Effectiveness of Behavioral Highway Safety Countermeasures (2008)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5Behavioral change techniques used in highway safety countermeasures basically fall into one of the following four categories: 1. Voluntary action (public information, education, mass media, training); 2. Laws, regulations, policies; 3. Laws plus enhancements (enforcement plus publicity); or 4. Sanctions and treatments (fines, points, jail, alcohol school, license suspension). Changing Driver Behavior There are many issues and challenges that need to be con- sidered with respect to changing driver behavior. One issue is that safe driving practices and protective behaviors (such as helmet use) have to be practiced on each trip. Measures that have only a short-term effect with no lasting behavioral change contribute little. Long-term effects are much harder to achieve than immediate ones, and some behaviors are more difficult to implement than others. For example, converting motor- cyclists to helmet use requires them to purchase a helmet and always wear it, which may seem uncomfortable and intrusive. Potential users may be opposed to helmet use for intellectual and emotional reasons. In comparison, wearing an already available seat belt, which is less intrusive than a helmet, should be an easy sell. In general, most people know what they are supposed to do on the highway in terms of safe driving practices; it is not a matter of lack of knowledge. What people actually do, how- ever, is guided by attitudes, motivations, lifestyle factors, and assumptions about risk. Veteran drivers have well-developed habits that pose a challenge to change. Moreover, from risk perception research it is known that in very familiar activities such as driving, there is a tendency to minimize the possibil- ity of bad outcomes as a way of allaying personal concerns (Douglas, 1985). People underestimate risks that are supposed to be under their control, insulating themselves by creating “illusory zones of immunity” around everyday activities (Jasanoff, 1998). This sense of subjective immunity is bol- stered by the belief of most people that their own driving skills are superior (Williams, Paek, and Lund, 1995). Crashes happen, but to other drivers; the highway safety problem is a problem of the other driver. We want those other drivers to behave on the highways since they are a threat to us and, in that context, safety messages are for them, not us. In general, people have an opti- mistic bias, thinking that they are less likely than others to suf- fer misfortunes (Maibach and Holtgrave, 1995). Therefore, it is not surprising that this so-called “third-person effect” is found in a number of health realms. People viewing health messages believe the message is for others, not themselves (Davison, 1983). This is likely to be even more of a factor in regard to highway safety, given the psychological tendency of people to protect themselves by minimizing the possibility of harm to themselves resulting from the everyday activity of driving. Finally, crashes, especially those that produce injuries, are extremely rare events per mile driven. Speeding, driving while impaired, running red lights, and other dangerous and illegal behaviors generally have no downside. In this sense, drivers are rewarded every time they complete a trip involving these actions. All of these factors, taken together, pose significant barriers to influencing driving behavior. That said, some groupings can be expected to have more effective countermeasures than others. For example, laws are generally more effective than requesting voluntary actions in terms of producing behavior change; enhanced laws should be more effective than laws alone. Within each of the four cate- gories, however, some countermeasures work and some do not. Cautions About the Countermeasures As indicated earlier, one issue in determining the effec- tiveness of a countermeasure is that one type of program or C H A P T E R 3 Countermeasure Classification

approach involving that specific countermeasure may work, and another may not. This can happen in all four of the countermeasure groupings. For example, in the laws area, it is possible that a law that works in one state will not work in another, since laws (e.g., seat belt laws, administrative license revocation [ALR] laws) vary in terms of coverage and penal- ties. In the laws/enhancements area, enforcement programs can vary in intensity and duration, and may be differentially effective. In the treatment/sanctions area, alcohol treatment programs can differ markedly. However, this is most likely to be an issue in the grouping for voluntary actions only where programs promoting a specific action can vary widely, rang- ing from a passive public information campaign based on materials sent through the mail to multiple face-to-face in- teractions involving sophisticated behavior change models, and possibly involving other inputs. The latter may work; the former may not. Another warning concerning effectiveness ratings was raised in Countermeasures That Work (NHTSA, 2007b), namely that evaluation studies generally examine and report on high-quality implementations of countermeasures, so that the effectiveness data are likely to show the maximum effect that can be realized. That is, the countermeasure in question may not