Advertisement

Crash Risk and Risky Driving Behavior Among Adolescents During Learner and Independent Driving Periods

      Abstract

      Purpose

      Novice adolescents' crash rates are highly elevated early in licensure, despite substantial practice driving during the learner period. The objectives of this study were to examine the variability in measures of driving risk among adolescents during the learner and early independent driving periods and evaluate how risk varies by driving experience, gender, time of day, and road surface conditions.

      Methods

      Objective driving data were collected in a naturalistic cohort study of 90 adolescent drivers with learner driving permit and 131 experienced adult drivers. Participants’ private vehicles were equipped with data acquisition system documenting driving kinematics, miles driven, and video recordings of the driver and the driving environment. Crash/near-crash (CNC) and kinematic risky driving (KRD) rates were calculated during the learner and early independent driving periods by gender (female/male), time of day (day/night), and road surface conditions (wet/dry) for adolescents and adults.

      Results

      CNC and KRD rates of adolescents were similar to adult drivers during the learner period (CNC: incident rate ratio [IRR] = 1.67, confidence interval [CI] = .98–2.82 and KRD: IRR = 1.04, CI = .78–1.40, respectively), but dramatically higher in the first year of independent driving (CNC: IRR = 6.51, CI = 4.03–10.51 and KRD: IRR = 3.95, CI = 2.96–5.26, respectively), and particularly elevated the first 3 months of licensure. Adolescent KRD rates were higher for males than females and invariably higher than adult rates during day and night, wet and dry conditions.

      Conclusions

      While the learner driving period was relatively safe for adolescents, the transition to independent driving was typified by a dramatic increase in risk among adolescents that was higher than adult rates overall and under varying driving conditions.

