Paper No.

01-2047

PREPRINT
Duplication of this preprint for publication or sale is strictly prohibited without prior written permission of the Transportation Research Board

Title: CRASH RATES AND TRAFFIC

MANEUVERS OF YOUNGER DRIVERS

Authors: Adam Kirk (ajkirk0@pop.uky.edu)

Nikiforos Stamatiadis (nstamat@engr.uky.edu)

Department of Civil Engineering University of Kentucky Lexington, KY 40506-0281 Ph. (859) 257-8012 Fax (859) 257-4404

Transportation Research Board 80th Annual Meeting January 7-11, 2001 Washington, D.C.

Kirk & Stamatiadis ABSTRACT

Page 2

While over the past decades the population of younger drivers has been decreasing, their crash rates have increased. Past research has associated their higher crash rates to societal influences and youthful behavior. The objective of this research is to identify the specific driving maneuvers whose unsuccessful undertaking results in specific types of crashes involving these drivers. Four types of crashes were identified as the most prominent for young drivers including crashes at intersections, rear end, crashes resulting from passing maneuvers and single vehicle crashes. The analysis was performed examining the Kentucky crash database for the 1994-1996 period using the quasi-induced exposure method. The results showed that for all crashes there is a general trend of decreasing involvement with increasing age, which indicates that their inexperience is the largest single contributor to their increased crash rates. Of significance is the fact that for all crashes a dramatic decrease of involvement after the first year of driving between the years of 16 and 17 is observed. This may be indicative of a steep learning curve in the first years of driving regarding the ability to control a vehicle. Therefore, very little can be made to improve this phenomenon. Increasing the level of awareness among young drivers about these issues and their likely crash involvement seems to be the only viable approach. However, preliminary efforts from the graduated license show that some of these trends seem to be reduced indicating a possible impact on the crash rates of young drivers.

KEYWORDS: Safety, Younger drivers, Crash rates, Crash types.

Kirk & Stamatiadis INTRODUCTION

Page 3

Over the past decades, the population of young drivers has been continuously decreasing. In 1970, drivers between 15 and 20 years accounted for 11.7% of the U.S. driving population while in 1998, 6.8% (1). Thus, during the 1970-1998 period young drivers showed a 42% reduction in their driving population which was the only age group experiencing a population decline. Despite this decrease, the crash rate of younger drivers has increased during the last decade when considering crashes per licensed drivers and they were involved in more fatal and all crashes than any other age group (2). Past research has associated the higher crash rates of younger drivers to their societal influences and youthful behavior (3). Teenage drivers may as a result of their limited driving experience and high risk taking behavior cause more crashes. Given this background the objective of this research is to identify the specific driving maneuvers whose unsuccessful undertaking results in specific types of crashes involving younger drivers, defined here as drivers 20 years old or younger. This crash analysis will be conducted with crash records from the Kentucky Sate Crash Database. Potential counter-measures will be discussed based on this analysis, which can be evaluated in a future work using younger driver focus groups. Younger drivers have shown high crash involvement rates and tend to be overrepresented in specific types of crashes involving certain traffic maneuvers and conditions (4). A traffic maneuver is defined here as the intended action of the driver who caused the crash. Consideration of these maneuvers allows for the determination of possible deficiencies among age groups and can provide additional insight in determining whether age is a factor in crash occurrence. In order to conduct this type of analysis, one needs to disaggregate crash records to further understand which conditions or maneuvers are problematic for each group. A focal point of this research is the correlation of the causal factors to driver and specific maneuver characteristics. Crash rates disaggregated into these types of categories would be useful metrics for establishing the relative strengths or deficiencies among groups. This will provide invaluable direction for policy development aimed at improving highway safety.

