If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Center for Injury Research and PreventionDepartment of Pediatrics, Division of Emergency Medicine, the Children’s Hospital of PhiladelphiaCenter for Clinical Epidemiology and BiostatisticsPerelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania
New Jersey (NJ) implemented the first Graduated Driver Licensing (GDL) decal provision in the U.S. in May 2010. An initial study reported a 1-year post-decal decrease in the crash rate among NJ intermediate drivers aged <21 years. Longer-term analysis is critical for policymakers in other states considering whether to implement a decal provision.
To evaluate the longer-term (2-year) effect of NJ’s decal provision on overall and age-specific crash rates of young drivers with intermediate licenses.
Monthly per-driver police-reported crash rates during January 2006–June 2012 were estimated. Specific crash types included injury, midnight–4:59am, single-vehicle, multiple-vehicle, and peer passenger crashes. Negative binomial modeling compared pre- versus post-decal crash rates, adjusting for age, gender, calendar month, gas price, and 21- to 24-year-old licensed driver crash rates; piecewise negative binomial regression models accounted for pre-decal crash trends among intermediate drivers. Analyses were conducted in 2013.
The adjusted crash rate for intermediate drivers was 9.5% lower in the 2-year post-decal period than the 4-year pre-decal period (95% CI=0.88, 0.93). Crash rates decreased 1.8% per year before the provision and 7.9% per year in the post-decal period (p<0.001 for difference in slopes). For several crash types, effects appeared to be particularly strong for 18- and 19-year-olds. An estimated 3,197 intermediate drivers had crashes prevented.
NJ’s decal provision was associated with a sustained decline in intermediate driver crashes. Future research should aim to better understand the causal mechanism by which NJ’s decal provision may have exerted an effect.
Motor vehicle crashes are the leading cause of death and acquired injury among U.S. adolescents.
To address this public health problem, U.S. states have implemented Graduated Driver Licensing (GDL) systems, including an “intermediate” licensure phase in which novice drivers may drive without adult supervision but are restricted from engaging in higher-risk behaviors such as carrying multiple passengers and nighttime driving. Evaluations of GDL and passenger and night restrictions have consistently shown crash reductions among 16- and 17-year-old drivers.
However, the effectiveness of GDL restrictions may be limited by the inability of police officers to enforce them—as restricted drivers can only be identified via traffic stop and license inspection—and teens’ willingness to comply with them.
Enhancing the impact of GDL on reducing crashes requires efforts to increase both compliance with and enforcement of restrictions.
Modeled after international jurisdictions, New Jersey (NJ) implemented the first GDL decal provision in the U.S. in May 2010, requiring all drivers aged <21 years with a learner’s permit or intermediate license to display highly visible decals on their vehicles. By making such drivers easily identifiable to police, the provision is hypothesized to facilitate enforcement of GDL restrictions and increase the likelihood that novice drivers avoid higher-risk driving behaviors, thereby reducing their crash risk. Previous to NJ’s implementation, GDL decal provisions had not been formally evaluated.
The provision’s initial implementation was accompanied by significant media attention and required use of the decal even if young drivers entered the GDL system prior to implementation (i.e., no “grandfathering”). These factors may have contributed to initial estimates of the provision’s impact. Evaluation of whether the reduction in crash rates can be sustained after media attention has ceased and as a new cohort of teens enters the licensing process is critical to guide policymakers in other states as they consider decal provisions. The objective of this study was to evaluate the longer-term (2-year) effect of NJ’s decal provision on the police-reported crash rate among young intermediate drivers. In addition, effects on important crash types—injury, post-midnight, single-vehicle, multiple-vehicle, and peer passenger crashes—were estimated overall and by age.
New Jersey’s Graduated Driver Licensing System
New Jersey has one of the most comprehensive GDL systems in the U.S. and the highest minimum licensure age of 17 years.
Adolescents progress through three licensing phases: (1) learner’s permit, eligible at a minimum age of 16 years and a 180-day minimum holding period; (2) intermediate (known as probationary in NJ) license, eligible at a minimum age of 17 years, a 365-day minimum holding period, and subject to the following restrictions: (a) one-passenger limit unless a parent/guardian is in the vehicle, (b) ban on driving between 11:01pm and 4:59am, (c) ban on driver use of hand-held and hands-free interactive wireless communication devices, and (d) required seat belt use for all vehicle occupants; and (3) unrestricted (basic) license, eligible at a minimum age of 18 years following completion of Phases 1 and 2. NJ is the only state that applies full GDL rules to all newly licensed drivers aged <21 years; in other states, new drivers aged ≥18 years are exempt from GDL restrictions.