      Keywords

      See Related Editorial on p. 521
      Implications and Contribution
      Adolescents’ early independent driving was associated with dramatic increase in risky driving behaviors compared to the learner period and to adult experienced drivers. Findings highlight the possible benefits of purposefully maintaining some adult presence in the vehicle during early stages of independent driving.
      Adolescent drivers have disproportionately elevated crash rates compared to other driver age groups, representing 6% of U.S. licensed drivers and 9% of fatal crashes [
      National Highway Traffic Safety Administration (NHTSA)
      Traffic safety facts 2014: Young drivers.
      ]. With the exception of the oldest age group, adolescent drivers also have the highest crash rates per mile driven compared to other age groups [
      National Highway Traffic Safety Administration (NHTSA)
      Traffic safety facts 2013: Young drivers.
      ].
      In recognition of the high crash rates of novice adolescent drivers, graduated driver licensing (GDL) policies have been adopted by nearly all U.S. states. GDL policies consist of learner, provisional licensure, and full licensure stages, which aim to decrease crash risk among novice adolescent drivers by requiring extended supervised practice and limiting exposure to high-risk conditions, whereas adolescents gain independent driving experience and develop safe driving skills. Most evaluations have concluded that GDL policies, particularly those with the strictest provisions, reduce crash rates of novice adolescent drivers [
      • Chen L.-H.
      • Baker S.P.
      • Li. G.
      Graduated driver licensing programs and fatal crashes of 16-year-old drivers: A national evaluation.
      ]. However, crash rates remain high among novice adolescent drivers even in states with strong GDL policies [
      • Curry A.E.
      • et al.
      Comparison of older and younger novice driver crash rates: Informing the need for extended graduated driver licensing restrictions.
      ]. At present, there is no evidence that the amount and quality of the learner period, in which adolescents can drive only with supervision of an adult driver, are associated with independent driving safety [
      • Ehsani J.P.
      • Bingham C.R.
      • Shope J.T.
      The effect of the learner license Graduated Driver Licensing components on teen drivers’ crashes.
      ].
      Presumably, extensive supervised practice driving, as required by GDL policies would prepare adolescents to drive more safely once licensed. However, analyses of crash records indicate that the crash rates of adolescents in the early months of driving are disproportionately elevated [
      • Curry A.E.
      • et al.
      Comparison of older and younger novice driver crash rates: Informing the need for extended graduated driver licensing restrictions.
      ,
      • Masten S.V.
      • Foss R.D.
      • Marshall S.W.
      Graduated driver licensing and fatal crashes involving 16-to 19-year-old drivers.
      ]. Previous studies on crash records from the mid-1990s and early 2000 reported low crash rates during learner driving period, but much higher rates immediately after licensure that declined over the first year of independent driving [
      • Mayhew D.R.
      • Simpson H.M.
      • Pak A.
      Changes in collision rates among novice drivers during the first months of driving.
      ,
      • Lewis-Evans B.
      Crash involvement during the different phases of the New Zealand Graduated Driver Licensing System (GDLS).
      ]. Using U.S. survey data, McCartt et al. described a similar pattern of high crash rates right after licensure, which declined over time [
      • McCartt A.T.
      • Shabanova V.I.
      • Leaf W.A.
      Driving experience, crashes and traffic citations of teenage beginning drivers.
      ]. Inexperience and young age both may contribute to the elevated crash risk of adolescents. Curry et al. found that crash rates were high early in licensure, regardless of the age at licensure, but highest for the youngest novices [
      • Curry A.E.
      • et al.
      Comparison of older and younger novice driver crash rates: Informing the need for extended graduated driver licensing restrictions.
      ]. In a naturalistic driving study of 42 licensed drivers at the age of 16–17 years old, crash and near-crash (CNC) rates were high during the first 18 months of licensure relative to experienced adults (parents) driving the same vehicles, on similar roads and driving conditions [
      • Simons-Morton B.G.
      • et al.
      Crash and risky driving involvement among novice adolescent drivers and their parents.
      ]. Of course, crash rates were higher for some novices than others [
      • Guo F.
      • Simons-Morton B.G.
      • Klauer S.E.
      • et al.
      Variability in crash and near-crash risk among novice teenage drivers: A naturalistic study.
      ,
      • Shope J.T.
      • Bingham C.R.
      Teen driving: Motor-vehicle crashes and factors that contribute.
      ] and rates among adolescent drivers vary according to driver gender [
      • Williams A.F.
      Teenage drivers: Patterns of risk.
      ], time of day [
      • Bingham Cnd.
      • et al.
      Do as I say, not as I do: Distracted driving behavior of teens and their parents.
      ,
      • Gershon P.
      • O'Brien F.
      • Zhu C.
      • Simons-Morton B.G.
      Multi-level predictors of teenage risky driving.
      ,
      • Goodwin A.H.
      • Foss R.D.
      • Harrell S.S.
      • O'Brien N.P.
      Distracted driving among newly licensed teen drivers.
      ,
      • Rajaratnam S.M.W.
      • et al.
      Teen crashes declined after Massachusetts raised penalties for graduated licensing law restricting night driving.
      ], road conditions [
      • Carney C.
      • McGehee D.V.
      • Harland K.
      • et al.
      Using naturalistic driving data to assess the prevalence of environmental factors and driver behaviors in teen driver crashes.
      ], and possibly other factors. Novice adolescent drivers tend to drive in a relatively risky manner that may be due to inexperience, resulting from skill and/or judgment deficiencies, experimentation, and/or general risk-taking propensity [
      • Williams A.F.
      Teenage drivers: Patterns of risk.
      ]. The best evidence of this behavior is from naturalistic driving study measures of elevated gravitational force event rates (i.e., speeding, rapid acceleration, sudden deceleration, swerves, and hard turns), termed kinematic risky driving (KRD). Previous naturalistic driving research found that KRD rates were substantially higher among young drivers compared to older experienced drivers [
      • Simons-Morton B.G.
      • et al.
      Crash and risky driving involvement among novice adolescent drivers and their parents.
      ], and prospectively associated with crash risk [
      • Simons-Morton B.G.
      • et al.
      Do elevated gravitational-force events while driving predict crashes and near crashes?.
      ]. Several studies have reported relatively higher KRD rates among adolescent males than females [
      • Prato C.G.
      • et al.
      Modeling the behavior of novice young drivers during the first year after licensure.
      ,
      • Gershon P.
      • O'Brien F.
      • Zhu C.
      • Simons-Morton B.G.
      Multi-level predictors of teenage risky driving.
      ], consistent with data on self-reported risky driving [
      • Rhodes N.
      • Pivik K.
      Age and gender differences in risky driving: The roles of positive affect and risk perception.
      ]. KRD is conveniently measured in naturalistic driving research and serves as an objective measure of risky driving behavior and a logical intervention target for reducing novice adolescent crash risk [
      • Carney C.
      • et al.
      Using an event-triggered video intervention system to expand the supervised learning of newly licensed adolescent drivers.
      ].
      The literature to date is limited by the lack of objective longitudinal data on driving performance, measured by KRD and CNC rates, during the learner driving period and the transition to independent driving. The purpose of this research is to examine the variability in CNC and KRD rates among novice adolescent drivers relative to experienced adult drivers and with respect to driver gender, time of day, and road surface conditions during the learner and early independent driving periods.