Kirk & Stamatiadis LITERATURE REVIEW

Page 4

Over the years the high crash rates of young drivers have caused concern within the population of drivers and the highway safety community. Realizing that no two driving populations have similar driving characteristics, it can be concluded that none are alike in the type of crash involvement. This literature review was conducted to identify possible characteristics and identify trends of the crash involvement of young drivers. Areas and circumstances which showed definite cause for concern as well as areas in which the significance of occurrence is questionable have been noted and examined. These areas of significant overrepresentation are single vehicle crashes on rural highways and movements at intersections related to gap acceptance (4,5,6). Questionable areas remain as to side impact collisions, rear end crashes, and dangerous overtaking and passing (5, 6, 7, 8). Additional factors that examined included nighttime driving and occupancy rates. Significant research has been completed investigating the disproportionate number of younger drivers in single vehicle crashes when compared to older drivers, since such crashes remain to be the number one fatal crash for this age group (1). Youth single vehicle crashes involving alcohol have decreased steadily from 1982 to 1994 by 55% while in the same time period a decrease of 21.8% in youth single vehicle crashes was observed (4). This obviously shows a predisposition to single vehicle crashes for this age group related to factors other than alcohol impairment. Speeding is another frequent cause of crashes for drivers 18-24 years old, with the highest speed observed for drivers under 20 years of age (6). This is indicative of the propensity of young drivers to be involved in crashes from loss of control of the vehicle, which could be interpreted as having a high rate of speed for the roadway conditions. Largely due to this increased speed, youths have a much greater frequency citing loss of control as the primary reason of a crash (28.5%) as compared to older drivers (16.9%) (8). Even though young drivers as a whole have a higher risk than older drivers, an examination of characteristics of sub-groups which are at an even higher risk within this population, can lead to a better understanding of the causal factors for this attributed increased risk (6). While past research has shown mixed results regarding the significance of rear end collisions among young drivers, the majority of findings support an increased crash propensity.

Kirk & Stamatiadis

Page 5

A study on Swedish drivers found the primary collision factor to be the increase of speed with a decrease in age. Drivers under 20 years of age had the highest observed speed and it was shown that young drivers endorsed higher speed and dangerous overtaking procedures (6). This was expressed in the overrepresentation of youths in crashes with improper vehicle following (tailgating), careless lane changes, and careless passing. Another study found that young drivers, while having the same involvement as more mature drivers, had a higher representation as the rear ender than as the rear endee (7). Their increased likelihood to be involved in rear end crashes was attributed to their greater confidence of driving at higher speeds and following with smaller gaps while at the same time their awareness of the situation was decreased (8). Another area of concern for young drivers has been that of movements at intersections; especially turning left into oncoming traffic. This problem however, seems to have different reasons depending on gender. Broughton found that young females were involved in more crashes at intersections than were males (5). In relation to these incidences, a gender related difference in visual perception was cited, demonstrating problems of females with spatial perception and orientation. Thus females were more susceptible to poor performance in gap acceptance tasks at intersections, which may explain the overrepresentation of young females in crashes occurring at intersections (8). It is should be noted that older drivers are more inclined to be involved in crashes involving the violation of right of way than younger drivers. However, the potential severity levels of such crashes and the likelihood of incorrect gap estimation by young drivers pointed toward further investigation of crashes at intersections. While the scope of this review was initially for specific crash types, two additional areas of concern emerged: the time of the day of the crash occurrence and vehicle occupancy rates for crashes involving youth. A gender difference is once again obvious in terms of crash rates for these condition (9). Nighttime driving due to the fact that it requires operating a vehicle with lowered visibility at a time when there is a higher likelihood for a driver to be fatigued, is naturally a more dangerous time to be driving. Research showed that nighttime driving was associated with a higher risk of crash than daytime driving, but this risk was even more pronounced for young drivers (9). The same study also found that Friday and Saturday nights had the highest crash

Kirk & Stamatiadis rates again with a disproportionate number being young drivers. This was especially

