the model possessed convergen<sup>t</sup> and discriminant validities.

**Table 6.** Heterotrait–monotrait (HTMT).


Harman's one-factor test [91] was conducted to examine the potential for common method variance. Common method variance is observed when only one factor arises from a factor analysis, or when the first factor describes more than 50% of the variance. Therefore, all items for the constructs were introduced into the factor analysis. The unrotated matrix shows that the first factor explained 38% of the variance. Thus, common method variance was not an issue in this study.

### *4.3. Assessment of the Hierarchical Incivility Construct*

Incivilities were treated as a second-order construct, comprising two first-order reflective constructs (physical and social incivilities) that represent 13 items. Physical incivilities (R2 = 90.6%) and social incivilities (R2 = 79.8%) reflected the degree of the explained variance of the hierarchical construct. The entire path coefficient from its incivilities to its dimensions was significant at *p* < 0.01.

### *4.4. Assessment of the Structural Model*