      Methods

      Participants

      This study included a total of 90 novice adolescent participants with learner driving permit, and 131 of their parents (i.e., the experienced drivers' group) recruited in local newspapers and high schools in southwestern Virginia. All adolescent participants were recruited soon after obtaining a learner permit, and all adults held unrestricted licensure. Adolescent participants received $800 for completing the study, paid to them in installments as they completed key milestones. Identical twins and adolescents with diagnosed attention deficit hyperactivity disorder were excluded from the study. No other selection criteria were applied. Parental consent and adolescents assent were obtained, and all procedures followed an approved institutional human subject protocol (see Ehsani et al. [
      • Ehsani J.P.
      • Li K.
      • Grant B.J.
      • et al.
      Factors influencing learner permit duration.
      ] for detailed description of participants' recruitment and screening).

      Vehicle instrumentation

      Participants’ private vehicles were equipped with Data Acquisition System developed at the Virginia Tech Transportation Institute (Blacksburg, Virginia). The Data Acquisition System included the following: (i) multiple sensors to record vehicle speed, multiaxis acceleration, lane position, and Global Positioning System data; and (ii) video cameras to monitor the driver's face, hands, and body position, driver's forward and rear views, and the car dashboard. Data were collected from 2010 to 2014, for a period of up to 21 months, which included a minimum of 9 months of learner driving and up to 12 months of independent driving. Data were collected continuously and were downloaded periodically. During postprocessing, the data were reduced and then coded by trained staff viewing the video and kinematic data.

      Measures

      Crash/near-crash rates

      Near-crash events were defined as any situation that required a last moment maneuver to avoid a crash. Crash events were defined as physical contact between the driver's vehicle and another object. The threshold value used to identify possible CNC events were gravitational units ≥.65 g. CNC rates were calculated for each 3-month period (or quarter) per 10,000 miles. Research indicates that near crashes serve as reasonable proxies for crashes [
      • Guo F.
      • Simons-Morton B.G.
      • Klauer S.E.
      • et al.
      Variability in crash and near-crash risk among novice teenage drivers: A naturalistic study.
      ]. Accordingly, due to low rates of police-reported crashes in naturalistic studies, crashes and near crashes are commonly combined to form a stable measure of driving risk.

      Kinematic risky driving

      G-force events were classified as KRD based on the following thresholds: (i) longitudinal acceleration (≥.30 g, rapid acceleration); (ii) longitudinal deceleration (≤−.45 g, sudden brake); (iii) lateral negative turn (≤−.50 g, hard left turns); (iv) lateral positive turn (≥.50 g, hard right turns); and (v) yaw (±6° per second, angular velocity). KRD rates were calculated for each quarter per 1,000 miles.