Page 6

troublesome considering the fact that young drivers had the lowest travel distance of all groups studied and the highest frequency of crash at night when exposure is considered (8,9). It was noted that graduated licensing tries to reduce crashes of this age group by reducing exposure through limiting driving at especially hazardous times of the day. Even more alarming is the effect that occupancy has among young drivers, which was examined recently (10). For older age groups, increased passenger occupancy decreased the likelihood of a crash. However, the increased passenger occupancy increased the likelihood of crashes of drivers 16-19 years old with a dramatic increased risk when there are 3 or more passengers in a car. Young drivers were two times as likely to be involved in a crash when there was one or more passengers and 16-19 year old males remained to be the population at greatest risk in all categories. In a study of Swedish young drivers, risky driving (i.e. speeding and dangerous maneuvering) was shown to coincide with the fact that peer and societal pressures made younger drivers feel the need to engage in these types of activities (11). Peer pressure is obviously more pronounced when there is a high peer occupancy rate and it increases significantly in the presence of passengers (7). A unique study done in East Germany allowed the effects of experience and age to be studied independently (12). Following reunification of Germany many older people had their first opportunity to acquire a driving license. Thus, a large population of older novice drivers was available to be compared with the crash rates of younger novice drivers. The difference in crash rates was shown to be mainly the result of age related differences and the effect of lack of experience, while present, was found to be very much in the background as compared to age. This clearly shows a characteristic of youth, which places the young driving population at high risk, unrelated to their lack of driving experience. Overall, several specific types of crashes and crash settings have been identified and have been shown to contribute to the higher crash involvement of young drivers. These are single vehicle crashes, rear-end and passing crashes, incidents at intersections involving small gap acceptance, and incidences at night. Perhaps of significant importance is the important role occupancy plays in increasing the risk of crashes of young drivers. It has also been determined

Kirk & Stamatiadis that the main cause of such crashes is the relative young age of the drivers and not their inexperience. In conjunction with a more precise estimation of the crash involvement, countermeasures may be developed in order to reduce the number of youth casualties and fatalities from roadway incidences.

Page 7

METHODOLOGY From the literature review, four main types of crashes have been identified as having either high or questionable involvement of young drivers. From past research, it is also evident that there may exist different causal factors for each of the four types of crashes. The factors of period of the day, number of occupants, and driver gender have also been identified as possible influences on youth driving characteristics. Through examination of the influence of these factors on each category of crashes it is hoped to gain a better understanding of the crash characteristics of young drivers in each of the four main types of crashes. Although computerized databases can yield sufficiently accurate estimates of the frequencies of crashes sub-divided by many roadway, driver, and environment characteristics, correspondingly accurate estimates of crash exposure for these categories are often difficult or impossible to make. Moreover, investigators sometimes disagree about which exposure measure is most appropriate for each specific application. The traditional methods are based on estimating the amount of vehicle-miles traveled (VMT) by simply multiplying the average daily traffic (ADT) with the length of the roadway. However, the use of VMT's calculated in this manner prohibit the development of exposure metrics for specific driver and situation combinations, such as the period of the day, driver/passenger characteristics and roadway features. To overcome some of the problems and limitations in estimating exposure by driver and situation type from exogenous values such as travel distance, drivers licensed, and vehicles registered, methods have been developed that derive exposure estimates from the crash database itself. These techniques are coming into more wide-spread use and have been recently validated against more conventional techniques (13). These quasi-induced exposure techniques have been developed specifically to obtain first-order approximations of relative travel by different classes

Kirk & Stamatiadis

Page 8

of road users, which are acceptable surrogates for vehicle-miles of travel when estimates are made for conditions during which the mix of road users is relatively constant. This approach is ideal for this application, since dissaggregate travel estimates are not available. This study examines crash databases for the Commonwealth of Kentucky between the years of 1994-1996. Examination of data within this time period allows for recent crash data as well as data reflecting youth crash involvement before the graduated licensing program was instituted within the state at the end of 1996. The measure of the accident causing propensity used in this quasi-induced exposure methodology is the relative accident involvement ratio (RAIR). The RAIR is calculated by taking the ratio of the percentage of at-fault drivers in a specific subgroup to the percentage of not-at-fault drivers from the same subgroup. The assumption is that the distribution of not-responsible drivers is a representative sample of the travel exposure of all drivers in that group. In order to develop this ratio it was necessary to define at-fault and not at-fault drivers. The issue of assigning fault in a crash is determined from information contained in the accident database. Each accident entry includes a human factor category that indicates what each driver did to contribute to the accident. If any contribution to the crash occurrence is evident, then the driver is considered to be at fault. To avoid cases where biases may be introduced due to miscoding of the at-fault driver, accidents where both drivers or neither driver had a contributing human factor are excluded from the analysis. There were a total of 12% of crashes that were eliminated from the database due to inability to clearly identify the responsible driver, which is similar to the percentage for drivers over 20 years old (14%). The factors of age, gender, period of day, and number of occupants were also deemed appropriate for further analysis. The age presented herein is that of the age of the driver as reported in the Kentucky crash database. This study only examined the ages of 16 through 20 to give a representative sample of the young driving population and to exclude crashes resulting from the influence of alcohol experienced in driver populations above the legal drinking age of 21. Relative accident involvement ratios were then examined for the effect of each of the previously listed factors on the four types of crashes examined. The effect of these factors have also been examined singularly and in combination with one another, in order to define special

