Additionally, supplementary LGMs of respective changes in alcohol involvement and personality that dropped the last point of assessment (i.e., age 35) were also estimated, allowing linear (as opposed to freely estimated) slopes for the constructs of interest to be specified. This further reduces any interpretive ambiguity surrounding the estimates of correlated change between the slopes of alcohol involvement and personality and the relation of the slope parameters to other covariates. The results of these supplementary analyses yielded findings that are entirely consistent with the results presented below, with the exception that they specifically pertain to changes in problematic alcohol involvement and personality modeled linearly from ages 18 to 29 (rather than from 18 to 35 with a freely estimated slope loading at age 35). These supplemental analyses yielded findings that are comparable in pattern, magnitude, and significance level and lead to the same types of inferences, bolstering confidence that findings are not being driven by nonlinear aspects of change observed toward the end of the observation period.3