work, or work as well, with lesser efforts. Also, it should be noted that while a particular approach may not work by itself, it may facilitate acceptance of an approach that will reduce injury (e.g., public information and education [PI&E] may affect public acceptance making passage of a law more likely). The remainder of this chapter will discuss each counter- measure category in turn, along with the criteria derived that distinguish effective and ineffective countermeasures within that category. Then, each countermeasure within that group will be rated as follows: • Proven effective; • Likely to be effective; • Effectiveness is Unknown/Uncertain/Unlikely; or • In a few cases, Proven Not to Work. Countermeasure Categories Class 1: Voluntary Action A popular approach in the behavioral field has been to urge people to take appropriate actions through public informa- tion, educational programs, mass media, and training used alone. Given the barriers to change discussed earlier, it is not easy to change driver behavior in this manner. This subject is treated extensively in Public Information and Education in the Promotion of Highway Safety (Williams, 2007a), which forms the basis for the following discussion. Most of the countermeasures in the Voluntary Action Group involve communications. Historically, many of these efforts have been of poor quality, consisting of passive messaging, sloganeering, exhorting people to do—or not do—some behavior, and delivered to an undifferentiated audience over the short term. The simplistic assumption is that if individuals are made aware of behaviors that will enhance their personal health or safety and urged to adopt these behaviors, they will do so. Seemingly logical, this sequence of events is unlikely to happen. It is well established that information-only programs are unlikely to work, espe- cially when most of the audience already knows what to do. Therefore, highway safety messages conveyed in signs, pam- phlets, brochures, on buttons, etc. may increase awareness of the health issue being addressed and reinforce social values, but are unlikely to have any effect on behavior. Behaviors that are particularly difficult to change, such as getting a motor- cyclist to buy and use a helmet, are least likely to be affected by advice or urgings to do so. Lecture-oriented education programs that are information- only in nature also are likely to be ineffective, as are short- term programs and messages delivered only once or twice. Extreme fear or scare techniques also are likely to have no more than a short-term emotional effect, especially when directed at adolescents. Programs recommending driver behavior that are more likely to be effective include public information programs that involve careful pre-testing of messages to make sure the message is relevant to the group being addressed and care- ful delineation of the target group to make sure the messages reach the target group in sufficient intensity over time. These are the aspects involved in successful social market- ing programs. In the education arena, some success (mostly in other health areas) has been achieved through programs using theory-based behavior change models, and interactive methods to teach skills to resist social influence through role playing, behavior rehearsal, group discussion, and other means. However, even high-quality public information and education programs rarely work by themselves to change in- dividual behavior, although their contribution can be criti- cally important when combined with other prevention efforts (e.g., in support of law enforcement or as part of broader community programs). According to the research literature (Williams, 2007a), programs involving voluntary actions that work on their own include those targeting children, whereas programs targeting teenagers or adults are not likely to work. Unlike adults, children do not have well-developed safety behavior patterns and so are more amenable to change. Programs also work that communicate health knowledge not previously known. One example of this is the shift of children from front to rear seats to avoid air bag 6

inflation dangers, a “new” knowledge that was largely driven by public education programs. Programs where the com- municator has some control over resources or over the audience also are more likely to be successful. These would include employer programs, parents influencing their chil- dren, and alcohol servers influencing patrons. Finally, high- quality public information and education programs that are part of broad-based community programs have also been successful. The 38 voluntary action countermeasures (the largest group of any of the four categories) are listed below, sorted according to their effectiveness rating. Note that this group also includes three items that research has clearly shown do not work to reduce crashes and, in fact, can increase them: novice driver education (when that education leads to licensure at an age which is younger than would otherwise be the case without the education), skid training for novices, and traffic viola- tor school in lieu of penalties. Regarding the category of Unknown/Uncertain/Unlikely, see Appendix A for the ration- ale and references to further separate this group into: (+) some basis for thinking that it might work; (0) unknown or no opinion; and (−) some basis for thinking that the counter- measure will not work. Proven • School pedestrian training for children; • Programs to get parents to put children in rear seats; • Booster seat promotions; and • Child bicycle helmet promotions. Likely • Responsible beverage service and • Parent guiding teen licensing. Unknown/Uncertain/Unlikely • Child pedestrian supervision training for caregivers (+); • Child safety clubs (+); • Bicycle education for children (+); • School-based alcohol education programs to reduce drink- ing and driving (0); • PI&E for elderly drivers (−); • PI&E for low belt users (+); • Motorcycle education and training courses (−); • Formal driver education courses for elderly drivers (−); • Bike fairs, rodeos (+); • Driver training about sharing the road with bicycles (−); • Teaching bike rules/safety in driver education (−); • Education encouraging bicyclists to increase their con- spicuity (−); • Education to encourage pedestrians to increase their con- spicuity (−); • Driver education in regard to pedestrians (−); • Programs to teach driver awareness about motorcyclists (−); • PI&E about driver fatigue (−); • PI&E about distracted driving (−); • PI&E on sleep disorders for general population and phy- sicians (−); • Employer programs for shift workers, medical interns (+); • Alternative transportation for alcohol-impaired drivers (+); • Designated driver programs (0); • Motorcycle helmet use promotion programs (−); • PI&E on drinking and motorcycling (−); • Education to encourage motorcyclists to increase their conspicuity (−); • Programs to help police detect impaired motorcyclists (0); • Communications and outreach regarding impaired pe- destrians (−); • Extreme fear and scare tactics in youth programs, e.g., fake deaths, mock funerals (−); • High school driver education (not leading to early learning/ licensing) (0); and • School bus training for children (+). Proven Not to Work • High school driver education (leading to early learning/ licensing); • Advanced driver education, skid training; and • Traffic violator school in lieu of penalties. Class 2: Laws, Regulations, Policies Many of the demonstrable gains in changing behavior in ways that reduce motor vehicle injuries have come through laws and regulations. The power of laws is illustrated by the abrupt changes in behavior that occur coincident with their introduction. For example, on the day British Columbia’s seat belt use law went into effect, belt use was 30 percentage points higher than it had been 24 hrs earlier (Williams and Robertson, 1979). Not all laws work, however. Laws that work best incorpo- rate elements associated with high deterrent capabilities. That is, they are well known to the public, and they are enforceable laws, based on easily observable behavior and objective criteria (e.g., motorcycle helmet use laws). This leads to the expectation that not complying with the law will result in apprehension and sanctioning. Also advantageous are laws where enforcement is done not only by the police, but by parents (e.g., bicycle helmet laws for children, or grad- uated licensing laws for adolescents). Department of Motor Vehicles (DMV) rules that have to be followed, and ordinances 7

and other across-the-board policies also are more likely to work. Policies work that force changes that result in positive outcomes. For example, motorcycle helmet laws force riders to wear a helmet. Laws less likely to work on their own are those that are not well known, or for which the behavior is not easily observ- able by police and therefore not easily enforced (e.g., open container laws). Laws that apply only to a portion of the pop- ulation performing the behavior (e.g., motorcycle helmet laws that apply only to young motorcyclists) are difficult to make effective, especially when the penalties are weak. Laws where the criteria are not explicit also are less likely to be successful (e.g., aggressive driving, fatigue, and distracted driving laws). Proven • Bike helmet laws for children; • Graduated driver licensing (GDL); • Extended learner permit; • Night restrictions (for young drivers); • Passenger restrictions (for young drivers); • Administrative license revocation laws; • BAC test refusal penalties; • Primary seat belt law; • Speed limits; • Motorcycle helmet laws; and • Reduced speed limit regarding pedestrians (proven in Europe). Likely • Ice cream vendor ordinance; • Local primary seat belt laws; • Adult bike helmet laws; • License renewal policies for elderly drivers; and • License actions for underage alcohol violations. Unknown/Uncertain/Unlikely • General cell phone laws (+); • Open container laws (0); • Lower BAC limit for repeaters (+); • Cell phone laws as part of graduated licensing (+); • Belt use as part of graduated licensing (+); • Motorcycle licensing laws, especially in regard to having a valid license (0); • Belt laws with significant exclusions (0); • Keg registration laws (0); • Medical advisory boards for elderly drivers (0); • Aggressive driving laws (−); • Driver fatigue and distracted driving laws (−); • Referral of elderly drivers to licensing agencies (+); • Licensing screening and testing for elderly drivers (+); and • Licensing restrictions for elderly drivers (+). Class 3: Laws Plus Enhancements If the public to whom the law applies is not aware of the law, or there is little