      Environmental measures

      The following factors were coded for each driver and each trip: road surface conditions (dry/wet); and time of day (day/night), which were determined by the recorded times of sunrise and sunset of the day the trip occurred.

      Statistical analysis

      Mixed-effects Poisson regression models with a driver-specific random intercept were used to calculate CNC and KRD incident rates (IRs). The logarithm of mileage driven by each participant was added to the models as an offset to adjust for the variability in miles driven. We also assessed the three-way interactions among the driver type, time, and each of the following factors: gender, time of day, and road surface conditions. IRs for adolescents, adults, and time periods were compared as incident rate ratios (IRRs). We defined statistical significance at the level of 95% confidence interval (CI).

      Results

      Sample characteristics

      The novice adolescent drivers' group included 49 females and 41 males with an average age of 15.6 years (SD = .2) who drove an average of 5445 miles (SD = 3829), a total of ∼490,000 miles. A total of 217 CNCs (148 near crashes and 69 crashes with 9 police-reported crashes), 2.4 (SD = 3.0) per driver, and 18,378 KRD events, 107.7 (SD = 128.2) per driver, were documented. The adult drivers' group consisted of 63% female drivers, mean age 47.5 (SD = 6.3), with mean driving experience of 31.3 years (SD = 6.4), who drove a total of 1,027,357 miles in the study vehicles, twice that of the adolescent participants. Adult drivers had 112 CNCs (84 near crashes and 28 crashes with 2 police-reported crashes), an average of 1.2 (SD = 1.7) per driver, and 5,272 KRD events with mean of 58.6 (SD = 70.7).

      Crash and near-crash rates

      As shown in Figure 1, CNC rates of novice adolescents during the first two quarters of the learner period were higher, and much more variable, than the rates of experienced adult drivers (IRR = 5.01, CI = 1.84, 13.61 and IRR = 2.31, CI = 1.10, 4.89, respectively). Following, adolescent CNC rates declined, and did not differ from adult rates, during the last two quarters of the learner period. Overall, during the entire learner period the mean CNC rates of novice adolescent and experienced adult drivers were reasonably similar (IRR = 1.67, CI = .98, 2.81). CNC rates of adolescents during the first quarter of independent driving were eight times higher than their rates in the last quarter of the learner period (IRR = 8.32, CI = 2.01, 34.34). During the independent driving period, CNC rates of the novice adolescent drivers were substantially higher than the rates of the experienced drivers in each quarter (Q1: IRR = 5.43, CI = 2.85, 10.36; Q2: IRR = 7.53, CI = 3.38, 16.76; Q3: IRR = 4.27, CI = 1.78, 10.25; and Q4: IRR = 14.15, CI = 3.40, 58.84) and overall (IRR = 6.51, CI = 4.03, 10.51), with no decline over the 12-month study period.
      Fig 1
      Figure1Quarterly mean incidence rates and CI for CNC rates per 10,000 miles by license status (learner or independent) for novice adolescents and experienced adult drivers. The vertical dashed line represents adolescents’ licensure time.
      The analysis indicated that CNC rates for both novice adolescents and experienced adult drivers did not differ significantly according to driver gender, time of day, or road surface conditions (data available upon request).

