«by Johnathon P. Ehsani A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Health Behavior ...»
Analytical strategy For each state, the models were estimated using the natural logarithm of the monthly (or quarterly) fatal crash rate per 100,000 population. Using the natural logarithm, the coefficient representing the intervention effects (!) is directly interpretable (using the formula 100 x [e! – 1]) as the percentage change in the post-intervention series relative to the pre-intervention series (McDowall, McCleary et al. 1980). Results presented are based on the models using the natural logarithm of fatal crash rates as the primary outcome variable.
The analyses were conducted in three stages. First, a linear regression model was estimated for the teen driver crash rates and the covariates: adult crash rates, gas prices, and GDL laws. Second, the model for each state was statistically adjusted for trends and seasonal variation. Autoregressive and moving average orders were identified using auto-correlation and partial-auto-correlation functions of the series residuals. Finally, the original regression model was re-estimated with the inclusion of the autoregressive or moving average orders identified in the second stage. Outliers were also detected and controlled for in the final model. Analyses were conducted using the SCA Time Series and Forecasting System, a specialized time-series analysis software package (Scientific Computing Associates 2011).
The fatal crash rates across five year intervals for states that implemented GDL intermediate license requirements (passenger restriction or nighttime driving restriction) independently of other GDL components during the period 1990 to 2009 are presented in Table 3.3. Teen drivers’ fatal crash rates were generally higher than adult drivers’ fatal crash rates for all states for most years, with the exception of Rhode Island. There was considerable variation among states’ teen and adult fatal crash rates, with teen crash rates highest in Nebraska, and lowest in Rhode Island. Adult drivers’ fatal crash rates were highest in North Carolina and lowest in New Hampshire.
Teen drivers’ crash rates were typically highest in 1990, and usually twice as high as adult crash rates in the same state. Both teen and adult crash rates declined over the study period, although the decline was more pronounced among teens, such that by 2009, teen crash rates were lower or comparable to adult crash rates for most states in the study sample.
Passenger Restriction The results of the analysis of the effect of the passenger restriction provided little support for the first hypothesis (Table 3.4). In Rhode Island, the introduction of a law that restricted driving to a single passenger below the age of 21 for the first 12 months of intermediate licensure was followed by a 46% reduction [100 x (e -.6175 – 1) = - 46%] in teen driver passenger fatal crash rates that approached significance (p =.06). This effect was observable when using quarterly, not monthly, fatal crash rates.
There were no significant changes in the overall or passenger fatal crash rates of 16- and 17-year-old drivers corresponding to the introduction of a passenger restriction for the remainder of the states in the sample. Dose-response relationships between the duration of the passenger restriction or the number of passengers allowed and a reduction in teen drivers’ fatal crash rates were not observed. Adult crashes explained some of the variability in teens’ fatal crash rates in Colorado, Missouri, North Carolina and Utah.
Nighttime Driving Restriction Based on the results of the statistical model used for this study, the second hypothesis was rejected. The introduction of nighttime driving restrictions did not result in a reduction of teen drivers’ overall fatal crashes or fatal nighttime crashes in Utah or Nebraska (Table 3.5). Adult crashes explained some of the variability in teens’ fatal crash rates in Utah.
Additional findings In Connecticut, the introduction of the six-month learner license in January 1997 was followed by a significant decline in both the overall teen driver fatal crash rate and the passenger fatal crash rate, declining by 16.6% and 20.4% respectively. In North Carolina, the introduction of a 12 month learner permit and extended nighttime driving restriction in December 1997 was followed by a 15.6% decline in the overall teen driver fatal crash rate, and a 15.5% reduction in the teen driver passenger crash rate, both statistically significant.
The purpose of this study was to examine the independent effects of GDL intermediate restrictions, namely passenger and night restrictions, on teen drivers’ fatal crashes. This purpose was challenging to accomplish for several reasons. Drawing from a potential study population of fifty states, we identified eight instances where a passenger restriction was implemented independently of any other GDL component, creating a natural experiment where intervention effects could potentially be measured.
Each state had existing GDL components in place, however, when the passenger restriction was introduced, meaning we could not estimate the independent effect of that restriction, but instead, only the additive effect of the passenger restriction. We identified no cases where a nighttime driving restriction was introduced as a standalone restriction.
By loosening the inclusion criteria for the evaluation of nighttime driving restrictions, we identified two instances where the restriction was implemented as the first intermediate restriction in a GDL system, although simultaneously with a required number of supervised driving hours in the learner phase. Using this carefully selected sample, we tested the effect of each restriction using an analytical approach that accounts for long terms trends, a technique that has been recommended but infrequently applied in GDL evaluation research (Hartling, Wiebe et al. 2004).
The observed decline in passenger fatal crashes in Rhode Island is notable for being larger than the previously reported effects in other studies (Masten and Hagge 2004; Zwicker, Williams et al. 2006; Chaudhary, Williams et al. 2007). Rhode Island’s passenger restriction was among the most stringent in the study sample in terms of its duration (12 months), and allowance of only a single passenger. Colorado’s passenger restriction was also for 12 months, but allowed no passengers for the first six months and a single passenger for the second six months. Based on epidemiological evidence suggesting that crash risk increases with each additional passenger in the vehicle (Chen, Baker et al. 2000), the crash risk for Rhode Island teen drivers in the first six months of independent licensure should be partially elevated, relative to teen drivers in Colorado. The fact that a significant decline in fatal passenger crashes was not observed in Colorado means we could not identify an association between the number of passengers allowed in a teen driver’s vehicle and a decline in crashes.
With the exception of Rhode Island, these analyses did not demonstrate a decline in teen drivers’ fatal crashes following implementation of a passenger restriction.