Kirk & Stamatiadis

Page 9

groups of drivers and driving conditions which promote exceedingly unsafe characteristics. Logistic regression analysis was then performed in order to determine statistical significance of the results of the study at the 0.05 level. This analysis allows for the use of a dichotomous independent variable (at-fault or not-at-fault driver) and probability of the at-fault driver is modeled as P (driver is at fault) = 1/(1+e-z) where: Z = B0 + B1X1+ B2X2+ ... + BNXN and Xs = driver factors (e.g. age), confounding factors (e.g. time of crash), and interactions. A multivariate approach was used for evaluating the impact of each of these independent variables on the crash rates. The RAIR was statistically tested with the logistic regressions using the SPSS statistical package (14). The effects of the various independent variable on crash propensity are presented in the following section.

CRASH ANALYSIS Left Turns Past research shows that there is a higher risk involved in making left turning movements at intersections as compared to other intersection movements due to crossing two directions of traffic and estimating speeds of oncoming traffic. Therefore, by examining this movement it is possible to define factors in the driving characteristics of young people to better understand their higher propensity for crash occurrence at intersections. Left turn movements were defined to be any crash taking place at an intersection where the at-fault driver/vehicle was completing a left turning movement on a roadway consisting of two or more lanes. This definition excludes rear end crashes, which are examined separately in another section of the study. No differentiation was taken into account for the presence of turning lanes or for the type of operating traffic control devices at the intersection, since this information was not available from the crash data. The possible influencing factors of age, gender, occupancy rates, and period of the day were examined for this movement. Definite conclusions can be drawn from the effect of age on the crash propensity of younger drivers. The data in Table 1 show a general trend of decreasing crash involvement as the driver ages. However, within this trend a sharp decrease can be seen between the ages of 16 and

Kirk & Stamatiadis

Page 10

17, while after the age of 17 this decrease is dramatically slowed. From these data, it is then reasonable to conclude that crash likelihood decreases rapidly within the first few years of driving. The cause for this decrease is however unclear. This decrease could correspond to an increased experience level of the driver by enhancing one's ability to control the vehicle and perfecting visual search patterns or could arise from an increased maturity level, which would reduce the risky behavior of young drivers - an influencing factor commonly attributed to this age group. A comparison between these drivers to drivers over 20 years indicates that younger drivers exhibit higher rates that the remaining driving population (Table 1). While previous research findings showed that the female driving population was at a greater risk than the male population, this study found no statistical difference between genders (Table 1). Another factor examined was the period of the day when the crash occurred. The statistical analysis showed no significant difference between day and night crashes (Table 1). This is especially surprising considering the higher volume of traffic that is likely to be present during the daytime period. This could possibly be explained by the operational characteristics of these movements. The vehicle at greater risk would be the first in line, since it would initiate a left turning movement into oncoming traffic. However, in times of high traffic volumes, leftturning vehicles would most likely move in queues and thus, there is a decreased risk for vehicles in the middle of the queue. At nighttime, due to lower volumes of traffic, the likelihood of being the initiating vehicle is greater. This added risk of nighttime driving could be offset by the lower volume of traffic, when compared to that of the high volume of daytime traffic. Finally, the effect of occupancy rates was examined and no statistically significant differences were noted in the overall group of young drivers nor when split by age categories (Table 1). Therefore, the overall contributing factor to the increased involvement of youth in left turn crashes at intersections may be shown to be influenced more heavily by perception tasks as well as their ability to control the vehicle.

