Because the correlations between the exogenous variables in our model are comparatively high, there is a potential risk that our results may not be stable with small changes in the data(Cohen et al. 2003). Thus, we reanalyzed our model with 20 different datasets, in which 5% of the cases had been removed randomly(Homburg et al. 2007). The results from the stability tests strongly suggest that the correlations between the exogenous variables do not compromise the validity of our results. For all datasets, the pattern of results is consistent with our hypotheses. In addition, we assessed the stability of the results by splittingthe entire sample into previous users of mobile payment services(n1 = 583) and previous non-users of mobile payment services(n2 = 864). All hypothesized relationships were supported in both subsamples (p 6 .05), which further underlines the robustness of our results.Finally, we assessed the issue of multicollinearity by looking at the magnitude of the bivariate correlations between the exogenous variables (see Table 3). All correlations are far below the common cutoff value of .8 (Berry and Feldman 1985). Together with the results obtained from the stability tests, this indicates that multicollinearity is not a problem in our study.