IV. What Caused the Decrease?

This chapter examines several factors that may have contributed to the youth drinking and driving decrease. These factors include very specific legislation directed at youth drinking or driving after drinking, programs to provide youth with the knowledge, skills, and attitudes to make healthy choices in many areas including drinking and driving, and broad factors that may have influenced youth attitudes and behavior. Some of these factors have been evaluated extensively with high-quality studies using accurate data; others have not. As a result, the effects of some factors are known quite accurately while the influences of others are highly speculative.

A. Minimum Legal Drinking Age 21 Laws

History of the laws

In 1933, the 21st amendment to the United States Constitution repealed the prohibition of the manufacture and sale of alcoholic beverages, thereby allowing each state to regulate how and by whom alcohol could be consumed. At that time, most states set the minimum legal drinking age at 21, the standard legal age of adulthood. New York opted for an MLDA of 18 and a few states had a 21 year-old legal age but permitted certain beverages (low strength beer) to be purchased at age 18. Additionally, some states continued for some years to prohibit alcohol or allowed the prohibition of its sale as a local option.

The Vietnam era brought about a national change in the age of majority to 18 years and many states followed by reducing the legal drinking age. By 1982, only 14 states retained a MLDA of 21; the MLDA in the other 36 states was 18, 19, or 20.

In the 1970s and early 1980s MLDA laws became a traffic safety issue. Several studies in the 1970s showed that youth traffic crashes increased as states lowered their MLDA. In addition, the "blood borders" between states with different MLDAs caught public attention after highly-publicized crashes in which youth below the legal drinking age would drive to an adjoining state with a lower MLDA, drink legally, and crash on their way home.

Advocacy groups urged states to raise their MLDA to 21. Several did so in the late 1970s and early 1980s, but others did not. To encourage a national drinking age, Congress enacted the Uniform Drinking Age Act in 1984, which provided that states that failed to raise their MLDA to 21 would lose a portion of their federal-aid highway construction funding. In response, all remaining states increased their MLDA to 21 by 1988. In terms of population coverage, most of the drinking age increase took place in a short time period. In 1985, about 30 percent of the U.S. population age 16-20 was covered by a MLDA 21 law. By 1987, over 90 percent were covered, and all were covered by 1988 (see Figure 28). Toomey, Rosenfeld, and Wagenaar (1996) summarize this history as well as many of the evaluation results that follow.

Figure 28. 
Percent of the U.S. Population Age 16-20 Covered by MLDA 21 Law

Graph of Figure 28. Percent of the U.S. Population Age 16-20 Covered by MLDA 21 Law

Effects of the Laws

The effects of drinking age law changes on traffic crashes, injuries, and fatalities have been studied extensively. These effects are relatively easy to evaluate for several reasons. Each law applied to all drivers in an entire state as of a specific date, so crash results can be compared within the state, before and after the law, and with other states that did not change their law at the same time. Each reduction or increase in a state's drinking age provided a new opportunity to evaluate effects. Finally, evaluations can use large traffic crash data files. In particular, FARS has provided uniform national data on fatal crashes since 1975.

The United States General Accounting Office (1987) reviewed and synthesized results from all 49 studies that had adopted MLDA 21 by 1986. They concluded that "raising the drinking age has a direct effect on reducing alcohol-related traffic accidents among youths affected by the laws, on average, across the states" and that "raising the drinking age also results in a decline in alcohol consumption and in driving after drinking for the age group affected by the law." They note that the traffic accident studies they reviewed were high-quality. While the studies used different evaluation methods, they produced "remarkably consistent" results. Additional studies since 1986 have reached the same basic conclusions (Toomey, Rosenfeld, and Wagenaar, 1996).

Two recent studies deserve special note. O'Malley and Wagenaar (1991) used FARS and Monitoring the Future data to investigate how the MLDA affects youth drinking and youth drinking and driving. They compared states with a MLDA of 21 to states with a lower MLDA (during the years before all states raised their MLDA to 21). They also compared behavior in states that changed their MLDA and states that did not. Among their conclusions:

In short, O'Malley and Wagenaar conclude that MLDA 21 laws reduce alcohol consumption, which in turn reduces crashes, and there is a carryover effect even after persons reach the age of 21.

Voas, Tippetts, and Fell (1999) use FARS data for all states from 1982 to 1997 to estimate and compare the effects of MLDA 21, zero tolerance, and other traffic safety laws (per se, administrative license revocation, and safety belt use). They find substantial effects for both MLDA 21 and zero tolerance laws.

Reasons for the Effects

The evidence reviewed briefly above shows unequivocally that MLDA 21 laws reduce youth drinking and driving, as measured by traffic crash involvements. But the way in which MLDA 21 laws have produced these results may not be completely straightforward. The laws make it illegal for youth to purchase, possess, or consume alcohol (individual state laws differ in precisely what they prohibit). But, much as national prohibition did not stop drinking, MLDA laws have not eliminated alcohol use by youth: the data in Section IIIC show that most youth drink, and a majority drink at least monthly.

The basic method for implementing MLDA 21 laws is for anyone selling, serving, or otherwise providing alcohol to a young person to verify the person's age. Retail establishments (liquor, grocery, and convenience stores; restaurants, bars, taverns, sports arenas, etc.) can require that identification be checked. Still, many do not: for example, a 1991 study found that 97 out of 100 liquor outlets in Washington, DC sold alcohol to 17- and 18-year olds (Preusser and Williams,1991). In areas where identification is checked regularly, many youth have responded by acquiring false identification cards. Verifying a young person's age at parties and other informal gatherings is considerably more problematic. In all settings, identification checking is done most effectively when some organization (retail establishment, college, private club) is responsible for selling or providing alcohol and when that organization faces a substantial legal liability if they serve underage youth. But even then, false identification can subvert the MLDA.

