banner of Methodology for Determining Motorcycle Operator Crash Risk and Alcohol Impairment

3. Detailed Report of Discussion

Cohort Study

A cohort study would attempt to select and follow a representative sample of riders and record their drinking and riding behavior over time. This type of study has never been attempted or accomplished for motorcycle riders. BAC levels for trips that result in crashes, close calls, or   other measures of unsafe riding are compared to BAC levels for trips that do not. Recent advances in technology have made this more of a possibility than it may have been in the past. Devices such as alcohol interlocks used to prevent impaired drivers from starting and operating cars can be modified for use in a cohort study of motorcyclists. These devices could require that riders provide breath measurements before starting the motorcycles. If a breath measurement is not provided, the motorcycle would not start, but regardless of the outcome, once a breath measurement is provided, riders could proceed.  If the motorcyclist was prevented from riding after drinking, subsequent rider behavior could not be recorded. Other on-board instrumentation could be used to record physical conditions that indicate crashes or close calls (e.g., moving motorcycle becomes stopped on its side). Global Positioning Satellite (GPS) equipment could be used to record such information as crash locations and speeds. Participation might be increased through the incentive of providing the instrumented vehicles to riders for free, at a discount, or as a loan for the period of the survey. Alternatively, surveys could take the place of much of the technology mentioned for data collection. Reports of drinking behavior or crashes/close calls could be elicited through surveys or recorded in logbooks. It may be possible to get more accurate data by using frequent phone surveys (e.g., twice weekly) rather than less-frequent surveys (e.g., monthly mail-in surveys). Internet survey instruments might also be used. Issues of liability and human subject protection would significantly affect the feasibility of such a study. For example, some would consider it unethical to create a situation where riders can record a high BAC on an interlock device that then permits them to start the motorcycle and ride it. If instrumented motorcycles or other incentives are provided only to subjects who can be expected to occasionally ride after drinking, there is potential to view the study as encouraging drinking and riding. The fact that BACs are being recorded also creates a potential legal issue in the case of a crash.


If such a study could be done with a sufficient number of participants and highly accurate data, it could be the most accurate way of determining relative risk curves for motorcycle operation.

Participating riders have the potential to be both the crash case and the comparison case. There is a potential for recording far more data than could be obtained using archival data. These additional data could strengthen the study and may also be of interest to potential project sponsors (e.g., motorcycle manufacturers). Even if much of the data come from surveys of participants (rather than automated data collection via instrumented vehicle), the response rate would likely be far higher than for the Survey studies described below, because participants have agreed in advance to participate.


As with any study in which participants are aware that they are being observed, participation in the study may change the behavior it is attempting to study.

There is no guarantee that all pertinent data would be collected. If an interlock device is used, participants may attempt to circumvent it when they have been drinking, or they may choose to use other forms of transportation, including different motorcycles, when planning an outing that may involve riding after drinking.

Other people could use the participant’s motorcycle. If an alcohol interlock is used to capture BAC data it might take advantage of technology which requires the user to learn a coded system of blowing in order to start the engine. Still, it would be possible for someone who borrows the motorcycle to learn the system, or for the owner to blow for another rider.

Participants may not accurately report crashes and close calls.

Any other data collected through a survey has the potential to be inaccurate if the subject incorrectly recalls or misreports the facts. In particular, if BAC measures were collected using survey methods, the data would be no more accurate than for the survey study (described below).

There is a potential for self-selection bias due to some riders agreeing, and others not agreeing, to participate in the study. There is also the problem of identifying potential subjects in a way that does not bias the sample.  For example, recruiting subjects through licensing rolls would preclude having unlicensed riders in the study.

To the extent that data are collected by means of survey instruments, the study would require  OMB clearance. The use of human subjects creates a requirement for review by an IRB committee, which may choose not to permit it. Steps would need to be taken to prevent survey data from being used against participants in the case of a crash. To use crashes as a dependent variable, many riders and/or many years of data may be necessary to obtain a large enough sample of crashes. A study that takes several years of data collection is not very attractive to agencies that want the research question answered as soon as possible. Use of instrumented vehicles also creates cost/maintenance issues.


The cost of such a study would vary depending on the extent to which automated data collection systems are involved. If instrumented vehicles are used, it may be that motorcycles would be purchased and instrumented out of project funds and provided to study subjects as an inducement to participate. Cost could conceivably include purchasing and maintaining motorcycles, alcohol interlocks, crash detection equipment, GPS equipment, and other potentially expensive technologies. Prices of some of these technologies may come down in the future. Costs might also include incentives for subjects to participate.

Another element that adds to the potential cost of this methodology is the large number of people and/or time needed to generate enough crashes or close calls to have significant results.