The second modification of Figure 2 was a consequence of the fact that the distributions of alcohol-related variables are often skewed, which could affect the SEM results. Therefore, the analysis was repeated after using transformations selected based on the topography of the specific item. Thus, inverse transformations were used for the SRE score, weekly drinking quantity and alcohol problems; log transformations were applied for maximum drinks; and square root transformations for the PEER variable. Once again, all significant paths in Figure 2 remained, no new paths were added, and only minor changes in path coefficients were obtained.
