Although distracted driving among teenagers is of great concern to traffic safety professionals and has received considerable media attention in recent years, rigorous research on this issue has been limited. Most of the research to date has concentrated on the risks associated with teen passengers and driver cell phone use.

Almost no research has examined the many other potential driving distractions often believed to be common and problematic among teenage drivers. In an ongoing study with the AAA Foundation, we collected a sizeable dataset of video data on teen drivers during the provisional licensing stage of GDL.

This in-vehicle data provided a unique opportunity to study distracted driver behaviors and potentially distracting conditions among young, beginning drivers. For the present analysis, we sampled and coded video data with the specific purpose of studying the nature and prevalence of distracted driving among teenagers. The study addressed a number of questions:

 Which distracted driver behaviors are most common among teenage drivers?
 Do males and females differ in how often they engage in distracted behaviors, or the kinds of distractions they experience?
 Do distracted driver behaviors vary based on the number of passengers and the characteristics of those passengers (e.g., teens vs. adults vs. young siblings)?
 Are distracted driver behaviors more common during certain times of day or week (e.g., weekday vs. weekend), and do these behaviors bear any relation to the amount of traffic or other characteristics of the driving environment?
 Do drivers who engage in distracted behaviors spend more time looking away from the roadway than drivers who are not distracted?
 Are distracted driver behaviors associated with serious incidents such as near collisions, or events involving hard braking or swerving?

Methods

The data used to address these questions were collected during a previous investigation of 50 families of novice drivers (Goodwin, Foss, Margolis & Waller, 2010). Event-based data recorders were placed in the vehicles of participating families at the outset of the learner stage so parent and teen behaviors during practice sessions could be directly observed.

These data recorders, obtained from DriveCam, collected video, audio and accelerometer data when a triggering “event” occurred such as sudden braking or an abrupt turn. The data recorders were returned to the family vehicles during the initial 6 months of unsupervised driving, a very high risk time for new drivers.

The data reported in the present study are from this initial period of independent driving. Because vehicles were sometimes shared, we also have data on some more experienced teen drivers – the siblings of the original target teen. In total, the sample for this study included 52 drivers: 38 newly licensed teens as well as 14 high-school aged siblings.

It is important to note this was a “naturalistic” study of teen driving behavior. No interventions were conducted with participating families. During the 228 total months data recorders were installed in vehicles during the intermediate license stage (6 months x 38 vehicles), 24,085 driving clips were recorded for the 52 teens in the study.

A sample of 7,858 clips was selected for coding. Clips with passengers were oversampled to ensure a sufficient sample size for comparisons between different passenger combinations (e.g., driving with teenage peers versus siblings or parents). A coding system was developed to analyze the selected video clips.

This system included detailed information about electronic device use by drivers, other distracted driver behaviors (e.g., adjusting controls, personal hygiene), and distracting conditions that may occur when passengers are present (e.g., horseplay, loud conversation).

Results

Frequency of Electronic Device Use and Other Distracted Driver Behaviors In total, teenage drivers were observed using an electronic device in 6.7% of all driving clips. Nearly twice as many were operating (or suspected of operating) an electronic device than were observed holding a cell phone to their ear (4.3% versus 2.3% of clips).

The frequency of electronic device use varied considerably by driver. Nine drivers (17%) did not use an electronic device in any of their driving clips. By comparison, six drivers (12%) were observed using an electronic device in over 15% of their clips.

Females were twice as likely as males to be using an electronic device. The other distracted driver behaviors we examined included adjusting controls, eating or drinking, personal hygiene, reading, turning around, reaching for an object, and communicating with someone outside the vehicle.

All of the behaviors we examined were relatively rare. Adjusting controls in the vehicle was the most common behavior; reading was the least common. Females were more likely than males to be observed adjusting controls or reaching for objects in the vehicle.

Males were approximately twice as likely as females to turn around while driving. Altogether, excluding electronic devices, teenage drivers engaged in at least one of the distracted driver behaviors in 15.1% of all driving clips.

Once again, the frequency of distracted behaviors varied considerably by driver, but the incidence was not concentrated so heavily among a small subset of drivers as was the case with use of an electronic device.

Frequency of Distracting Conditions Involving Passengers

Loud conversations were evident in 12.2% of driving clips when passengers were present. Horseplay was less common, at 6.3% of clips. Other potentially distracting conditions, such as dancing by passengers or physical contact between the driver and passengers, were quite rare.

There were few, if any, notable differences for male and female drivers in distractions involving passengers. As might be expected, the frequency of potentially distracting conditions depended not only on the presence of passengers, but also on who those passengers were.

Compared to when one teenage peer was in the vehicle, loud conversation and horseplay were more than twice as likely when teens were carrying multiple teenage peers. Conversely, the likelihood of loud conversation and horseplay were markedly less likely with one sibling passenger or when a parent/adult was present.

We also examined whether the frequency of driver electronic device use and other distracted driver behaviors (e.g., adjusting controls, personal hygiene) was related to the combination of passengers. Generally speaking, electronic device use and other distracted driver behaviors were most common when teens were carrying no passengers.

