weakly significantp < 0.1
For models with at least a weakly significant marginal effect (i.e., p < 0.1), we then assessed the significance of the odds ratio for each pair of levels of each predictor by Wald tests and Wald confidence intervals.
For models with at least a weakly significant marginal effect (i.e., p < 0.1), we then assessed the significance of the odds ratio for each pair of levels of each predictor by Wald tests and Wald confidence intervals.
In contrast, those reliant on mass media tended an opposite shift in attitude, although this effect was relatively minor and carried marginal significance, as indicated by the corresponding odds ratios (see Table 4 ).