Science-Based Road Safety Research

Module 1.5

Science-Based Road Safety Research

Module 1.5

Science-Based Road Safety Research

Learning Objective Duration 45 Minutes

At the conclusion of this module, participants will be able to:

Describe and value science-based road safety research and its application as fundamental to achieving further improvements in road safety.

NCHRP 17-40, June 2010

Page 1 of 9

Module 1.5

Science-Based Road Safety Research

In this module we will discuss several topics including:

·  Traditional methods

·  Engineering and public health science-based methods

·  Quality data and data systems

·  Data driven approaches to road safety

·  The importance of professional support for science-based approaches

Show the goal of safety research. The goal of safety research is to illuminate ways to reduce the negative externalities associated with motor vehicle related crashes such as fatalities, injuries, property damage, and congestion.

Ask: What words could be added, deleted, expanded, etc? Do you have any questions about this statement?

Science is not the only method used in making transportation investment decisions. Other methods include expert opinion, conventional wisdom, and collective experience. However, the goal of this module is to illustrate that a scientific approach will result in the most measurable and positive outcomes.

Improving road safety in the long-term requires the careful measurement and analysis of the transportation system. There are several important reasons that careful and deliberate analysis is the only justifiable method for pursuing improvements to road safety:

•  Crashes by themselves vary randomly across time and space. Thus, a system location in any given observation period (e.g. an intersection during the last year) could record a very high number of crashes yet be relatively safe. Conversely, a site (e.g. pedestrian crossing) could record a very low number of crashes in a given period (e.g. school year) and be considered high risk.

•  Guesswork about the causes and contributing factors of road crashes are often incorrect.

•  Personal experiences and our limited ability to internally process data can lead to wrong conclusions.

•  Competing methods, such as expert opinion, conventional wisdom, and collective experience can at times lead to counter-productive recommendations.

As a result, road safety professionals need to develop and rely on state-of-the-practice data collection, management, and analysis techniques to make improvements in road safety and to reduce the fatalities, injuries, and property damage associated with motor vehicle crashes.

Because transportation is so prevalent to everyday experience, it is quite easy to believe that expertise is obtained through personal observation. This is readily evident when attending public meetings and members of the general public claim to know how to solve transportation problems, including congestion, air quality, and safety.

Personal observation can skew the view of the transportation system. If important safety decisions are based on such experiences, the result is typically far from optimal. For example, the general public typically reacts to a school bus related crash with significant horror and shock, and often demands that safety belts be installed in school buses.

However, careful analysis has repeatedly shown that installing safety belts in school buses is one of the least cost-effective safety investments and in some cases may even be detrimental to safety.

Individuals have notoriously poor intuition when it comes to processing statistical information. That is, people generally do not have the ability to estimate risk probabilities and appropriately consider relative risks from experiential data.

For example, many individuals report being far more fearful of a terrorist attack than being killed in a motor vehicle crash — both of which are quite unpredictable events. In 2001, for example, about 2,992 people in the US died from terrorist attacks mainly because of the attack on the World Trade Center on September 11, but 42,196 people died on US roadways. In subsequent years the number of US citizens killed in terrorist attacks declined considerably while road fatalities remained relatively constant until the economic problems in 2007.

Examples of how poorly Americans assess risk in their lives include: (other earthly citizens perform similarly):

•  Americans worry (and the media worries) about the Mad Cow disease pathogen in hamburger meat (no US deaths yet reported), yet cholesterol contributes to heart disease and kills 700,000 Americans annually.

•  Americans (and media) worry about Avian Flu (no US deaths to date), yet the common flu contributes to 36,000 deaths annually.

•  20% of all adults smoke;

•  About 20% of drivers and 30% of backseat passengers don’t use seat belts; and

•  67% are overweight or obese.

Risk assessment is complicated by a number of complex behavioral and physiological factors, including:

We exist in a modern world with a pre-historic brain: the brain has not yet grown accustomed to living in a predator-free environment, and the central nervous system has not kept pace.