Rear End Rear end crashes were considered because it has been hypothesized that the reason for a higher incident of youth drivers in rear end crashes is due to a feeling of greater confidence driving at

Kirk & Stamatiadis

Page 11

higher speeds and following with shorter gaps. Thus, due to this behavior the young driver is more often to be found as the rear ender as opposed to being the rear endee, i.e they are more likely to be at fault. This higher involvement may be indicative either of their experience deficit or motivational factors that promote risk-taking behavior. First, the factor of driver age was examined, which was determined to be of statistical significance (Table 1). A continuous decline of the rear end crash involvement for all age groups examined was noted. This trend is significant because it demonstrates a continuous trend of safer driving throughout all five years of age. Thus, the reason for the decline should be based on a model of a more gradual and longer learning curve than that of the intersection crash model. However, these drivers still show higher rates compared to drivers over 20 years (Table 1). The influence of driver gender on crash involvement was also examined. The data showed that males are having higher involvement ratios than females - a statistically significant result (Table 1). However, when the populations were further divided and studied by both age and gender there is no significant difference in crash involvement between the genders of the same age group. The statistical analysis for the time of day showed that night ratios are higher than those of the day (Table 1). When age is considered statistical significance was only found for the ages of 16, 17 and 18 years. Also, when the raw data is examined it is found that approximately 84% of all rear end crashes in which youth were involved (both being at fault and not at fault) occurred during daytime as opposed to sixteen percent of which occurred at night. This was somewhat expected, since the major volume of traffic is present during the day especially during peak traffic conditions. It would be expected that higher volumes of traffic would also contribute to a higher crash involvement rate due to the added congestion. However, the rates of the young drivers reflect an increased involvement during the nighttime with lower traffic volumes. Therefore, it could be hypothesized that the increased involvement of the drivers during the nighttime is a result of motivational factors and not an experience related behavior. This trend is also shown to decrease during the ages of 19 and 20, perhaps this demonstrates an increased maturity of the driver. The analysis of occupancy rates showed that all groups of occupants have significance in determining crash involvement (Table 1). Again when the group of drivers is further subdivided by occupancy and age all statistical significance is lost.

Kirk & Stamatiadis

Page 12

Thus it is likely that youth drivers driving alone are willing to take different (and possibly more) chances than when others are present.

Passing Another type of maneuver in which youths are said to have a higher incidence rate is that of head-on crashes caused when the driver attempts to pass or overtake other cars. In this study only passing on two-lane, two-way roads was examined, since this is the only road where such a crash can be observed. Due to possible discrepancies in the reported information given by crash databases, passing crashes were taken to be any crash on a two lane two way road away from an intersection that involved two vehicles and was not classified as a rear end crash. While this approach considers sideswipe crashes as well as head on crashes, it is believed that a sideswipe is a crash type resulting from an attempt to overtake where the driver misjudged the available distance. The cause for the broadening of the scope for this study was to overcome the effect of false reporting of crash causes and initial maneuvers The same decreasing trend with age is evident as that observed in rear end and intersection crashes: there is a decreasing trend with increasing age which is also statistically significant but also a higher general rate when compared to drivers over 20 years (Table 1). The dramatic reduction in the crashes after the initial year of driving between 16 and 17 years may be demonstrative of the potential presence of a sharp learning curve within the first few years of driving. The gender of the driver analysis showed that there is significance in terms of driver gender, where male drivers have much higher involvement rates in passing crashes than female drivers (Table 1). When subdivided further into age categories, while no statistical significance is found, some interesting observations can be made (Figure 1). First, there seems to be a rather sizable difference between male and female drivers in the first year of driving with males having higher involvement rates than females. From the literature review it is evident that young males view themselves as much more skilled drivers and have higher confidence levels in their driving than do females. With these increased confidence levels, males would therefore be more apt to engage in passing maneuvers and to engage in them when there is less room for correction.