Wolfson, Wagenaar, and Hornseth (1995) investigated MLDA enforcement in 1992. They interviewed law enforcement officers in four states (Kentucky, Michigan, Montana, and Oregon) regarding underage drinking, MLDA enforcement, and community attitudes regarding underage drinking. The officers reported that MLDA enforcement is not a community priority; indeed, they found a general acceptance of youth drinking in their communities. They reported that most underage drinkers obtained their alcohol from legal-age purchasers. When youth were arrested for MLDA violations, the officers felt that the penalties were light and were applied unevenly, so had little deterrent value. Impaired driving violations, on the other hand, were procedurally simpler to enforce and the penalty of a driver's license suspension or revocation was an effective deterrent. Overall, the officers felt that drinking and driving by youth had decreased in their communities over the past decade but that drinking by youth had not.


MLDA 21 laws clearly reduced youth drinking and driving. They appear to have done so both by reducing youth drinking directly and by encouraging youth to separate their drinking from their driving.

MLDA 21 laws reduced youth drinking both by reducing alcohol availability and by establishing the threat of punishment for alcohol use. Neither works particularly well in practice, as youth still can obtain alcohol relatively easily and underage drinkers are highly unlikely to be detected and punished. But both have had some effect.

But MLDA 21 laws probably had other effects beyond the straightforward prohibition and attempted punishment of alcohol use by youth. As listed in Chapter IIIB, 11 states have had MLDA 21 laws since the repeal of prohibition. These states also saw substantial reductions in youth drinking and driving after drinking in the 1980s. Furthermore, youth driving after drinking decreased more than youth drinking.

This suggests that MLDA laws may have helped influence youth attitudes about drinking and driving. The principal reason for raising the drinking age to 21 was to reduce traffic crashes. Some youth and some parents may have consciously or unconsciously absorbed some of these beliefs: that youth drinking is not a problem unless it results in dangerous actions, of which by far the most dangerous is drinking and driving. Underage drinking is generally accepted, but underage drinking and driving is not. The widespread debate over the legal drinking age also may have had some "spillover" effect in states where MLDA 21 was already in place.

However, the observations that youth drinking and driving decreased substantially more than youth drinking, and that youth drinking and driving after drinking both decreased in states which had MLDA 21 laws throughout the 1980s, suggest that MLDA 21 laws were not the only influence on youth drinking and driving during this period.

B. Zero Tolerance Laws

History of the Laws

Zero tolerance laws are a logical combination of Minimum Legal Drinking Age laws that prohibit drinking by youth and per se laws that make it illegal to drive with a BAC exceeding a specified level. If it is illegal for youth to drink, then it should be illegal for youth to drive with any positive BAC. A zero tolerance law sets a maximum BAC of 0.02 or less for youth and suspends or revokes an offender's driver's license. Most zero tolerance laws use an 0.02 BAC limit rather than an absolute 0.00 limit to allow for small measurement errors in BAC test instruments and to avoid challenges from youth who claim they have taken medication with small amounts of alcohol.

During the 1980s several states enacted zero tolerance laws applying to drivers under their MLDA of 18 or 19. Other states set a BAC limit for young drivers below the legal limit for older drivers but above the now standard 0.02 zero tolerance limit: for example, New Mexico's BAC limit was 0.05 for drivers under 18. In the National Highway Systems Designation Act of 1995, Congress used the same strategy as for MLDA 21 to encourage zero tolerance laws: states failing to enact a zero tolerance law by 1999, at a BAC limit of 0.02 or lower, covering all persons under 21, would lose a portion of their federal-aid highway construction funds. Most zero tolerance laws took effect between 1994 and 1998, and by 1998, all states had zero tolerance laws in place (see Figure 29).

Figure 29. 
Percent of U.S. Population Age 16-20 Covered by Zero Tolerance Law

Graph of Figure 29. Percent of U.S. Population Age 16-20 Covered by Zero Tolerance Law

Effects of the Laws

Several evaluations have shown that zero tolerance laws reduce youth drinking and driving crashes. Hingson, Heeren and Winter (1995) studied 12 states that lowered their BAC limit for some young drivers before 1991. They found a 22 percent reduction in single vehicle nighttime fatal crashes (a standard surrogate for crashes involving alcohol) in states with a 0.00 BAC limit, compared to a 2 percent reduction in comparison states; a 17 percent reduction in states with a 0.02 BAC limit compared to a 4 percent increase in comparison states; and no effect in states with a 0.04 to 0.06 BAC limit. Blomberg (1992) found an 11 percent reduction in crash-involved drinking drivers in Maryland. In six counties that publicized the zero tolerance law heavily, youth alcohol-related crashes decreased by 50 percent. Voas, Tippetts, and Fell (1999) found that zero tolerance laws reduced the proportion of underage drinking drivers in crashes by 24 percent.