      Kinematic risky driving rates

      KRD rates during the learner period were consistently low, with about five events per 1,000 miles each quarter for both novice adolescents and experienced adult drivers (Figure 2). With the transition to the independent period, adolescent KRD rates increased to their highest value of about 20 events per 1,000 in the first quarter, around four times higher than the quarter before licensure and four times higher than that of adult drivers. Adolescent KRD rates declined significantly from the first to the third quarter of independent driving, then increased somewhat in the fourth quarter. KRD rates for adolescents over the entire independent driving period were higher than that of experienced drivers during the same period (IRR = 3.95, CI = 2.96, 5.26).
      Fig 2
      Figure2Quarterly mean incidence rates and CI for KRD rates per 1,000 miles by license status (learner or independent) for novice adolescents and experienced adult drivers. The vertical dashed line represents adolescents’ licensure time.
      Figure 3 illustrates the KRD rates by license status and gender for novice adolescents and experienced adult drivers. During the learner period, KRD rates did not vary by gender among adolescents or adults. However, KRD rates for the entire independent driving period were higher among male than female adolescent drivers (IRR = 1.73, CI = 1.16, 2.60). For both female and male adolescents, KRD rates were three times higher the first quarter of independent driving compared to the last quarter of the learner period. While KRD rates for adolescent females decreased with time and were significantly lower the last quarter compared to the first quarter of independent driving (IRR = .63, CI = .57, .69), KRD rates of male adolescents did not follow this trend and were slightly higher in last quarter compared to the first quarter (IRR = 1.11, CI = 1.02, 1.21). Finally, during independent driving KRD rates of adolescents were consistently higher than adult drivers of the same gender (females: IRR = 3.62, CI = 2.49, 5.27; males: IRR = 4.46, CI = 2.87, 6.92).
      Fig 3
      Figure3Quarterly mean incidence rates and CI for KRD rates per 1,000 miles by license status (learner or independent) and gender for novice adolescents and experienced adult drivers. The vertical dashed line represents adolescents’ licensure time.
      As shown in Figure 4, adolescents had higher KRD rates when driving during the day than at night during both learner and independent driving periods (IRR = 1.54, CI = 1.26, 1.89 and IRR = 1.32, CI = 1.25, 1.39, respectively). Adolescent daytime KRD rates were significantly higher than those at night during the second and third quarters of learner period. However, in the last quarter of the learner period there were no significant differences in KRD rates based on time of day. KRD rates among adolescents were significantly higher during independent driving than learner period for both day and night time driving (IRR = 4.01, CI = 3.42, 4.70 and IRR = 2.65, CI = 2.05, 3.43, respectively). KRD rates of the experienced drivers were relatively stable throughout time and did not differ according to time of day. Comparing day and night KRD rates of adolescent and adult drivers indicated that while there were no differences in rates between adolescent and adult drivers during the learner period (day: IRR = 1.10, CI = .82, 1.48; night: IRR = .85, CI = .61, 1.20), in the independent period adolescents’ day and night KRD rates were consistently higher than those of adult drivers (day: IRR = 4.09, CI = 3.06, 5.46; night: IRR = 3.65, CI = 2.72, 4.91).
      Fig 4
      Figure4Quarterly mean incidence rates and CI for KRD rates per 1,000 miles by license status (learner or independent) and time of day for novice adolescents and experienced adult drivers. The vertical dashed line represents adolescents’ licensure time.
      Differences in adolescent KRD rates according to road surface conditions were evident only during the independent driving period (Figure 5). Throughout the entire first year of independent driving, adolescent KRD rates were higher compared to the rates in the learner period for both dry (IRR = 4.02, CI = 3.68, 4.38) and wet (IRR = 3.19, CI = 2.55, 3.99) road surface conditions. The transition of adolescents from learner to independent driving was characterized by significant increase in KRD rates for both wet and dry road surface (IRR = 2.45, CI = 1.69, 3.55 and IRR = 3.78, CI = 3.27, 4.37, respectively). Adolescent KRD rates on wet road surface did not change across the independent driving period. In contrast, their rates on dry weather did change over time, decreasing gradually in the second and third quarters and increasing at last quarter of the independent period (IRR = .89, CI = .83, .94, IRR = .81, CI = .76, .87 and IRR = 1.15, CI = 1.07, 1.24). No differences were found between wet and dry KRD rates of adolescents and experienced drivers during the learner period (wet: IRR = 1.01, CI = .71, 1.46; dry: IRR = 1.05, CI = .78, 1.41). However, in the independent period, adolescent KRD rates on wet and dry road surface were consistently higher than those of experienced drivers (wet: IRR = 2.62, CI = 1.93, 3.56; dry: IRR = 4.24, CI = 3.18, 5.65).
      Fig 5
      Figure5Quarterly mean incident rates and CI for KRD rates per 1,000 miles by license status (learner or independent) and time of day for novice adolescents and experienced adult drivers. The vertical dashed line represents adolescents’ licensure time.
      Adult passengers were present in almost all learner period driving, as required by law, but present in only 5% of the independent driving trips. Previous research indicated that novice adolescents drive more safely with adult passengers [
      • Simons-Morton B.G.
      • et al.
      Crash and risky driving involvement among novice adolescent drivers and their parents.
      ]. Therefore, we repeated the same analysis for KRD while removing trips with an adult passenger from the independent adolescent driving data. The results followed the same trends reported with small differences in the estimates (data not shown).