This finding contrasts with existing research that has generally reported significant declines in injury and fatal crashes following the introduction of a passenger driving restriction (Masten and Hagge 2004; Rice, Peek-Asa et al. 2004; Cooper, Atkins et al.
2005; Chen, Baker et al. 2006; Zwicker, Williams et al. 2006; Chaudhary, Williams et al.
2007; McCartt, Teoh et al. 2010; Fell, Todd et al. 2011). One explanation may be the presence of additional GDL components that were unaccounted for in previous evaluations. For example, California’s passenger restriction was introduced simultaneously with a 12 midnight to 5 a.m. driving restriction, and a six month learner license period with 50 hours of required supervised driving (Insurance Institute for Highway Safety 2011), making the exact contribution of the passenger restriction difficult to quantify. Another explanation could be suboptimal compliance with or enforcement of a passenger restriction. In Colorado and Utah, the passenger restriction was limited to secondary enforcement, which means that officers could not stop teens for driving with passengers, and could only issue a citation for a passenger restriction violation if a teen driver was stopped for another reason (National Highway Traffic Safety Administration 2006). In four states (Connecticut, Maine, North Carolina and Utah), passenger restrictions do not apply to siblings, which could further complicate enforcement efforts.
Survey research suggests that both novice teen drivers and their parents are less supportive of passenger restrictions than nighttime restrictions (Ferguson and Williams 1996), and this attitude may translate into lower compliance with passenger restrictions (Williams, Leaf et al. 2006).
The importance of identifying and implementing effective policies that reduce passenger fatal crashes cannot be underestimated. Despite the presence of passenger restrictions in 44 states and the District of Columbia, passenger crashes and teens driving with passengers remain a significant problem, according to FARS data. Recent estimates suggest that over 40% of 16- to 17-year-old drivers in fatal crashes were transporting teens with no adult occupant in the vehicle (Williams, Ferguson et al. 2005), and 60% percent of teenage passenger deaths in 2009 occurred in vehicles driven by another teenager (Insurance Institute for Highway Safety 2012). These data suggest that compliance with, and enforcement of, existing passenger restrictions may be inadequate. Approaches to enhancing compliance with passenger restrictions need to balance the epidemiological evidence that every additional peer passenger increases crash risk, with the reality that a complete passenger ban is likely to be violated. Given that one passenger is associated with a modestly higher crash risk (Chen, Baker et al.
2000), one could question whether lower compliance with a complete passenger ban could be offset by higher compliance with a limit of one passenger.
Our hypothesis that the introduction of a nighttime driving restriction would be followed by a decline in teen drivers’ nighttime and overall fatal crashes was not supported. The hours of the nighttime driving restriction in the study sample could provide an explanation for this finding. While the hours between 12 midnight and 6 a.m.
have a highly elevated crash risk, the total number of fatal crashes occurring during that time is small, and may be insufficient to yield either statistically or practically significant changes (McKnight and Peck 2002). The bulk of the nighttime crash problem among novice teen drivers is skewed toward the hours between 9 p.m. and 12 a.m. and not those after midnight (Foss and Goodwin 2003; Insurance Institute for Highway Safety 2012). Currently, in the majority of jurisdictions where a nighttime restriction is in place in the United States, it begins at midnight or later. Several states also allow exemptions for non-recreational driving (work or school-related), and ten states have only secondary enforcement of the nighttime driving restriction (National Highway Traffic Safety Administration 2006). This suggests that reductions in teen drivers’ nighttime crashes in the United States may be possible if existing nighttime driving restrictions began earlier, and the nighttime driving restriction was subject to primary enforcement.
This study identified significant declines in teen drivers’ overall and passenger fatal crash rates in North Carolina following the implementation of a 12-month learner permit and 9 p.m. to 5 a.m. nighttime driving restriction in December 1997. In Connecticut, the introduction of a six-month learner license holding period in January 1997 was followed by a sizeable significant decline in overall and passenger fatal crash rates. While these findings do not address the primary research hypotheses of this study, they suggest that declines in passenger crashes cannot be attributed to passenger restrictions alone, if other GDL components are already in place. In reality, the significant declines in passenger crashes as a result of passenger restrictions reported in previous evaluations were likely the result of synergistic or combined effects from several simultaneously implemented GDL components rather than the effect of passenger restrictions alone.
With the exceptions of studies from North Carolina (Foss 2009) and Utah (Hyde, Cook et al. 2005), this research represents the first state-level evaluations of GDL to be conducted for the states in this study sample. In addition, this study adds to the literature by finding that with exception of a single state, individually implemented passenger and nighttime driving restrictions did not result in reductions in teen drivers’ fatal crashes in the states studied, a finding that contrasts with existing research on the effect of passenger and nighttime driving restrictions (Lin and Fearn 2003; Cooper, Atkins et al.
2005; Chaudhary, Williams et al. 2007; Fell, Todd et al. 2011). It is possible that the declines observed in previous studies are due to the synergy resulting from the combination of multiple GDL components, rather than the passenger or nighttime restriction itself.
The sample for this study was limited to the instances where states introduced an intermediate license driving restriction independently of other GDL components. In every instance that a passenger restriction was implemented in this sample, however, it was introduced within an existing GDL system of varying strength. Consequently, the additive effect of multiple co-existing GDL components could not be disentangled. Further, the findings of this study are based only on fatal crashes involving teen drivers. Fatal crashes represent a small and atypical subset of all crashes, and the etiology of fatal crashes may differ from that of less serious crashes (Lam 2003).
Future research examining the effect of intermediate license restrictions on teen drivers’ should extend the analysis to include injury and property damage only (PDO) as well as fatal crashes. Currently, this data is available in state’s crash records, which include all crashes occurring on public roadways involving a fatality, disabling injury, non-disabling injury, possible injury or property damage valued at $1000 or more.