Reasons for the Effects

As with MLDA 21 laws, the reasons for these clear zero tolerance law effects may not be completely straightforward. Zero tolerance laws begin with several advantages. They prohibit something that's clearly defined (driving after drinking any alcohol) and that is commonly understood to be dangerous. The public generally supports zero tolerance laws: for example, Balmforth (1999) reports 45 percent of all adults agreeing that BAC limits should be lower for drivers under 21. Driver license suspension or revocation often provides an administrative penalty (not involving the courts) that clearly motivates youth: loss of your driver's license is a severe punishment indeed.

However, actually enforcing zero tolerance laws is difficult. As with any impaired driving violation, police first must observe something in a person's driving that provides grounds to make a traffic stop. Young drivers at low BACs may not exhibit driving behavior that police can link with probable impairment. Ferguson, Fields, and Voas (2000) studied zero tolerance laws in five states and concluded that these laws have not changed impaired driving enforcement practices. This means that young drivers at relatively low BAC levels who are in violation of zero tolerance laws are unlikely to be stopped by police unless they commit some other traffic violation. Some zero tolerance law provisions do make it easier for police to issue a citation to a young driver. For example, in California a zero tolerance law citation can be issued at the roadside, using evidence from a preliminary breath test equipment; a citation for the standard driving under the influence of alcohol offense requires an evidential breath test.

Despite these difficulties, youth are quite aware that they risk being stopped and sanctioned if they drink and drive. Balmforth (1999) reports that males age 16-20 believe that about one-third of impaired drivers will be stopped by police; females believe about one-half will be stopped. And about 12 percent of males and 8 percent of females aged 16-20 reported that they were stopped by police for a suspected drinking and driving violation in the past year. It is not clear, though, that youth know about their state's zero tolerance law or understand that it prohibits driving after drinking any alcohol. Balmforth reports that only 19 percent of the driving population knew that the BAC limit was lower for youth, and only about 4 percent knew the correct BAC level for youth.

Three further studies shed some light on youth attitudes and behavior toward zero tolerance laws. First, Grosvenor, Toomey, and Wagenaar (1999) surveyed high school seniors in 15 upper Midwest communities on their drinking and driving attitudes and behavior. They found that 40 percent of the students reported a fairly high likelihood of being caught by police if they drove after drinking; 65 percent expected the penalty to be arrest or driver license suspension. Among binge drinkers, those who believed the probability of being caught after drinking and driving was high were less likely to drink and drive.

Grube and Voas (1996), on the other hand, find that driving after drinking by youth is influenced more by friends' approval or disapproval and by expectations regarding physical risks than by fear of arrest and sanction. They suggest that law enforcement's lack of influence may be due to the low likelihood that a drinking driver will be stopped by police.

Most recently, McCartt, Leaf, Preusser, and Farmer (2000) surveyed Florida high school juniors and seniors in 1996 and 1998, before and after the state's zero tolerance law. They found that self-reported driving when drinking and riding with friends who were drinking both decreased significantly after the law took effect, but not by substantial amounts: for example, the proportion reporting they had ever driven after drinking dropped from 23.7 percent to 20.0 percent.


Zero tolerance laws definitely have reduced youth drinking and driving. They likely did so for two reasons: by deterring youth through the fear of losing their driver's license if they drive after drinking, and also by reinforcing the broad community disapproval of driving after drinking. Zero tolerance laws likely strengthened the attitudes raised by MLDA 21 laws: underage drinking is normal and generally accepted, but drinking and driving is not. Zero tolerance laws directly address the ultimate goal of reducing drinking and driving, completely bypassing the intermediate issue of underage drinking.

C. Law Enforcement

The relationship between law enforcement, for MLDA and drinking and driving laws, and youth drinking and driving was examined further using data provided by the FBI. The data set contained the number of DWI arrests and the number of liquor law violation arrests of persons under 21 years of age in each state annually for 1989 through 1995. A general linear model was constructed using DWI arrests, liquor law arrests, year and state as independent variables and the number of youth drinking drivers involved in fatal crashes as the dependent variable. The model showed no significant relationship between these enforcement measures and youth drinking drivers involved in fatal crashes.

This result is not especially surprising as previous studies have not found a strong relationship between DWI arrest totals and alcohol-related crashes. It is generally agreed that laws are most effective in deterring the behavior they prohibit if the public believes that violators are highly likely to be arrested and punished. Good enforcement programs seek to increase the public's perception of enforcement levels, not just raise arrest levels. Some enforcement efforts such as checkpoints produce few DWI arrests but create substantial publicity; other strategies can raise arrests but have little or no effect on public perceptions. In short, arrest levels generally are not a good measure of public perceptions of enforcement.

D. School and Community Youth Programs

Minimum drinking age and zero tolerance laws attempt to reduce youth drinking and driving through deterrence: passing laws prohibiting the behavior, publicizing the laws to advertise the consequences of violating them, enforcing the laws and applying penalties to the violators - in short, by attempting to motivate youth through fear of legal consequences. The other major strategy used over the past 20 years is to motivate youth not to drink and drive through positive means: by education on crash and injury risks posed by drinking and driving and the effects of alcohol use and abuse, by providing positive role models that discourage alcohol use, by establishing youth norms that do not include alcohol, and by encouraging youth activities that do not involve or lead to alcohol use. This strategy typically is implemented through school or community programs.