enforcement, or little perceived enforce- ment, positive effects of laws can be diminished or eliminated. Thus, the effects of laws can be enhanced by special enforce- ment programs, publicity about the law and its enforcement, and—in some cases—by special equipment such as passive alcohol sensors to enhance enforcement. When one or more of these elements is combined with laws that are easily enforce- able, success is likely. Success is less likely when laws are not easily enforceable because the criteria for enforcement are vague or the behavior is difficult to observe. Proven • Sobriety checkpoints; • Saturation patrols for alcohol-impaired driving; • Preliminary breath test devices; • Passive alcohol sensors; • Short, high-visibility belt law enforcement; • Automated enforcement for speed, red light running; • Mass media support of alcohol enforcement or other programs; • PI&E supporting enforcement of seat belt laws; and • Community programs, including age 21 enforcement. Likely • Integrated enforcement (alcohol, seat belts, speeding); • Zero-tolerance enforcement; • Vendor compliance checks for age 21 enforcement; and • Sustained seat belt enforcement. Unknown/Uncertain/Unlikely • Aggressive driving enforcement (+); • GDL enforcement (+); • Enforcement of pedestrian rules targeted to drivers and pedestrians (−); • Enforcement of bike rules (−); and • Enforcement against unapproved motorcycle helmets (+). Class 4: Sanctions and Treatments Special penalties and treatments also can supplement laws. Sanctions that are well known to violators, have a high 8

probability of being imposed, and have a high degree of in- trusiveness (i.e., involve a real amount of money or time) are most likely to work. If there is low intrusiveness, if sanctions are not well known to violators, are unlikely to be applied, or if the penalty is not very meaningful, success is unlikely. Proven • Aggressive driving, speeding penalties (e.g., suspension, warning letters); • Restrictions on plea bargains; • Court monitoring; • Mandatory attendance at alcohol treatment; • Close monitoring of DUIs; • Alcohol interlocks; • Brief interventions—alcohol; • License plate impoundment; • Vehicle immobilization; and • Vehicle impoundment. Likely • Increased belt use law penalties and • Simplifying and streamlining DUI statutes. Unknown/Uncertain/Unlikely • Vehicle forfeiture (+); • GDL penalties (0); • Driving under the influence (DUI) fines (0); • DUI jail (0); • High BAC sanctions (+); and • DWI (driving while intoxicated) courts (+). Summary Overall, 45% of the 104 countermeasures are considered effective (33% proven; 12% likely). By comparison, the other 55% are less likely to work. This assessment is based on the fact that evidence for effectiveness is uncertain or unknown and/or the criteria for what is likely to work are not met (52%), or because research indicates that these countermea- sures increase crashes (3%). The following percentage of countermeasures are rated Proven or Likely to work in each class of countermeasures: • Sixteen percent of Class 1: Voluntary Action; • Fifty-three percent of Class 2: Laws, Regulations, Policies; • Seventy-two percent of Class 3: Laws Plus Enhancements; and • Sixty-seven percent of Class 4: Sanctions and Treatments. By topic area, there are differences in expected effective- ness. Effectiveness is most likely in the occupant restraint group where 82% of the countermeasures are rated Proven or Likely; followed by alcohol (67%); aggressive driving/ speeding (50%); young drivers (38%); bicycles (33%); pedes- trians (30%); elderly drivers (14%); motorcycles (11%); and distracted/fatigued drivers (none at this time). How effective are those countermeasures rated as Proven or Likely? Ideally, for all countermeasures rated as Proven, and for many rated as Likely, it would be possible to derive a numerical estimate of the effect size, the expected percentage reduction in injuries. However, it is not always possible to estimate this num- ber. Of the 47 countermeasures rated as Proven or Likely, about half of the outcomes relate not to reductions in crashes or in- juries, but to some intermediate measure (e.g., reductions in re- cidivism, increases in arrests or convictions, decreased drinking, increases in seat belt use). It is possible to estimate the impact of increases in seat belt use to decreases in injuries, but for many other intermediate measures, there is no credible way to do so. There also are a few cases where the expected effect relates to crashes or injuries, but not enough information is available to extract a numerical estimate of the effect. In addition to the effect size, there are other important factors in determining the overall impact of any countermea- sure. One of these factors is the size of the population affected. For example, a measure affecting the general population can have more impact than one affecting a specific subgroup (e.g., teenage drivers only). Another is the expected duration of the effect. For example, although the effects of laws can vary over time, depending on such factors as the amount of public- ity and enforcement, their permanence gives them an advantage compared with programs that are one-time efforts. Duration can also refer to the length of time the positive effects of a pol- icy last on individuals affected, for example, license suspension. 9