Next, a multivariate logistic regression was performed to examine
the main effects of riding with a drinking driver during
adolescence on young adult DUI after the effects of key confounding
factors were taken into account. Pairwise correlation analyses
indicated weak to moderate correlations amongst the predictor
variables (r ranged from −.12 to .49), indicating no signs of multicollinearity
amongst these risk factors. As shown in Table 2,
in the multivariate model, riding with a drinking driver during
adolescence was associated with young adult DUI, increasing the
likelihood by 62%. Other significant risk factors included being
male, older in age, sensation seeking and higher family SES. Males
were two and a half times more likely to DUI than females.