History of Youth Traffic Safety and Alcohol Programs

SADD. The single best known youth program is SADD, originally Students Against Driving Drunk. SADD was founded in 1981 as a high school-based program to reduce youth drinking and driving. Its original model included school assemblies, a student club (called a SADD chapter), alcohol-free activities, a 15-session curriculum for use in the sophomore year, and a Contract for Life between youth and parents in which youth who have been drinking and need a ride home agree to call parents and parents agree to provide the ride. SADD maintained national and state offices and provided program materials to school SADD chapters but did not control state or chapter activities. Consequently, state SADD organizations and local chapters operated in rather different ways with goals and activities adapted to local needs and opportunities. SADD's national goals have expanded beyond drinking and driving to other health and safety issues, as shown by its name change to Students Against Destructive Decisions.

SADD grew rapidly. By 1994 there were SADD chapters at an estimated 16,000 of the nation's high schools. Some state SADD organizations were well-organized with an effective state coordinator; others were not. In some states, SADD activities evolved under another name, sometimes with a continuing association with SADD, sometimes not: STAND in Colorado and SAFTYE in Washington are two examples.

State youth programs. At the same time that SADD was expanding rapidly, state traffic safety activities directed at youth drinking and driving also grew, with assistance from federal funds. Each state receives federal funds each year for highway safety activities under Section 402 of the Highway Safety Act of 1966. States usually allocate some of these funds to youth programs. In addition, some states qualified for additional funds for several years in the 1980s and 1990s under the Section 408 and 410 alcohol incentive grant programs and used portions of these funds for youth programs.

A good view of how state impaired driving programs directed at youth grew, and what they included, can be found in a series of NHTSA assessments of how these federal funds were used (LaHeist, 1998). For these assessments, NHTSA selected one state from each of the 10 NHTSA regions. In each state, NHTSA traced how federal highway safety funds were used from 1980 to 1993. The information following, taken from these assessments, summarizes youth alcohol program activity across the 10 states and provides selected examples from each state.

Each of the 10 states developed and implemented a large number of youth impaired driving programs. Practically all were supported by federal funds. Programs usually began in a few areas and, if successful, spread statewide. One of the most popular was Project Graduation, which organizes alcohol-free prom and graduation celebrations at high schools and which was conducted extensively in each of the 10 states. SADD chapters were formed in most states. Several states developed media or other education programs for elementary and intermediate schools. All states conducted public education activities directed at youth drinking and driving. Many states integrated youth program activities into community traffic safety programs.

Colorado developed a "refusal skills" drug and alcohol education program which was part of the regular school curriculum by 1992. The first SADD chapter was formed in 1984. Colorado's SADD chapters reorganized in 1998 as STAND (Students Taking A New Direction) and have spread throughout the state. BACCHUS (Boost Alcohol Consciousness Concerning the Health of University Students) and GAMMA (Greeks Advocating Mature Management of Alcohol) chapters were active on every major college campus by 1994, promoting designated driver and Safe Spring Break programs and the National Collegiate Alcohol Awareness Week.

Connecticut began Project Graduation activities in 1986. Using the design and materials of Nationwide Insurance's "Prom Promise," 50 schools participated in 1994. A group of Middletown police officers, calling themselves the "Blue Crew," produced an award-winning rap video and poster on drinking and driving that was distributed to all Connecticut high schools and libraries. The NHTSA "Team Spirit" leadership program was introduced in 1994.

Kansas implemented a pilot drug and alcohol prevention program in Wichita schools in 1982 that reduced school suspensions for alcohol by more than 30 percent by 1985. Training conferences spread the program to other areas. Kansas established a Governor's Center for Teen Leadership in 1989 to promote drug- and alcohol-free lifestyles through education and training. SADD chapters grew from 3 in 1985 to over 200 by the early 1990s. Radio and television PSAs directed at youth drinking and driving were developed and aired.

Nevada conducted a Governor's Student Safety Program in 1980 for high school students and advisors. SADD in Nevada began in 1984. Friday Night Live programs provide youth with alcohol-free activities on weekends.

New Jersey youth activities began in 1979 as S.O.B.E.R. (Stay Off the Bottle, Enjoy the Road), a community-level public information campaign involving MADD, RID, SADD, PTAs, the Federation of Women's Clubs, and other civic organizations. The Teen Institute of the Garden State (TIGS) was begun in 1987 to train students in alcohol and drug prevention strategies and leadership skills. The Governor's Youth Advisory Task Force on Alcohol and Drug Prevention was established in 1993 to oversee youth activities.

New Mexico's youth drinking and driving activities began in 1984, when the first SADD chapters were started. An Albuquerque school-based drinking and driving education program was begun in 1985 and spread to six other school districts by 1987. The Traffic Safety Education and Enforcement Fund, legislated in 1990, funded youth public information campaigns. Teen courts and Friday Night Live programs were active in several communities.

North Carolina formed the Governor's Youth Advocacy and Involvement Office in 1983 to organize youth groups and educate them about drinking and driving. SADD grew rapidly, to 325 chapters by 1990. Youth Safety Councils were formed in all state high schools to address drinking and driving and other issues. Many communities conducted Project Graduation activities annually. The college community of Chapel Hill instituted a Drive-A-Teen program to provide rides as an alternative to driving after drinking.

Ohio conducted over 1,300 high school programs on drinking and driving from 1986 through 1991. There were 670 SADD chapters by 1994 with a full-time state SADD coordinator. Project Graduation was conducted in 500 high schools by 1991; BACCHUS was active in 37 colleges by 1990. A "None for Under 21" campaign in 1994 promoted Ohio's zero tolerance law. Cops in Shops programs help enforce MLDA 21.

