The following analyses are based on the sample of 1956 respondents
from 3 grade cohorts described above. To compare the
frequencies and means of exposure and potential confounder variables
in drink driving and non-drink driving groups, chi square
tests for categorical variables and t-tests for continuous variables
were performed. The unadjusted association between riding with
a drinking driver during adolescence and young adult drink driving
was then examined in a bivariate logistic regression. A series
of logistic regressions was also performed to identify associations
between each potential confounder with young adult DUI. A multivariate
logistic regression predicting young adult drink driving was
then performed to test the predictive value of riding with a drinking
driver during adolescence while accounting for the influence of
potential confounding risk factors. To identify differential effects
between subgroups within the sample, interactions between gender,
age and cohort with adolescent exposure to riding with a
drinking driver were also tested. However, none were significant
and so the results are pooled across all participants