*5.3. Structural Model Results*

Before measuring the structural model, we measured co-linearity (ensuring sufficient independence between variables), and the VIF for all the variables was below 3; thus, co-linearity was not a major issue in this study. After ensuring the reliability and validity of the model, we tested the hypotheses using the structural model. The path coefficients and significance test results from the structural model are shown in Table 3. Overuse of DCTAs had a positive and significant effect on awareness of the consequences, ascription of responsibility, and perceived life inconvenience, with H1a, H1b, and H1c being supported. Awareness of the consequences had a significantly positive effect on the ascription of responsibility and personal norms, thus supporting H2a and H2b. Ascription of responsibility had a significantly positive effect on personal norms, supporting H2c. Personal norms had a positive impact on the willingness to consistently cooperate with the government in COVID-19 prevention, supporting H2d. In addition, none of the control variables had a significant effect on the users' intention to consistently cooperate with the government in COVID-19 prevention.


**Table 3.** Assessment of the structural model.

Abbreviations: CTO—COVID-19 Tracking App Overuse; ACS—Awareness of Consequences; ARE—Ascription of Responsibility; PLI—Perceived Life Inconvenience; PNO—Personal Norm; CRC—Continue to Cooperate Intention to Prevent COVID-19.

Finally, we evaluated the goodness of fit (GOF) of the model using the standardized root mean square residuals (SRMR). The SRMR value for the model is 0.068, which is less than the threshold value of 0.08. Thus, the fit of the model is satisfactory [58].

#### *5.4. Moderating Effect Results*

The perceived life inconvenience was used as a moderating variable; its moderating effect was measured through two steps in this study. First, we measured the significance of the moderating effect; second, we measured the strength of the moderating effect by calculating F2 as follows: (R2 interaction model − R2 main effects model)/(1 − <sup>R</sup><sup>2</sup> main effects model). If F<sup>2</sup> is between 0.02 and 0.15, it indicates a small moderating effect; if it is

between 0.15 and 0.35, it indicates a moderate moderating effect; and if it exceeds 0.35, it indicates a high moderating effect [59,60].

The moderating effects are shown in Table 4. The perceived life inconvenience significantly reduced the effect of awareness of the consequences on the ascription of responsibility (β = −0.158, *p* < 0.01), thus supporting H3a. Perceived life inconvenience also significantly reduced the effect of ascription of responsibility on personal norms (β = −0.158, *p* < 0.01), thus supporting H3c. However, perceived life inconvenience had no significant moderating effect on awareness of the consequences and personal norms, and H3b was rejected (β = −0.078, n.s.).


Abbreviations: ACS—Awareness of Consequences; ARE—Ascription of Responsibility; PLI—Perceived Life Inconvenience; PNO—Personal Norm.

Slope plots are provided as part of the moderating effect analysis to provide a more visual response to the enhancing/weakening effect of the moderating variable on a specific relationship. We performed slope analysis on the significant moderating relationships. The results are shown in Figures 5 and 6. Perceived life inconvenience significantly reduced the predicted effect of awareness of the consequences on the ascription of responsibility, with a "medium" effect size (<sup>β</sup> <sup>=</sup> −0.158, *<sup>p</sup>* < 0.01, 0.02 < F<sup>2</sup> = 0.023 < 0.15). Perceived life inconvenience significantly reduced the impact of ascription of responsibility on personal norms, with a "high" effect size (<sup>β</sup> <sup>=</sup> −0.158, *<sup>p</sup>* < 0.01, 0.35 < F<sup>2</sup> = 0.072).

**Figure 5.** Simple slope analysis (PLI\*ACS -ARE).

**Figure 6.** Simple slope analysis (PLI\*ARE -NO).