Pennsylvania held Youth Traffic Safety Council conferences in 1980 featuring drinking and driving. Forty school districts participated in an alcohol and drug education project in 1982-83. Ten regional youth traffic safety conferences were held in 1987. SADD had 400 chapters by 1989. An active network of community traffic safety programs conducts local public education and other program activities directed at youth.

Washington conducts youth drinking and driving activities through SAFTYE (Stopping Automobile Fatalities Through Youth Efforts), which began in 1974 and expanded substantially when a youth program coordinator was hired in 1985. By 1986, SAFTYE was active in over 120 high schools statewide. It conducted public education campaigns and held annual student conferences.

NHTSA's Report to Congress on Youth Alcohol Traffic Safety (1995) provides a similar brief summary of each state's federally-funded youth traffic safety activities during fiscal year 1994. The report tabulates activities in several broad categories and notes which states conducted activities in each. The categories and the number of states active within each are:

MADD's Rating the States. The Rating the States assessments compiled by Mothers Against Drunk Driving (MADD) provide an interesting, although subjective, measure of the strength of state youth programs (Mothers Against Drunk Driving, 1991, 1993, 1996, and 1999). In these assessments, a panel of MADD members and national drinking and driving experts collected information on each state's drunk driving laws and activities. The panel assigned each state and the nation as a whole a "report card" grade from A to F in each of about 10 categories. The categories differed slightly from year to year, but always included Youth Legislation, Prevention, and Education (in 1991 the category included only prevention and education, with youth legislation included elsewhere).

For the nation as a whole, Youth received the highest grade of all categories in each of the last three assessments: B- in 1993, tied with three others; B+ in 1996 along with one other; B+ in 1999 by itself. In 1991, Youth received the second highest grade of B, trailing one category and tied with none.

The MADD assessments are not evaluations in any sense. They subjectively compare the relative strengths of various drunk driving program components - law enforcement, legislation, public information, youth programs, etc. - against MADD's goals in each. The youth component contains legislation (zero tolerance) and enforcement (of MLDA 21 laws) as well as prevention and education programs. Nevertheless, when youth programs consistently receive top grades, it's clear that MADD's panel considers youth drinking and driving activities to be extensive and effective.

Community programs. Community traffic safety programs, or CTSPs, also developed and spread in the 1980s. These were organized and operated locally, sometimes with state assistance. Most conducted some youth impaired driving activities, as noted in some of the previous state program summaries. Two CTSP groups have been described and evaluated in some detail and are discussed in the following section.

Section 410. Section 410 of the 1991 Intermodal Surface Transportation Efficiency Act (ISTEA) provided grants for impaired driving countermeasures to qualifying states. To qualify, a state must meet a specified number of criteria: for example, in 1992, states qualified by meeting four out of five criteria. Over the years, the number of available criteria and the number required to be met for qualification changed somewhat. But one criterion directed at youth remained the same. To meet it, a state must issue distinctive driver licenses to persons under 21 and must conduct programs directed at youth drinking and driving, including a program for alcohol retailers and servers, a strategy to enforce the MLDA 21, and an underage drinking prevention program that involved youth participation.

In 1992, 19 states qualified for grants by meeting four out of five criteria. In 1997, 38 states qualified by meeting five out of seven criteria. In each year, 1992 through 1997, every qualifying state met the youth criterion. No other single criterion was met this frequently. For example, in 1997 all 38 qualifying states met the youth criterion, while the next most frequently used criterion was met by 31 states (Leaf and Preusser, 1998). This observation provides additional evidence that each state took substantial actions directed at youth drinking and driving.

State expenditures on youth programs. States have spent a substantial amount of both federal and state funds on youth drinking and driving prevention activities over the past 20 years. Information on these expenditures for the fiscal years 1993-1995 are provided in NHTSA's Report to Congress (1995). Congress appropriated $8 million for each fiscal year 1994 and 1995, and $9.2 million for fiscal year 1996, to be used by the states for youth drinking and driving prevention. NHTSA reports that in fiscal year 1993, before this additional appropriation, states spent $10.0 million of federal funds on youth drinking and driving (using a combination of Section 402, 408, and 410 funds). With the additional appropriation, the total rose to $21.1 million in fiscal year 1994 and to $21.7 million in fiscal year 1995.

Section 410 spending. Leaf and Preusser (1998) track Section 410 spending on youth drinking and driving for the six fiscal years 1992-1997. During these years, a total of $93.3 million in Section 410 funds was awarded to states. Approximately 15 percent, or $14.7 million, was used for youth activities: $3.4 million for enforcing laws, $3.2 million for school and workplace programs, $1.3 million for youth leadership development, $1.1 million for family-based programs, $0.8 million for public education directed to youth, $0.6 million for community youth program activities, and $4.3 million for other activities.

Effects of Youth Programs

The volume and variety of youth drinking and driving program activity is barely suggested by the preceding information. Unfortunately, most of these activities have not been evaluated, so the evidence of their effects is meager. This section reviews what's available.

SADD. Two studies attempted to evaluate SADD's activities and effects. Klitzner et al (1994) examined SADD programs in two schools (one each in California and New Mexico). They concluded that neither school implemented the model SADD program well, student participation in SADD was low, and comparisons with similar schools without SADD chapters showed no evidence for SADD effects on any drinking and driving measure.