Kirk & Stamatiadis

Page 13

However, between the ages of 17 to 19 years old the female drivers actually show a higher involvement than males. This could be explained by the hypothesis of females being more comfortable behind the wheel at these ages and thus are engaging in riskier activity. Therefore, while males are shown to have higher involvement than females over the entire age group in terms of passing maneuvers, this may be attributed to possible discrepancies during the first year of driving. Passing was also examined with respect to the period of the day in which the crash occurred and no statistical significance was found between the day and night periods. The analysis of the effect of occupants on crash involvement showed no significance in terms of occupancy rates (Table 1). This is especially surprising due to the fact that the crashes resulting from a passing maneuver are most often thought to result from motivational factors, such as engaging in risky maneuvers, and the belief that increased peer occupancy rates increase the tendency to participate in risky behavior.

Single Vehicle Crashes Perhaps one of the most troubling of all crashes studied is that of the single vehicle crashes, due to its high frequency of occurrence among young drivers and the higher severity levels resulting from such a crash. While this crash is due to the sole interaction of the driver and the roadway, it can be hypothesized that there are some contributing factors prompting the driver to engage in risky behavior or the driver is not fully capable of controlling the vehicle. One factor that may be present however is that of inattention of the driver causing him/her to run off of the road and crash. The effect of age on crash involvement was first examined and the overall decreasing trend of crash propensity with age was also noted here (Table 1). These data clearly show that after the first year of driving there is a dramatic decrease in the involvement rates of young drivers, which may be indicative of a sharp learning curve or of a sharp decrease in motivational factors placing drivers at risk. Even though younger drivers show a decrease in their crash rates as they age, they still exhibit significantly higher crash rates than the remaining driving population (Table 1).

Kirk & Stamatiadis

Page 14

Another factor examined was that of the gender of the driver and the statistical analysis showed that males have much higher involvement rates than females (Table 1). This is in accordance to findings discussed in the literature review where males are believed to have a higher likelihood to engage in risky behavior and to take chances. However, when broken down by age and gender a new trend appears. Figure 2 shows that females in the 17 and 18 year old age groups have a higher involvement in single vehicle crashes than males from the same ages. In fact, the statistical analysis showed significance only for the age group of 17 years, where females have a higher involvement rate than males. This finding may indicate a different maturing period between genders, where males may curve their risk taking behavior after their first year of driving. This difference may also reflect a different starting point in driving behavior, where females start driving more frequently at a later age than males. This may be indicative of the possibility that the female driving population is becoming more comfortable after the initial year of driving and thus, engaging in higher risk driving, closely paralleling the trend of the younger males. The findings of the analysis examining only the periods of the day when the crash occurred statistically support the literature findings (Table 1). However, when broken down by age groups, 16, 17 and 18 year old drivers had a higher crash rate during the daytime hours as opposed to the nighttime hours. Finally the effect of the number of occupants on single vehicle crash involvement was examined. While statistical significance was not found for any group of occupants some brief observations can be made from examining the rates in Table 1. These data show that there is an increasing effect of occupancy rates on the crash involvement especially with three or more occupants. This may be due to an increased peer pressure to drive in a more risky behavior or may be due to an added distraction in the car leading to driver inattention

CONCLUSIONS AND DISCUSSION From the data presented here, many hypotheses can be made as to the involvement of young drivers in each of the four types of crashes examined. When the different crash types are examined collectively general trends of youth driving characteristics also emerge and the influence of different factors on crash propagation is identified. Also, areas in which unclear

