Outline of Results, Methodology, and Data Limitations
Phase I Study on Distracted Driving

Below is a distribution of distracting activities. Listed next to the percentages are the 95% confidence intervals.

Distribution of Distracting Activities
Activity Percentage
Outside person, object, or event 29.4% (plus or minus 4.7%)
Adjusting radio/cassette/CD 11.4% (plus or minus 7.2%)
Other occupant 10.9% (plus or minus 3.3%)
Moving object in vehicle 4.3% (plus or minus 3.2%)
Other device/object 2.9% (plus or minus 1.6%)
Adjusting vehicle/climate controls 2.8% (plus or minus 1.1%)
Eating and/or drinking 1.7% (plus or minus 0.6%)
Using/dialing cell phone 1.5% (plus or minus 0.9%)
Smoking related 0.9% (plus or minus 0.4%)
Other distractions 25.6% (plus or minus 6.0%)
Unknown distraction 8.6% (plus or minus 5.3%)

Below is a listing of driver "attention status" for all crashes, based on weighted data from the National Highway Traffic Safety Administration's Crashworthiness Data System. Listed next to the percentages are the 95% confidence intervals.

Driver Attention Status
Attention Status Percentage
Attentive 48.6% (plus or minus 5.4%)
Distracted 8.3% (plus or minus 1.2%)
Looked but did not see 5.4% (plus or minus 1.4%)
Sleepy or fell asleep 1.8% (plus or minus 0.8%)
Unknown/no driver 35.9% (plus or minus 5.5%)

Study Methodology

The study analyzed data from the National Highway Traffic Safety Administration's Crashworthiness Data System (CDS). This data includes only information on crashes in which at least one vehicle was damaged severely enough to require towing from the scene. CDS crash investigators collect data from medical records, visits to the crash scene, examination of the vehicles and interviews with drivers and witnesses. It is important to note that the CDS analysis for this study was vehicle-based rather than crash-based, and thus almost certainly understates the role of driver distraction in crashes.

In 1995, CDS investigators began describing the attention status of the driver at the time of the crash. Specific categories of distraction are identified and coded. The current study used data from 1995 through 1999, a total of 32,303 vehicles. Some CDS reports also have narratives that describe circumstances of the crash; this information was included in the study.

Study/Data Limitations

Missing data:

The CDS data has a high percentage of "missing," "unknown," and "other" data. In spite of extensive investigations, driver attention status is "unknown" for almost 36 percent of the drivers. In addition, 34 percent of the drivers known to be distracted were coded as "other" or "unknown" distractions. Thus, present estimates for known distracting events probably understate their true magnitude.

Sample size:

Because of small sample sizes, the data have large standard errors when weighted to reflect national estimates. For example, the estimates for cell phone use are based on only 42 reported cases.

Reporting bias:

Some distractions may be more readily reported than others. Drivers may also be more willing to admit to certain distractions, since some distractions are more socially acceptable than others.

Exposure data:

It is not known how much time drivers engage in various distracting activities, so relative risk cannot be determined. The CDS data only provides information on how often each behavior is a factor in crashes. More research is needed to document the frequency, intensity, and consequences of real-world driver distraction. Understanding driver distraction is especially important in light of new in-vehicle technologies.