Leaf and Preusser (1995) examined six schools with strong SADD programs (in Arizona, Ohio, and Wisconsin), matched with similar schools with no similar program. They found that students in the SADD schools were exposed to more information about drinking, drugs, and driving while impaired and were more likely to hold attitudes opposed to drinking and driving. Self-reported drinking and driving behavior was slightly, though not consistently, lower in SADD schools.

Community programs. Two studies show that well-organized community traffic safety programs can reduce youth drinking and driving. The first is the Massachusetts Saving Lives program discussed and evaluated by Hingson et al. (1996). The program operated in six Massachusetts communities beginning in 1989 and conducted many activities addressing all aspects of traffic safety. An evaluation compared results in these communities with five similar communities and with the rest of the state. The evaluation found that the proportion of 16-19 year olds reporting driving after drinking in the previous month dropped from 19 percent in 1988 to 9 percent in 1993 in program cities, a 40 percent decline relative to the rest of the state.

Communities Mobilizing for Change on Alcohol (CMCA) was directed very specifically at reducing youth access to alcohol and youth drinking (Wagenaar et al., 2000). It was conducted in seven Minnesota and Wisconsin communities with eight others serving as controls. Local CMCA organizations implemented many changes in community policies, procedures, and practices regarding alcohol service, backed up with extensive media and community support. The evaluation found that merchants in CMCA communities increased age identification checking and reduced sales to minors. The proportion of 18-20 year olds who drank in the past 30 days decreased 7 percent compared to the control communities.

Other evaluation evidence. As part of a guide to reducing youth alcohol use, Stewart (1999) reviews the evidence supporting the effectiveness of 36 different strategies. The three strategies directed specifically at driving all involve legislation and enforcement, and all "can be very effective": zero tolerance laws, sobriety checkpoints, and vigorous overall impaired driving law enforcement. The remaining strategies are directed at alcohol use in one way or another. The only strategies that have been proven effective also involve laws and enforcement: enforcement against retailers selling alcohol to youth, better laws prohibiting alcohol possession by youth, increased alcohol taxes, media campaigns supporting enforcement, and school policies regarding alcohol use. All other strategies, including the youth program activities discussed above, either have not been evaluated or evaluations have not found consistent effects. For example:


Clearly, states and communities conducted extensive youth drinking and driving programs in the past two decades. Other organizations concerned with traffic safety - insurance companies, automobile manufacturers, MADD, AAA, and many others - did the same through public education and specific program activities.

But there is little evidence of the effects produced by these activities. The CMCA and Massachusetts Saving Lives results show that community programs can be successful. But these two examples were well-organized and well-funded and certainly may not be representative of many other community programs. The SADD evaluations show that effective SADD chapters certainly affect students' knowledge and attitudes and may affect behavior. Again, it is not clear how many SADD chapters operate at this level. When states evaluate the effects of their overall youth impaired driving programs they typically examine the bottom-line measures of youth crashes or fatalities without attempting to disaggregate the effects of specific program components.

The conclusion: with the exception of Massachusetts Saving Lives, there is no direct proof that any of the myriad youth traffic safety program activities not involving laws and enforcement had any direct effect on youth drinking and driving. But there also is no proof that they did not. The accumulation of information, education, skills, role models, and the like provided by these programs may have been the crucial force in the youth attitude, behavior, and crash changes documented in Chapter III. We simply do not know.

E. Measures to Control Adult Drinking and Driving

Comparison of Youth and Total Drinking and Driving Changes

Drinking and driving by youth does not occur in isolation: it is strongly related to drinking and driving by older persons. Measures to reduce drinking and driving overall should also affect youth. One way to investigate this assertion is to compare each state's success in reducing drinking and driving overall with its success in reducing drinking and driving by youth.

In a previous study, Ulmer, Hedlund, and Preusser (2000) calculated each state's change in total alcohol-related traffic fatalities from 1982 to 1996. To reduce the effects of random variations in individual years, they fit a simple linear regression to the annual data and then calculated the percentage change in the predicted values from 1982 to 1996, in the same manner as was used for Table 2 of Chapter III. Across the states, the change ranged from a reduction of 67 percent to an increase of 22 percent. Mississippi had an apparent increase of 88 percent, but the Mississippi FARS data are suspect for some of this time period.

To compare with the results for youth, this change in total alcohol-related traffic fatalities for each state except Mississippi was matched with the state's change in youth drinking driver fatal crash involvements. The results are plotted in Figure 30. In addition, a simple linear regression was fit to these data and also is plotted in the Figure.

Figure 30. 
State Changes in Youth and Total Alcohol Related Fatalities
Graph of Figure 30. State Changes in Youth and Total Alcohol Related Fatalities (1982-1996-98)

The Figure has 49 points, one for each state except Mississippi. The horizontal axis measures the state's percentage reduction in total alcohol-related traffic fatalities from 1982 to 1996, from Ulmer, Hedlund and Preusser (2000), Table 3. The vertical axis measures the state's percentage reduction in youth drinking driver fatal crash involvements from 1982 to 1998, from Table 2, Chapter III. The regression line is:

youth % reduction in drivers involved = - 41.9 + 0.739 (total % reduction in fatalities)

Thus the regression model predicts that a state that reduced total alcohol-related traffic fatalities by 30 percent from 1982 to 1996 (a typical amount) would reduce youth drinking drivers in fatal crashes by

- 41.9 + 0.739 (- 30) = 64.1 percent

again a typical amount. The regression's r-square is 0.60, indicating a reasonable fit to the data.