Teen drivers used an electronic device in 8.1% of clips and engaged in other distracted behaviors in 16.9% of clips when driving alone. Not surprisingly, these behaviors were least common when a parent or other adult was in the vehicle.

Distractions and Characteristics of the Driving Setting

Overall, the frequency of distracted driver behaviors and distracting conditions varied little by day of week. However, noteworthy differences were observed by time of day. Distracted driver behaviors (e.g., adjusting controls, personal hygiene) showed a small, gradual decline over the course of the day.

By contrast, loud conversation and horseplay increased, especially at night. In the specific setting of driving at night on weekends with one or more teenage peers (and no adults or siblings), loud conversation was observed in 20.2% of clips and horseplay was observed in 11.2% of clips.

A few studies suggest drivers may be more likely to engage in potentially distracting activities when the driving environment seems “safer.” Consequently, we examined whether distracted behaviors were less common in more challenging conditions such as busy traffic or rain.

Overall, there was no clear relationship between the frequency of distracted driver behaviors or distracting conditions and the amount of traffic. The frequency of distractions was relatively similar in light or moderate/heavy traffic, or when there was no traffic. However, the frequency of distracted driver behaviors and distracting conditions was slightly lower during rain.

Distractions and Looking Away from the Roadway

We coded whether the driver looked away from the roadway at any point during the 10 seconds preceding the vehicle movement that triggered the camera to record. Drivers looked away from the roadway, at least briefly, in 45% of the driving clips where the vehicle was moving and a clear determination could be made of where the driver was looking.

Females were somewhat more likely to look away from the roadway than males. Most drivers who looked away from the roadway did so only briefly. A third (35%) of drivers who looked away did so for one second or less.

A similar proportion (31%) looked away for 1.25 to 2 seconds. However, 12% of drivers looked away from the roadway for at least four seconds during the 10 second period prior to the event. The median amount of time that drivers looked away was 1.50 seconds. We also measured the longest continuous glance away from the roadway among drivers who looked away.

In clips where drivers looked away from the roadway, half the time (51%) the longest glance was one second or less. In 39% of clips, the longest glance away was 1.25 to 2.0 seconds, and in 10% of clips the longest glance was more than 2 seconds. Next, we examined the relationship between distractions and looking away from the roadway.

Drivers were three times as likely to look away from the roadway when using an electronic device. In fact, drivers using an electronic device spent a full second longer looking away during the 10 seconds preceding the event that triggered the camera to record, than drivers who were not using an electronic device.

Drivers were also two and a half times as likely to look away when engaging in some other distracted driver behavior (e.g., adjusting controls, personal hygiene), and they were more likely to look away when there was loud conversation or horseplay in the vehicle.

Distractions and Potentially Serious Events In the final section of the report, we examined the association between distractions and potentially serious events, such as near collisions or events involving high g-forces (indicating hard braking or turning). Of the 7,858 driving clips, only 52 (0.7%) involved a serious incident.

Twenty-seven of the 52 teenage drivers had no serious incidents during the six month period the event data recorder was installed. By contrast, seven teens accounted for 58% of the serious incidents, with three teens having five incidents each. Drivers were approximately six times more likely to have a serious incident when there was loud conversation in the vehicle.

Although driver electronic device use and horseplay were also associated with driving incidents, the confidence intervals were too wide for these associations to be considered meaningful. Finally, we examined whether distracted driver behaviors and distracting conditions were associated with higher g-force events, defined as events in the top 10% of the g-force distribution.

Horseplay was consistently associated with high g-force events, whether the events were triggered by acceleration, deceleration, left or right turns. High g-force decelerations and left turns were also more common when loud conversation was present. Driver electronic device use and other distracted driver behaviors were not strongly related to high g-forces.

In fact, the general trend was for high g-force events to be less common when drivers were using electronic devices or engaging in other distracted behaviors.

Conclusions

This study is among the first to directly measure the occurrence of distracted driver behaviors and distracting conditions among teenage drivers. It also describes how distractions vary based on the presence of passengers, time of day, sex of the driver, and other potentially important factors.

Finally, it documents how distracting activities were related to several aspects of driving performance, including serious incidents. Similar to adults, teenagers engage in a wide variety of distracted behaviors while driving.

However, substantial individual differences were observed between teenagers in the frequency of distracted behaviors, and there was some evidence teenagers tempered these behaviors in a setting that places greater demands on the driver (rainy conditions).

The study also provides insight into the increased crash risk for teenage drivers when carrying passengers. The presence of teenage peers – especially multiple peers – sometimes resulted in horseplay and loud conversation in the vehicle. Both horseplay and loud conversation were particularly common after 9 p.m. on weekends, a time when much of teen driving may be “recreational.”

By contrast, carrying parents – and to a lesser degree siblings – was associated with a substantially lower likelihood of horseplay and loud conversation. Potentially distracting conditions in the vehicle such as horseplay went hand-in-hand with serious incidents and high g-forces. However, causality cannot be inferred.

Carrying multiple passengers may have caused these incidents, but it is also possible that riskier drivers are simply more likely to carry multiple, rowdy passengers. Finally, electronic device use and other distracted driver behaviors were strongly associated with looking away from the roadway, although electronic device use was only weakly related to serious incidents.