Young drivers are required to purchase (US$4) a pair of red, reflectorized decals before obtaining a permit and display them on both license plates; the fine for failing to do so is US$100. Two concurrent changes occurred in NJ’s GDL system: (1) a change in the start of the night restriction from midnight to 11:01pm; and (2) a limit of one family member passenger unless accompanied by a parent/guardian (previously, immediate family members were exempt from passenger limits). Given that intensive outreach efforts began in February and decals were available for purchase in April, February through April 2010 were excluded a priori. Thus, the pre-decal period was defined as January 2006 through January 2010 and the post-decal period as May 2010 through June 2012.
Data from two administrative sources—the NJ Motor Vehicle Commission licensing database and the NJ Department of Transportation crash database—were linked. These databases were previously described in detail.
The licensing database included the date each NJ driver obtained a learner’s permit and intermediate license through June 2012. The end of the intermediate phase was defined as the earliest transaction indicating a license upgrade that occurred at least 365 days after the start of the intermediate period. Periods of license suspension and date of death were also identified, if applicable. The two databases were linked via a hierarchical deterministic linkage, and 98.4% of crash-involved NJ drivers aged <21 years matched to a unique licensing record.
The primary outcome was the monthly rate of police-reported crashes among intermediate drivers aged <21 years; the numerator was the number of drivers who experienced a crash in a specific month and the denominator was the number of drivers on the 15th day of that month. Drivers whose licenses were suspended, who were unlicensed, or who were deceased on the 15th were not included in that month’s rates. The crash rate for 21- to 24-year-old licensed drivers was also estimated to account for general NJ crash trends in statistical models. This age group is likely to be the most similar to young drivers in their susceptibility to changes in driving patterns owing to contextual factors such as fuel costs and economic conditions. The daily average price of unleaded gas in NJ was obtained from the AAA Fuel Gauge Report; gas price trends have previously been found to be a useful proxy for changes in driving exposure among young drivers and in these analyses were intended to reflect month-to-month changes in driving exposure due to economic factors.
Crash subtypes included crashes in which there was a moderate or greater severity injury (as noted on the crash report); crashes that occurred from midnight through 4:59am; single-vehicle crashes; multiple-vehicle crashes; and crashes in which the driver was carrying only passengers aged 14–20 years (i.e., peer passengers). To further explore the potential contribution of the night restriction change, analyses of crashes occurring during 11:01pm–11:59pm—the time period affected by the night restriction change—and those occurring outside the relevant period (5:00am–11:00pm) were also conducted.
Average monthly police-reported crash rates per 10,000 drivers were estimated for the pre- and post-decal periods. To accommodate overdispersion in the monthly counts of crashes, a negative binomial model was used. Indicators included time period (pre- versus post-decal); gender; age (ref 21- to 24-year-old licensed drivers versus 17-, 18-, 19-, and 20-year-old drivers); calendar month; and monthly average unleaded gas price; furthermore, the mid-month number of intermediate drivers was specified as the offset to estimate per-driver monthly crash rates.
Interactions between age and time period provided a “difference-in-differences” crash rate ratio that adjusted the pre- versus post-decal rate ratio among intermediate drivers for the analogous rate ratio among 21- to 24-year-old drivers.
Although informative, an overall pre–post model does not fully account for crash trends among intermediate drivers that existed prior to the law’s effective date. Thus, piecewise negative binomial regression models were also conducted among intermediate drivers for all crash types.
The slope of the regression line for the natural logarithm of estimated crash rates over the 4-year pre-decal period provided a baseline trend in intermediate driver crash rates before implementation. The slope was allowed to differ in the pre- and post-decal period, and a significant decrease in the slope coinciding with the provision’s effective date (after adjustment for potential confounders) would be consistent with an effect of the intervention. The use of a quadratic term to reflect potential non-linearity of post-decal slopes was explored, but it was concluded that slopes were well approximated as linear. Models were adjusted for gender; age (17–20 years); calendar month; and monthly average unleaded gas price. If the overall model showed a significant effect at the p=0.05 level, separate age-specific models were created. No evidence of effect modification by gender was found, as determined by likelihood ratio tests.
Finally, the number of intermediate drivers whose crashes were prevented in the post-decal period was estimated by subtracting the estimated number of crash-involved intermediate drivers in each post-decal month from the number of intermediate drivers who would have been predicted to be in crashes in that month under pre-decal conditions (i.e., counterfactual condition) and summing across months (∑post-decal months [rate predicted under pre-decal conditions × number of intermediate drivers] – [estimated post-decal rate × number of intermediate drivers]). Analyses were conducted in 2013 using SAS, version 9.3. This study was approved by the Children’s Hospital of Philadelphia’s IRB.