      Discussion

      This study is the first to provide objective longitudinal rates of CNC and KRD among adolescents over the learner period and the first year of independent driving. Adolescent CNC and KRD rates were compared to those of adults driving the same vehicles, over the same period of time, on similar roads, and under similar driving conditions. CNC and KRD rates of experienced adult drivers provided a reference point and placed in context the rates of novice adolescent drivers. Study findings indicate that adolescents drove relatively safely during the learner period, maintaining low CNC (after the first two quarters) and KRD rates that were similar to those of experienced adult drivers. Presumably, despite limited experience, relatively safe driving was fostered by the mandatory presence of a supervising adult, who likely assisted the novice in many ways, including anticipating maneuvers, maintaining driver attention to the driving task, preventing distraction, managing the vehicle environment, and encouraging driver self-control [
      • Simsons-Morton B.
      • Ehsani J.P.
      Learning to drive safely: Reasonable expectations and future directions for the learner period.
      ,
      • Simons-Morton B.G.
      • Ouimet M.C.
      What is the effect of passengers on teenage driving?.
      ].
      CNC and KRD rates increased dramatically with the transition from learner to independent driving, on average eight and four times higher, respectively, from the first quarter of independent driving to the last quarter of the learner period, consistent with previous research [
      • Mayhew D.R.
      • Simpson H.M.
      • Pak A.
      Changes in collision rates among novice drivers during the first months of driving.
      ,
      • McCartt A.T.
      • Shabanova V.I.
      • Leaf W.A.
      Driving experience, crashes and traffic citations of teenage beginning drivers.
      ,
      • Simons-Morton B.G.
      • et al.
      Crash and risky driving involvement among novice adolescent drivers and their parents.
      ]. It is not entirely clear to what this dramatic increase in CNC and KRD rates immediately after licensure could be attributed. It is possible that driving without the support of adult supervisors, experimentation, and exposure to new and diverse driving conditions contributed to the increase in rates. The first year of independent driving was vastly dangerous for novice adolescent drivers, with 6.5- and 4-folds higher CNC and KRD rates compared to experienced adults, which is consistent with previous findings [
      • Simons-Morton B.G.
      • et al.
      Crash and risky driving involvement among novice adolescent drivers and their parents.
      ].
      While the first 9 months of independent driving were characterized by a modest decline in KRD rates that is consistent with learning, CNC rates did not decline, and were largely invariable regardless of driving conditions. Previous self-report and police-report analyses indicated declines in crashes the first year of licensed driving. However, a recent analysis of crash data from the Second Strategic Highway Research Program Naturalistic Driving Study (SHRP2 NDS) indicated no such decline among novice teenage drivers [

      Simons-Morton BG, Gershon P, Ehsani, et al. Crash rates over time among novice and older drivers in the SHRP2 Naturalistic Driving Study. The Seventh International Symposium on Naturalistic Driving Research, Blacksburg, VA, August 28–30.