Our need for the ‘fight or flight’ response to risk has diminished (where the brain works very well but is not now needed), while the more longer-term risk assessment part of the brain is less well adapted (where it is most needed).

Some risks have short-term benefits or pleasures, and some are willing to trade short-term gains for long term risk. Risk is only a risk if you are exposed, and different market segments face risks differently (e.g. skydiving, scuba diving, flying in an airplane, Ebola, motorcycle riding, etc.)

Why science? The ultimate purpose is to learn and test new knowledge. But, what do we mean by science?

“In short, the justification for basic research in all fields lies in the knowledge-generating utility of scientific discoveries and in the well-founded anticipation—but not guarantee—that some of those discoveries will in the long run prove to be of great practical benefit.”

What sets science-based approaches apart from other methods of learning about the world?

·  Public scrutiny is widespread, making it more probable for inappropriate practices/methods/ conclusions to be detected and exposed (peer review process).

·  Scientists acknowledge a willingness to do without answers when satisfactory ones are not available.

·  Scientists recognize that doubt is not a pleasant outcome, and that certainty is an absurd one.

·  Scientists are sometimes willing to question the obvious, whereas most accept the obvious without question.

·  Scientists have the ability to test among competing answers to questions.

So if science is best, what characterizes the scientific process? The scientific process generally consists of the following steps:

1. Review the literature of existing theories and knowledge of the process under investigation. What is already known about the process under study (e.g., how, why, and when do safety restraints protect occupants)?
2. Postulate relationships (theories) between variables based on the research objectives. You must articulate what you want to learn/study (e.g., we believe safety restraint use differs across the population due to variation in the motives for using restraints).
3. Generate testable research hypotheses. How can you test your theory (e.g., we need to collect restraint-use rates and factors we believe affect it)?
4. Design a study (observation, survey, experiment) to obtain data appropriate for testing the relationships/theories.
5. Collect data.
6. Apply appropriate statistical methods and/or models to determine whether the data support or refute the hypotheses.
7. Correctly interpret the results of the analysis.

We have learned much about how to conduct steps six and seven and produce reliable and consistent results. (We will examine these steps further in Unit 4.) One aspect of conducting science in the public health and motor vehicle safety arena is that investigators rarely have the opportunity to conduct experiments. They are expensive and often require laboratories with expensive equipment, such as simulators, etc. Consequently, observational data dominates studies in our profession.

We conduct observational studies in environments where few factors that affect the outcome of interest can be controlled. Observational studies result in significant challenges for being able to arrive at definitive conclusions from research.

Examples:

·  In sampling restraint use, some vehicles have tinted windows and cannot be sampled. Other vehicles drive by the sampling location several times. This generally goes un-noticed; thus, those vehicles are counted more than they should be. Little is known about nighttime safety belt use, but it is estimated to be lower than daytime use. However, nighttime safety belt observations require special equipment and can be dangerous to operate.

·  In the study of crash occurrence, non-crashes are not generally observed or recorded, nor is the successful avoidance of crashes included in the sampled data.

·  For drivers who speed (a significant percentage in many studies), there are far more cases where speeding does not precede a crash compared to cases where speeding does precede a crash. In other words, many drivers exceed the speed limit regularly without getting involved in a crash, suggesting that speeding alone is not always sufficient cause for a crash but instead needs to be combined with other risk factors such as following too closely, driver inattention, driver fatigue, etc.

·  The percentage of commercial trucks can influence motor vehicle crash occurrence but cannot be controlled in the real world unless commercial vehicles are prohibited from using a specific road.

The importance of data quality cannot be overstated. This section highlights the extent to which science-based systems depend on high quality data and presents some important qualities of data collection that ensure high-quality data are obtained.

Science-based approaches for conducting safety research create a need for quality data and data systems. These science-based systems span all the way from the time of a safety event (e.g. a motor vehicle crash) to the management and provision of data to potential stakeholders via the Internet.