Over the study period, there was an average of 171,433 intermediate drivers per month. Overall, the observed average monthly rate of police-reported crashes among intermediate drivers decreased from 142.6 per 10,000 drivers in the pre-decal period to 126.5 in the post-decal period. After adjusting for gender, calendar month, age, and average gas price, the post- versus pre-decal rate ratio was 0.88 (95% CI=0.86, 0.89) for intermediate drivers and 0.97 (0.95, 0.99) for 21- to 24-year-old drivers. Upon further accounting for 21- to 24-year-old crash rates, the adjusted crash rate for intermediate drivers was 9.5% lower in the post-decal period (0.88, 0.93) than in the 4-year pre-decal period. The extent of reduction (after fully adjusting) varied by age: 13.3% lower among 17-year-olds (0.84, 0.90); 8.3% lower among 18-year-olds (0.88, 0.95); 8.5% lower among 19-year-olds (0.88, 0.95); and 4.1% lower among 20-year-olds (0.92, 1.00).
Results of piecewise regression models for intermediate drivers are shown in Table 1. Intermediate drivers’ adjusted crash rate decreased an estimated 1.8% per year (95% CI=2.4% decrease, 1.2% decrease) before the decal, whereas the rate of decrease over the 2 years post-decal was 7.9% per year (9.6% decrease, 6.1% decrease) (p<0.001 for difference in slopes). Monthly estimated crash rates, along with the counterfactual rate that would have been expected post-decal had the provision not been implemented, are shown for 18-year-old male drivers in Figure 1 (female trends were identical but at a lower rate). After accounting for existing trends, the provision did not appear to have an effect on injury and post-midnight crash rates among intermediate drivers (Table 1). Conversely, there were decreases in the slope of the crash rate for single-vehicle, multiple-vehicle, and peer passenger crashes. For these crash types, the provision appears to have had a particularly strong effect on 18- and 19-year-old drivers. For example, although there was no reduction in the single-vehicle crash rate of 18- and 19-year-old drivers during the 4-year pre-decal period, rates decreased 13.4% per year (19.7% decrease, 6.7% decrease) among 18-year-olds and 16.8% per year (24.1% decrease, 8.8% decrease) among 19-year-olds after implementation. Results of piecewise regression models for single-vehicle crashes of 18-year-old male drivers are shown in Figure 2.
Table 1Annual Percent Change (95% CI) in Adjusted Crash Rates (per 10,000 Intermediate drivers) and Ratio of Post- to Pre-decal Slopes
Overall, 2.5% of all intermediate drivers’ crashes occurred during 11:01pm–11:59pm. The rate of crashes occurring in the 11 o’clock hour decreased 6.7% per year (9.3% decrease, 4.0% decrease) in the pre-decal period and 16.5% (23.6% decrease, 8.7% decrease) in the post-decal period (p=0.04 for difference in slopes), whereas the rate of crashes occurring during 5:00am–11:00pm decreased 1.5% per year (2.1% decrease, 0.8% decrease) in the pre-decal period and 7.5% (9.3% decrease, 5.7% decrease) in the post-decal period (p<0.001 for difference in slopes).
Finally, it was estimated that crash involvement of 3,197 intermediate drivers was prevented in the 2 years following the provision, including 1,014 drivers aged 17 years, 1,220 drivers aged 18 years, 675 drivers aged 19 years, and 288 drivers aged 20 years. A conservative estimate, which excluded 138 drivers whose crash involvement during the 11 o’clock hour was prevented, is 3,059 drivers.
Extending findings reported in an initial evaluation,
these results indicate that there was a 2-year sustained reduction in the crash rate among young intermediate drivers after introduction of NJ’s decal provision. Their overall adjusted crash rate was 9.5% lower in the post-decal period than in the 4 years prior (relative to 21- to 24-year-old licensed drivers), and they experienced a significantly greater rate of crash reduction after implementation. For several crash types, associations were particularly strong for 18- and 19-year-olds. In light of some recent national GDL studies reporting increased crash rates among 18- and 19-year-olds in states with strong GDL laws,
these findings may speak to a possible mechanism of mitigating crashes among older teen drivers.
Results of this study should be interpreted in light of the hypothesized mechanisms by which decals would result in a crash reduction—police officers’ enforcement of GDL restrictions and teens’ compliance with restrictions and other traffic safety laws. There is limited information available about these specific mechanisms. Our previous analysis indicated that the GDL citation rate increased among intermediate drivers in the post-implementation period, but the increase appeared to be concentrated in the few months after implementation.
Findings of the current study—no additional reduction in post-midnight crashes after the decal’s implementation, few crashes during the 11 o’clock hour, and a reduction in daytime crashes similar to that of all crashes—suggest that the decal did not exert an effect primarily through increased compliance with NJ’s night restriction. We are conducting analyses to further evaluate the provision’s effect on enforcement and compliance.