      ]. Hence, the relatively high CNC and KRD rates documented in the current study may reflect persistent risk taking during the first year of licensure.
      KRD rates at the learner period did not vary by gender, nor by any evaluated environmental factors. In contrast, during independent period adolescents’ risky driving behavior varied with higher KRD rates under favorable environmental conditions, such as day time or dry roads. Gender differences were also observed, as adolescent males had higher KRD rates compared to adolescent females; but mainly adolescents were consistently higher than adults.
      Limitations of the research include the relatively small, regional, and volunteer sample. While CNC served as a useful measure of risk, most of the documented events were near crashes and few of the crashes were severe. As such, the findings should be interpreted in the context of risky driving and are not intended to provide estimates for crash rates. Analyses by environment conditions, such as time of day and road surface conditions, necessarily included smaller number of events. Therefore, minor variations over time and within groups should be interpreted with caution. Additionally, we were only able to follow the adolescents for 1 year of independent driving. Future research could benefit from larger samples and longer follow-up periods to better estimate adolescents’ crash rates, capture learning, examine change over time, and evaluate the impact of adult passenger during early independent driving.
      In conclusion, we find that novice adolescent risky driving, objectively measured as KRD and CNC, was higher during the first year of independent driving compared to the learner period and to experienced adult drivers. The learner driving period was relatively safe, suggesting that adolescents can drive safely with an adult in the vehicle. However, the transition to independent driving was typified by a dramatic and persistent increase in risk that was largely invariant by the driving conditions assessed. The lack of consistent declines in risk measures over the first year of independent driving is inconsistent with what would be expected if adolescents were learning from experience. This suggests that during the first year of driving on their own, adolescents either prefer risky driving or lack the judgment required to drive safely. As such, there are possible limits to the creation of policies that reduce exposure to risk by controlling the context of driving and driving conditions. However, future research should evaluate whether the presence of adult passenger during early independent driving stage could attenuate the increase in risky driving behaviors. In the absence of an adult, accelerometer technology is readily available to assess KRD events, providing immediate feedback to the novice driver, and routine reports to parents. This approach has been proven effective in reducing KRD rates of novice adolescent drivers [
      • Simons-Morton B.G.
      • Bingham C.R.
      • Ouimet M.C.
      • et al.
      The effect on teenage risky driving of feedback from a safety monitoring system: A randomized controlled trial.
      ]. Despite the advent of automated vehicles, young driver crash risk will remain a concern during this transition, mainly since adolescents tend to drive the oldest, least technologically advanced vehicles. Hence, there remains a need to develop creative prevention measures.

      Funding Sources

      This research was funded by the intramural research program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development , contract no. HHSN27520130026I .