Science-based methods are used, increasingly and broadly, throughout the entire sequence of events surrounding a crash, and science is being applied broadly in the supporting technologies needed to effectively manage a crash. Some of the ways science and technology rely on high quality data are outlined below.

•  Methods for reporting crashes - cell phone reporting, 911 systems, in-vehicle systems

•  Methods for responding to crashes - dispatching systems and methods, technologies for conveying/ reporting information from the ambulance to emergency rooms (in-ambulance camera systems and health monitoring systems), and methods for delivering crash victims to emergency rooms (air-lift programs, signal pre-emption programs, and efficient routing programs)

•  Methods for collecting victim and crash information - improved paper collection and electronic data collection at the scene and emergency medical services, emergency room, inpatient, and outpatient hospital records

•  Methods for managing and storing crash information - systems for managing, storing, and assuring quality of crash records

•  Methods for linking crash records with hospital records - probabilistic linkage of records, such as the Crash Outcome Data Evaluation System or CODES

•  Methods for analyzing crashes - before-after evaluations, observational studies, case-control studies, safety audits, etc.

•  Methods for disseminating crash information to stakeholder agencies

•  Methods for educating safety professionals

We will discuss data in far greater detail in unit 3.

Analysis of data is required to pursue reductions in motor vehicle related fatalities and injuries. Since the 70s, road safety professionals have seen steady reductions in the risk of driving normalized by the amount of driving, even though the total number of fatalities has remained relatively constant despite the fact that population and miles driven have continue to climb. Careful analysis of countermeasures is imperative to convince decision makers to support their implementation.

Several reasons exist for the U.S. experiencing a reduction in the relative risk of driving, but each of the effective countermeasures and programs relied on data collection and analysis to become accepted by decision makers. Scientific research with proven results is the antidote to political persuasion.

Question: Can you think of examples where scientific study has led to crash reductions?

Answer: Science has produced safer vehicles, e.g. airbags, safety belts, antilock brakes, electronic stability control, and in general better vehicle crash worthiness; safer drivers, e.g. tougher laws pertaining to safety belt and child safety seat use and impaired driving, improved driver education and licensing programs, and the graduated driver license laws; and safer roadway factors, such as cable median barrier to prevent cross-median crashes, enhanced pedestrian guidance systems, more forgiving roadside environments, improve crash barriers, such as guard rails, rumble strips and rumble stripes, and the safety edge.

General support for science-based road safety research would quickly become a priority if there were widespread support within the safety community. Unfortunately, this is not the case now. Many are unaware of appropriate scientific techniques for identifying and articulating problems or evaluating the outcomes of safety programs and countermeasures. That is part of the reason this course was developed; to create general awareness of and support for science-based approaches to the work of road safety.

The support of science-based research by safety-related organizations requires several prerequisites:

1. The need for the organization to assess safety (at any level) as part of their charge/mission.
2. The ability to assign, appoint, or hire a champion to the cause. A person (e.g. part-time or full time) or staff of people who can dedicate time to analyzing and assessing safety.
3. The resources necessary to educate and fund the staff and to provide the computer hardware and other resources necessary to support safety related activities.

DOTs, highway safety offices, other state agencies, metropolitan planning organizations, federal agencies, universities and colleges, professional member associates, industry associations, and many others can and should be recruited to support greater attention to road safety research.

The purpose of this module is to explore science-based approaches to safety research and evaluation as well as encourage the use of these methods when making transportation investment decisions.

The weaknesses of other methods, e.g. conventional wisdom, engineering judgment, expert opinion, and collective experience, were explored. These methods often produce results that are incorrect and counter-productive.

Science is a superior option because researchers are well trained in the use of appropriate scientific methods, and their results are subject to scrutiny by not only the research community (e.g., peer review), but also the wider public. Scientists are more likely to question conventional wisdom and have the ability to test among competing alternatives. Ultimately, scientists do not support unsatisfactory answers, even if satisfactory answers are not available.