Self-reported compliance with the decal provision itself is less than ideal—only about 40% of intermediate drivers indicated they “always” display their decal.
concluding that the provision does not jeopardize the safety of teen drivers. It is important to note that display of the decal is not necessarily in the critical pathway to crash reduction. Intermediate drivers may be motivated to drive more safely and comply with GDL restrictions and other traffic laws even when not displaying the decal to avoid being cited for violation of GDL provisions. As previously theorized,
decal provisions may also encourage safer driving behaviors among young drivers (beyond restricted behaviors) and among other drivers sharing the road with them.
A particular strength of the current study is the incorporation of licensing data, which by design accounts for monthly changes in the number of intermediate drivers and allows for rate estimates among intermediate drivers as opposed to relying on age as a proxy. Additional enhancements to prior analyses include the establishment of longer-term pre-decal trends, adjustment for gas prices, comparison to a younger cohort of older licensed drivers, and age-specific analyses.
The study also has several potential limitations. The ability to make causal inferences—and conclude that the post-decal decrease was attributed to the decal provision—depends on the comparability of pre- and post-decal intermediate driver groups and adequate control of confounders,
which include factors relevant to intermediate drivers that may have changed over the study period and the two concurrent changes in New Jersey’s GDL system. There were no other state-wide teen driver initiatives introduced during the post-decal period that would explain the large decline. We have attempted to control for variations in driving exposure due to seasonal effects and economic conditions by including calendar month and mean price of unleaded gas in NJ in the regression models.
The recent economic recession may have also affected crash rates via a reduction in driving. However, given that the bulk of the recession in the U.S. was in the pre-decal period (2007–2009), one might expect its effect on driving exposure—and thus crash reduction—in the pre-decal period to be greater than or similar to its effect in the post-decal period.
Finally, there were concurrent changes in NJ’s passenger and night restrictions that may have contributed to changes in crash rates above and beyond the effect of the decal provision. Although we were not able to account for the effect of the family member exemption, analyses of crashes by time of day demonstrated a significant decline in crashes apart from the expanded night restriction. We believe that we have taken into account the primary crash-related factors that may have differed in the pre- and post-decal periods; however, it is still possible that unobserved, unknown, or hard-to-measure confounders may bias the estimates of effect. Further, we cannot rule out the possibility that the decline was at least partly due to other aspects of NJ’s GDL revision (e.g., continual increase in awareness around teen driving safety in the post-decal period). Indeed, the estimated crash reduction in this study is somewhat larger than what might be expected based on previous GDL evaluations, strongly suggesting the need for additional research—both in NJ and future implementations—to better understand the complex role decal provisions may have in reducing the burden of teen driver crashes.
This study provides initial evidence that NJ’s introduction of a decal provision was associated with a sustained 2-year decline in the crash rate among intermediate drivers. However, NJ’s licensing system is unique in several ways; thus, it is unclear the extent to which effects observed in NJ would be realized in other states. Future research should aim to better understand the causal mechanism by which NJ’s decal provision may have exerted an effect; continue to track decal-related attitudes and behaviors of NJ novice drivers, their parents, and law enforcement; and monitor the experiences of several states experimenting with voluntary introduction of decals.
States that are currently considering implementing a decal provision should aim to introduce legislation in a manner acceptable to families and establish a clear evaluation strategy that begins well before the point of implementation.
The authors thank Rachel Myers, MS, Suzanne Hill, BA, and Christine Norris, BA, at the Center for Injury Research and Prevention for their work on the project and assistance with this manuscript; and Pam Fischer, MLPA, and Violet Marrero for their valuable insights on the implementation of NJ’s decal provision. The authors also thank the NJ Motor Vehicle Commission, the NJ Office of Information Technology, and the AAA New Jersey Automobile Club for assistance in providing data.
The study was funded by State Farm Mutual Automobile Insurance Company (State Farm®). State Farm reviewed this manuscript and supported the decision to submit the manuscript for publication but did not have a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or in manuscript preparation. The findings and conclusions are those of the authors and do not necessarily represent the views of State Farm. Allison E. Curry, Melissa R. Pfeiffer, and Konny H. Kim have no conflicts of interest to disclose. Michael R. Elliott and Dennis R. Durbin have received consulting fees from State Farm, the company that funded this study, for different projects.
Allison E. Curry has received grants from the AAA Foundation for Traffic Safety and the Insurance Institute for Highway Safety, stakeholders in this research topic. Michael R. Elliott has received grants from the AAA Foundation for Traffic Safety, a stakeholder in this research topic. Dennis R. Durbin has received grants from the Insurance Institute for Highway Safety and the Toyota Motor Company, North America, stakeholders in this research topic. No financial disclosures were reported by the remaining authors (Pfeiffer, Kim).