Kirk & Stamatiadis

Page 15

relationships exist as to crash characteristics can be identified in order to concentrate efforts of future research. The only major factor, among those examined, that was found to significantly contribute to crash involvement while conducting left turns was that of the driver's age. All other factors showed no statistical significance. This information lends itself to the possibility that left turning movements exhibit a crash characteristic affected mainly by the ability of the driver to handle the demands of maneuvering a vehicle through the traffic. This factor is consistent with the trend of decreasing crashes with increasing age, since drivers gain experience with the passing of time. All factors that were believed to influence the risk taking behavior of youth have been shown to have no significant effect and thus, such motivational or risk taking behaviors are not present or, at least, very insignificant during turning movements. Therefore, it could be concluded that the overall contributing factor at intersections is that of experience related abilities to estimate gaps and handle and maneuver the vehicle. From the data analyzed for rear end crashes there still exist many questions about the real cause and true characteristics of youth rear end crash involvement. While the general trend of decreasing involvement is coupled with increasing age, it is unclear as to whether ability, experience or maturity govern youth involvement in this type of crash. Due to significance found regarding the period of the day it may be hypothesized that the motivational factors are the overriding factors in this type of crash. Inconsistencies between occupancy rate findings may again suggest that such crashes are not due to motivational factors but instead due to the lower ability to handle the vehicle properly in higher demand situations. While it is obvious that motivational factors are responsible for the initial engagement and decision to pass another vehicle, they may not be responsible for the final execution of the maneuver. If motivational factors were responsible for increasing crash involvement, it would be expected that increased occupancy would further increase this risk by escalating the motivational factors. Due to the absence of this effect, it may be seen that during such a maneuver the driver will act responsibly within the limits of their perceived ability. This is evident in the gender related difference among 16 years old drivers. Therefore the cause of the crash may be more reliant on the perception ability of drivers to differentiate between safe and unsafe passing

Kirk & Stamatiadis conditions.

Page 16

The traditional understanding of youth single vehicle crashes does not easily compare to the findings of this study. While involvement was found to decrease with age, no other factor examined was found to be consistent with the previous research. New drivers, 16 and 17 years old, were found to have a higher involvement in single vehicle crashes during daytime than during the night. While this may be affected by stricter driving limitations imposed on youths by factors other than licensing requirements, it is largely inconsistent with the belief that the prime time for single vehicle crashes is during nighttime periods. Single vehicle crashes are also believed to be a male dominated crash. However, after the initial year of driving, females are shown to actually have a higher involvement rate than males. Perhaps this is indicative of latent risky driving behavior of females or a later age of females starting to drive more frequently. Through examination of all crash types there is a general trend of safer driving with increasing age throughout all 5 years of age studied. However, of significance is the fact that crashes at intersections, passing maneuvers and single vehicle crashes all demonstrate a dramatic decrease of involvement after the first year of driving between the years of 16 and 17. This may be indicative of a steep learning curve in the first years of driving regarding the ability to control a vehicle. Moreover, the more gradual trend of decreasing involvement crash after this age group also reflects a gradually maturing driver. The influence of driver gender also illustrates interesting results. While it has been previously shown that the male driving population has a higher involvement in crashes than that of the female population, this trend does not apply to all age categories. All crash categories exhibit an initial dominance by male drivers. However, all categories again reflect a greater involvement by females during the second consecutive year of driving. As previously mentioned this may be indicative of females becoming more confident by the second year of driving and thus engaging in higher risk activities. On the other hand, it may be reflective of a tendency for females to start driving on a regular basis at a later age than males. The effect of occupants on youth crash involvement provided inconsistent results and inconclusive data for the crashes examined. The data obtained showed either higher rates while driving alone or higher rates while driving with several passengers. Before any conclusions can

Kirk & Stamatiadis

Page 17

be drawn from these data, it is recommended that occupants be differentiated with respect to age to study the effect of adult and peer passengers on youth driving behavior. This differentiation would be necessary due to the understanding of differing youth behaviors while in the presence of parents as compared to friends. The important finding is that the higher crash involvement is due to the lack of experience and maturity among young drivers. Therefore, very little can be made to improve this phenomenon. Increasing the level of awareness among young drivers about these issues and their likely crash involvement seems to be the only viable approach. However, preliminary efforts from the graduated license show that some of these trends seem to be reduced indicating a possible impact on the crash rates of young drivers. Since driver education has not shown very promising results, this may be a more productive approach for improving their safety.

ACKNOWLEDGMENTS The authors wish to thank the Southeastern Transportation Center and the Federal Highway Administration for financial support (we take full responsibility for the conclusions drawn here).

