To test the mediation effects of our model, we ran the fully saturated model in SmartPLS per the recommendations by Gefen, Rigdon, and Straub (2011). We included four additional paths examining the direct effects of privacy concern and privacy policy to our dependent variables. Comparing the saturated model (Figure 5) with our hypothesized model (Figure 4), we first can see that all the significant paths in our hypothesized model also remain significant in the saturated model, with the only exception of the path between app engagement and emotional attachment. The only additional path that had direct effect on the dependent variables was privacy policy on emotional attachment.