The figure itself shows that most states lie close to the regression line. The few outliers are above the line: they did not reduce youth fatal crash involvements as much as the regression line would predict from their overall alcohol-related fatal crash reduction. Both the Figure and the regression show that, in general, the more a state reduced overall alcohol-related traffic fatalities, the more it reduced youth drinking driver fatal crash involvements.

Discussion and Conclusions

The likely reason for this relationship is that measures to reduce overall drinking and driving affect both youth and older drivers. The basic impaired driving laws apply to everyone; the special laws applying to youth merely add to these laws or make them stronger (as zero tolerance laws strengthen existing per se laws by lowering the BAC limit). Thus, these laws should deter youth as well as older drivers. Impaired driving law enforcement may affect youth to the extent that enforcement is present at the times and in the places where youth drive after drinking. Enforcement publicity sends a message that police are looking for all drunk drivers, not just those age 21 and above. General public education programs may reach youth if the messages and methods are appropriate to young as well as older drivers. Thus, states that took effective measures to reduce overall drinking and driving likely also saw the effect of these measures on youth drinking and driving.

Another factor also may be at work. States that took effective measures to address overall drinking and driving also may have taken special efforts to reduce drinking and driving by youth. Note, though, that all states adopted MLDA 21 and zero tolerance laws, the two measures proven to reduce youth drinking and driving. (Since 1980, 36 states raised their drinking age to 21 and all states enacted zero tolerance laws.) On balance, it is likely that some of the credit for reducing youth drinking and driving can be attributed to measures to reduce drinking and driving by everyone.

F. External Factors

Traffic safety in general, and drinking and driving in particular, may be affected by many broad economic, demographic, and social factors. This section discusses factors that may influence youth drinking and driving.

Changes in the Number of Youth

As documented in the section on national trends in Chapter III, the number of youth age 16-20 has decreased in the past 20 years while the number of older persons has increased substantially. When these changes are taken into account by comparing rates per population, the differences in drinking driver fatal crash involvements across age groups narrow. However, as was shown in Table 1, the decline in involvement rate for youth was considerably greater than among older drivers.

Economic and Travel Changes

Ulmer, Hedlund, and Preusser (2000) show the effects of travel, population, and employment changes on overall alcohol-related traffic fatalities. They report that state-by-state annual vehicle miles of travel (VMT), the numbers of persons employed, and the numbers of persons unemployed are associated with the overall level of alcohol-related traffic fatalities. That is, generally, favorable economic conditions yielded more alcohol-related fatalities than when the economy was less favorable.

Comprehensive state-by-state data on similar variables are not available for youth alone. To gain some insight into possible relationships between youth drinking and driving, and travel/economic conditions, a general linear models was constructed using state (as a categorical variable), year, VMT, persons employed and persons unemployed. Per capita beer consumption for the total population was also included. The dependent measure was 16-20 year old drinking drivers involved in fatal crashes. Thus, the dependent measure was specifically for youth while the independent measures covered the full population. Logarithmic transformations were used for the dependent measure and for VMT, persons employed and persons unemployed. An analysis of variance of the independent variables is shown in Table 16.

Table 16. Analysis of Variance for Model Variables
(Note: **p<.001)
Source df MS F
State 49 0.92 11.06**
Year 1 12.43 149.27**
VMT 1 0.13 1.60
Persons Employed 1 1.78 21.41**
Persons Unemployed 1 1.11 13.37**
Per Capita Beer 1 5.88 70.56**

All of the model variables except VMT were statistically significant. Employment and per capita beer consumption had a positive relationship to the numbers of 16-20 year old drinking drivers involved in fatal crashes, while year and unemployment had negative relationships.

These results suggest that as overall employment and income rise, youth employment and income also rises; more youth have access to cars, have discretionary funds to spend on gasoline and vehicle upkeep - and on alcohol. So it's highly likely that overall employment and income, and in particular youth employment and income, affect youth drinking and driving.

Youth Attitudes and Behavior

Chapter III shows clearly that youth attitudes regarding drinking and driving have changed substantially since 1980. It's also clear that youth knowledge, attitudes, and behavior have changed in many other ways during the same time. "Generation X" has given way to the "Millennials," raised on media and the Internet, cell phones and computers. The generation gap yawns as wide or wider than it ever has. Persons born before 1965 - the Baby Boomers of the postwar years, the Silent Generation born before them - have difficulty understanding today's teenagers (Howe and Strauss, 1993 label those born from 1961 to 1981 as the 13th Generation and describe some of their characteristics).

What changes in beliefs and practices in other areas affected youth drinking and drinking-driving attitudes and practices? Are youth today more politically conservative than in the 1960s? Are they more self-centered, interested in making money rather than saving the world? Do these changes reduce their risk-taking in general and their drinking and driving in particular? No studies or data addressing these issues were found. It's likely that there are substantial changes that have affected drinking and driving; it's possible that the observed changes in drinking and driving are merely one relatively minor manifestation of major attitude and belief shifts. On the other hand, it's entirely possible that drinking and driving changes brought about by measures such as MLDA 21 and zero tolerance laws helped change youth attitudes and behavior in other areas. No evidence one way or the other was found.


Changes over the past 20 years in where youth live and what they do undoubtedly affect youth drinking and driving. About 90 percent of young people now graduate from high school and about 62 percent of these graduates go on to college. By comparison in the early 1970s, about 85 percent of high school students graduated and 49 percent of these graduates went to college. Thus in recent years, more students are completing high school and more of these graduates are entering college. The underlying domicile, social and recreational patterns that go along with such changes may contribute to shifts in youth drinking and driving after drinking. For instance, it is likely that drinking on and around college campuses often can occur without requiring the use of an automobile. While plausible, such relationships cannot be quantified, however.