REFERENCES 1. U.S. Department of Transportation, Federal Highway Administration, Highway Statistics for 1975; and 1998. 2. National Highway Traffic Safety Administration. Traffic Safety Facts 1999Young Drivers, DOT-HS-809-099, U.S. Department of Transportation, Washington, D.C.; 2000. 3. Gregersen, N. P. & Berg, H. Lifestyle and Accidents among Young Drivers. Accident Analysis and Prevention; Vol. 26 No. 3, 297-303; 1994. 4. Lang, S., Waller, P. and Shope J. Adolescent Driving Characteristics Associated with Single Vehicle and Injury Accidents. Journal of Safety Research; Vol. 27 No. 4, 241-257; 1996. 5. Broughton, J. The Variation of Car Driver's Accident Risk and Age. Research Report 135, Department of Transportation; Road Safety Division, Crowthorne, Berkshire, England; 1998.

Kirk & Stamatiadis 6.

Page 18

U.S. Department of Transportation, National Highway Traffic Safety Administration. Understanding Youth Risk Taking. Interim Report. Silver Spring, Maryland; 1995.

7.

Catchpole, J., Cairney, P. and Macdonald, W. Why Are Young Drivers Overrepresented in Traffic Accidents. Australian Research Board Ltd. Special Report 50, Australian Road Research Board Ltd., Vermont South, Victoria, 1994.

8.

Trankle, U., Gelau, C. and Metker, T. Risk Perception and Age Specific Accidents of Young Drivers. Accident Analysis and Prevention; Vol. 22 No. 2, 119-129; 1990.

9.

Doherty, S. Candrey, J. and MacGregor, C. The Situational Risks of Young Drivers; The Influence of Passengers, Time of Day, and Day of Week on Accidents. Accident Analysis and Prevention; Vol. 30 No. 1, 45-52, 1998.

10. Aldridge, B., Himmler, M. Aultman-Hall, L., and Stamatiadis, N. The Impact of Passengers on Young Driver Safety. Transportation Research Record 1693, Transportation Research Board, 25-30; 1999. 11. Gregerson, N. P. Young Driver's Overestimation of Their Own Skill- An Experiment on the Relation Between, Training Strategy and Skill. Accident Analysis and Prevention; Vol. 28 No. 2, 245-250; 1996. 12. Bruhning, E. Kuhnen M. and Berns, S. Traffic Safety of Young People Internationally. Federal Highway Research Institute, Bergisch Gladbach, Germany, 13-28; 1996. 13. Stamatiadis, N. and Deacon, J. A. Quasi-Induced Exposure: Methodology and Insight. Accident Analysis and Prevention; Vol. 50, No. 1, 37-52; 1997. 14. Norusis, M. J. SPSS Advanced Statistics Users Guide, SPSS, Inc. Chicago, IL; 1990.

Kirk & Stamatiadis

Page 19

2.0

Rel. Acc. Inv. Ratio

1.5

1.0 MA LE FEMA LE 0.5

0.0 16 17 18 19 20

Dr iv e r Ag e

Figure 1. Relative Accident Involvement Ratios for Passing Crashes by Age and Sex

2 .0

Rel. Acc. Inv. Ratio

1 .5

1 .0 MA L E FEMA L E 0 .5

0 .0 16 17 18 19 20

D r iv e r Ag e

Figure 2. Relative Accident Involvement Ratios for Single Vehicle Crashes by Age and Sex

Kirk & Stamatiadis Table 1. Relative accident involvement rates for young drivers Crash type Category Age Subcategory 16 17 18 19 20 Gender Male Female Time of day Day Night No. of occupants 1 2 3 Age 16-20 Over 20 Left Turns 1.86 1.14 0.95 0.77 0.70 0.96 1.04 0.94 1.23 0.95 0.99 1.38 1.53 0.91 Rear End 1.42 1.19 0.99 0.86 0.75 1.13 0.84 1.01 0.94 1.07 0.85 1.04 1.58 0.81 Passing 1.38 1.05 1.00 0.91 0.77 1.08 0.88 1.01 0.98 0.97 0.87 0.93 1.68 0.90

Page 20

Single Vehicles 1.61 1.12 1.00 0.85 0.70 1.17 0.79 0.66 2.38 0.96 0.96 1.18 1.65 0.85

Note: Bold ratios are statistically significant at the 5% level.