      References

        • National Highway Traffic Safety Administration (NHTSA)
        Traffic safety facts 2014: Young drivers.
        US Department of Transportation, 2016 (. Washington, DC)
        • National Highway Traffic Safety Administration (NHTSA)
        Traffic safety facts 2013: Young drivers.
        US Department of Transportation, 2015 (Washington, DC)
        • Chen L.-H.
        • Baker S.P.
        • Li. G.
        Graduated driver licensing programs and fatal crashes of 16-year-old drivers: A national evaluation.
        Pediatrics. 2006; 118: 56-62
        • Curry A.E.
        • et al.
        Comparison of older and younger novice driver crash rates: Informing the need for extended graduated driver licensing restrictions.
        Accid Anal Prev. 2017; 108: 66-73
        • Ehsani J.P.
        • Bingham C.R.
        • Shope J.T.
        The effect of the learner license Graduated Driver Licensing components on teen drivers’ crashes.
        Accid Anal Prev. 2013; 59: 327-336
        • Masten S.V.
        • Foss R.D.
        • Marshall S.W.
        Graduated driver licensing and fatal crashes involving 16-to 19-year-old drivers.
        JAMA. 2011; 306: 1098-1103
        • Mayhew D.R.
        • Simpson H.M.
        • Pak A.
        Changes in collision rates among novice drivers during the first months of driving.
        Accid Anal Prev. 2003; 35: 683-691
        • Lewis-Evans B.
        Crash involvement during the different phases of the New Zealand Graduated Driver Licensing System (GDLS).
        J Saf Res. 2010; 41: 359-365
        • McCartt A.T.
        • Shabanova V.I.
        • Leaf W.A.
        Driving experience, crashes and traffic citations of teenage beginning drivers.
        Accid Anal Prev. 2003; 35: 311-320
        • Simons-Morton B.G.
        • et al.
        Crash and risky driving involvement among novice adolescent drivers and their parents.
        Am J Public Health. 2011; 101: 2362-2367
        • Guo F.
        • Simons-Morton B.G.
        • Klauer S.E.
        • et al.
        Variability in crash and near-crash risk among novice teenage drivers: A naturalistic study.
        J Pediatr. 2013; 163: 1670-1676
        • Shope J.T.
        • Bingham C.R.
        Teen driving: Motor-vehicle crashes and factors that contribute.
        Am J Prev Med. 2008; 35: S261-S271
        • Williams A.F.
        Teenage drivers: Patterns of risk.
        J Saf Res. 2003; 34: 5-15
        • Bingham Cnd.
        • et al.
        Do as I say, not as I do: Distracted driving behavior of teens and their parents.
        J Saf Res. 2015; 55: 21-29
        • Gershon P.
        • O'Brien F.
        • Zhu C.
        • Simons-Morton B.G.
        Multi-level predictors of teenage risky driving.
        in: Young driver subcommittee midyear meeting. Transportation Research Board, 2016
        • Goodwin A.H.
        • Foss R.D.
        • Harrell S.S.
        • O'Brien N.P.
        Distracted driving among newly licensed teen drivers.
        AAA Foundation for Traffic Safety, Washington, DC2012
        • Rajaratnam S.M.W.
        • et al.
        Teen crashes declined after Massachusetts raised penalties for graduated licensing law restricting night driving.
        Health Affairs. 2015; 34: 963-970
        • Carney C.
        • McGehee D.V.
        • Harland K.
        • et al.
        Using naturalistic driving data to assess the prevalence of environmental factors and driver behaviors in teen driver crashes.
        AAA Foundation for Traffic Safety, Washington, DC2015
        • Simons-Morton B.G.
        • et al.
        Do elevated gravitational-force events while driving predict crashes and near crashes?.
        Am J Epidemiol. 2012; 175: 1075-1079
        • Prato C.G.
        • et al.
        Modeling the behavior of novice young drivers during the first year after licensure.
        Accid Anal Prev. 2010; 42: 480-486
        • Rhodes N.
        • Pivik K.
        Age and gender differences in risky driving: The roles of positive affect and risk perception.
        Accid Anal Prev. 2011; 43: 923-931
        • Carney C.
        • et al.
        Using an event-triggered video intervention system to expand the supervised learning of newly licensed adolescent drivers.
        Am J Public Health. 2010; 100: 1101-1106
        • Ehsani J.P.
        • Li K.
        • Grant B.J.
        • et al.
        Factors influencing learner permit duration.
        Safety. 2016; 3: 2
        • Simsons-Morton B.
        • Ehsani J.P.
        Learning to drive safely: Reasonable expectations and future directions for the learner period.
        Safety. 2016; 2: 20
        • Simons-Morton B.G.
        • Ouimet M.C.
        What is the effect of passengers on teenage driving?.
        in: Caird J. Fisher D. Handbook of young driver research: Research, practice, policy, and directions. CRC Press, New York2016: 239-254
      1. Simons-Morton BG, Gershon P, Ehsani, et al. Crash rates over time among novice and older drivers in the SHRP2 Naturalistic Driving Study. The Seventh International Symposium on Naturalistic Driving Research, Blacksburg, VA, August 28–30.

        • Simons-Morton B.G.
        • Bingham C.R.
        • Ouimet M.C.
        • et al.
        The effect on teenage risky driving of feedback from a safety monitoring system: A randomized controlled trial.
        J Adolesc Health. 2013; 53: 21-26

      Linked Article