Pennsylvania records contained information on 5,690 individuals who were arrested between December 31, 1994, and December 31, 1997, on Use and Lose charges for whom a driver history record could be found. Table 5 describes demographic characteristics and driving history of the study population. Almost all of the persons involved were age 18 or older and were about equally divided among the individual years of age (18, 19 or 20). In Pennsylvania, persons under the age of 18 are usually processed in the juvenile court system. These cases were not available to the study.
The mean age of the subjects was 19.5 years (SE=0.86 years). Almost 90 percent were males and about 10 percent were females. Over two thirds had at least one previous traffic violation (prior to the Use and Lose arrest date), about one in four had a previous motor vehicle crash, and about 40 percent had at least one previous driver's license suspension.
Table 5. Study Population Characteristics, Pennsylvania (Total=5,690)
Almost all (5,607) of the arrest charges were for purchase, consumption, possession or transportation of alcohol (Section 6308). The remaining charges were for misrepresenting age to purchase alcohol or carrying false identification (83).
The driver record file was merged with the PCCD court data. This process revealed that one-half (2,851) of all of the Use and Lose charges had been made coincidental with a DWI arrest that had led to a DWI licensing action. Among these DWI-related cases, approximately one-half also underwent a Use and Lose license suspension. (Such suspensions are applied sequentially.) For the purposes of the analysis, this group of drivers was considered as having a DWI-related license suspension.
Among the remaining cases, 1,821 had no license action associated with the input arrest, 784 had undergone only a Use and Lose Suspension, and 234 had undergone a license action for another reason such as point count. Table 6 presents the summary statistics for these four groups of driver's license actions.
Subsequent driving behaviors of these individuals are summarized in Table 7. About 60 percent of the study population committed at least one subsequent traffic violation and about one fifth were involved in at least one subsequent motor vehicle crash.
Table 6. Four Driver's License Action Groups Following Input Arrest, Pennsylvania (Total=5,690)
Table 7. Distribution of Subsequent Violations, Crashes, and License Actions, Pennsylvania (Total=5,690)
Bivariate and Multivariate Analyses for the Probability of A Subsequent Traffic Violation
Age, gender, and driver's license action group after the input arrest were each evaluated separately and together in association with the odds of having a subsequent violation. Bivariate analyses involving each of the three population characteristics showed significant unadjusted associations with the odds of committing a subsequent traffic violation. Being younger or male or not having any license action for the input arrest were significantly associated with the higher odds of committing a subsequent violation. Compared to females, males were more likely (odds ratio [OR] =2.0, 95% CI, 1.68-2.37), and for every increase of age increment, older subjects were 0.76 times as likely as younger subjects to have a subsequent violation (95% CI, 0.72-0.81). For example, for an age of 18.8 (25th percentile) vs. 20.3 (75th percentile), the OR=1.51 (95% CI, 1.37-1.64). Compared to the group that had no driver's license actions, those who had a DWI- related action were the least likely (OR=0.50, 95% CI, 0.44-0.56), and those who had a Use/Lose-related or other license action were also less likely to have a subsequent traffic violation (OR=0.60, 95% CI, 0.52-0.73; OR=0.68, 95% CI, 0.52-0.91).
The adjusted associations between age, gender, driver's license action group, and the incidence of a subsequent traffic violation confirmed the results obtained from the bivariate analyses. For every increment of age increase, older subjects were 0.78 times as likely as younger subjects (95% CI, 0.73-0.83). For example, for subjects 18.8 years old (25th percentile) vs. 20.3 years old (75th percentile), OR=1.45 (95% CI, 1.32-1.60); males were more likely than females (OR=1.90, 95% CI, 1.60-2.27); the group with a DWI-related action was the least likely (OR=0.52, 95% CI, 0.46-0.60), and the groups with a Use/Lose-related or other license action were less likely than the one that did not have any driver's license action for the input arrest to commit a subsequent traffic violation (OR=0.61, 95% CI, 0.51-0.73; OR=0.65, 95% CI, 0.49- 0.88). Table 8 summarizes crude and adjusted associations between each of the independent variables and the outcome variable.
Table 8. Associations Between Age or Gender or Driver's License Action Group with a Subsequent Violation, Pennsylvania (Total=5,690)
Bivariate and Multivariate Analyses for the Probability of A Subsequent Crash
Unadjusted odds ratios of having a subsequent crash showed that having a license action for the input arrest was negatively associated with the probability of having a subsequent crash. Those individuals who had a driver's license action suspension were less likely than those who did not to be involved in a subsequent motor vehicle crash: for a Use/Lose-related driver's license suspension OR=0.63 (95% CI, 0.50-0.79), for those who had a DWI-related suspension OR=0.76 (95% CI, 0.66-0.88), and for those who had another license action OR=0.70 (95% CI, 0.48-1.01). However, the difference based on age or gender was not found to be statistically significant (p>0.050).
Multivariate analyses further confirmed that only the type of a driver's license action received for the input arrest statistically affected the odds of being involved in a subsequent 1st crash. Compared to the offenders who did not have any license actions, those who had a Use/Lose-related action were the least likely and those with a DWI-related license action were less likely to be involved in a subsequent crash (OR=0.64, 95% CI, 0.50-0.80; OR=0.79, 95% CI, 0.68-0.92). This difference between those who had other license actions and those who did not have any was marginally significant (OR=0.69, 95% CI, 0.47-1.01). Also, the difference in the incidence of a subsequent crash was not statistically significant based on age and gender (p>0.050). Since the parameter estimates of the significant predictor, "license action group", changed by less than 30 percent when age and gender were taken out of the model, the latter were not confounders and were not retained in the model. Table 9 presents summary statistics for the predictors of having a subsequent crash.
Table 9. Associations Between Age or Gender or Driver's License Action Group with a Subsequent Crash, Pennsylvania (N=5,690)