Economic, demographic, and social factors all undoubtedly influenced youth drinking and driving. The only influence that can be fully quantified is the change in the youth population relative to the total driving population. The other unquantified or even unidentified factors may have played a substantial role in reducing youth drinking and driving.

G. The Canadian Experience

To provide some perspective on youth drinking and driving changes in the United States since 1982, it's useful to examine Canada's experience over the same time.

Trends in Canada and the United States

Fatal crash data. Data from 1982 to date from the seven provinces for which these data are available (Alberta, British Columbia, Manitoba, New Brunswick, Ontario, Prince Edward Island, and Saskatchewan) were provided by the Transport Injury Research Foundation (TIRF). The data differ slightly from the United States FARS data in three ways. First, they record driver fatalities rather than driver involvements in fatal crashes. Second, they aggregate drivers age 16-19 rather than drivers under 21 as has been used for the United States. Finally, the TIRF file does not estimate alcohol presence for a driver without a BAC test. The proportion of fatally injured drivers with a positive BAC is calculated only for the drivers with a BAC test. Most drivers are tested: 76 percent of the age 16-19 driver fatalities in 1982 and 93 percent in 1997.

These differences may affect comparisons between the Canadian and United States data somewhat. For example, if BAC tests are available less frequently for sober drivers than for drinking drivers, then the proportion of fatally-injured drivers with a positive BAC calculated only from the tested drivers may overstate the proportion for all fatally-injured drivers. But these differences should not affect the trends over time or trend comparisons between the two countries.

Figure 31 shows the trend in Canadian fatally-injured drinking drivers aged 16-19. It looks rather similar to the United States trend of Figure 1, with a rapid decrease through about 1993 and no substantial change since then. (The Canadian trend fluctuates more from year to year than the United States trend since the absolute number of Canadian traffic fatalities is much smaller.) Figure 32 shows how very similar the two trends are by plotting both using a base of 1982 = 100 percent.

Figure 31. 
Canadian Driver Fatalities, Age 16-19, with Positive BAC
Graph of Figure 31. Canadian Driver Fatalities, Age 16-19, with Positive BAC

Figure 32. 
US and Canadian Trends, Percentage Change from 1982 
US: drivers under age 21 in fatal crashes with positive BAC (FARS) 
Canada: driver fatalities age 16-19 with positive BAC (TIRF)
Graph of Figure 32. US and Canadian Trends, Percentage Change from 1982

Figure 33 shows the trend in the proportion of fatally-injured Canadian drivers with a positive BAC. The trend is similar to the corresponding United States trend of Figure 3. The absolute percentages are higher in Canada: 67 percent in 1982 (compared to 43 percent in the United States) and 39 percent in 1997 (compared to 21 percent). Some of this difference may be a result of the different methods used to estimate alcohol involvement in the two data files. Some may in fact reflect higher drinking and driving rates in Canada. But the trends in the two countries again appear very similar. Figure 34 compares the trends directly.

Figure 33. 
Percent of Canadian Driver Fatalities, Age 16-19, with Positive BAC
Graph of Figure 33. Percent of Canadian Driver Fatalities, Age 16-19, with Positive BAC

Figure 34.
US and Canadian Trends, Percentage Change from 1982
US: percentage of drivers under age 21 in fatal crashes with positive BAC (FARS)
Canada: percentage of driver fatalities age 16-19 with positive BAC (TIRF)
Graph of Figure 34. US and Canadian Trends, Percentage Change from 1982

Figures 32 and 34 show the same thing: as measured by fatal crash data, youth drinking and driving decreases in the United States and Canada from 1982 to 1997 were virtually identical. Other data strengthen this conclusion. In both countries, the number of young drinking drivers in fatal crashes decreased more rapidly than the number of older drinking drivers. A roadside survey in British Columbia produced results similar to Roeper and Voas (1999): a much smaller proportion of drivers age 16-19 than older drivers had a positive BAC, or a BAC exceeding the legal limit of 0.08 (Mayhew and Simpson, 1999).

Survey data. Smart, Adlaf, and Walsh (1994) report on biennial surveys of about 4,000 Ontario high school students similar to the Monitoring the Future surveys in the United States. Table 17 summarizes self-reported drinking changes from 1979 to 1991 from the two surveys.

Since the Ontario data come from a sample of students in grades 7-13, while the United States data come from high school seniors, it's no surprise that overall self-reported drinking levels are lower in Ontario. The reductions, though, are generally similar: a modest reduction in annual drinking, substantial reductions in binge drinking, and reductions close to 50 percent in both daily drinking (at low levels in both countries) and driving after drinking.

Notes for Table 17: 

Table 17. 
Change in Self-reported Drinking and Driving 
after Drinking, High School Students
Drinking 1979 1991 Change
Annual drinking: United States




Annual drinking: Ontario




Daily drinking: United States




Daily drinking: Ontario




5 or more drinks: United States 
(2 weeks) 




5 or more drinks: Ontario 
(4 weeks) 




Drive after drinking: United States 
(2 weeks) 

Data from 1984



Drive after drinking: Ontario 

Data from 1983



Laws and Programs in Canada and the United States

The major difference between Canadian and United States laws affecting youth drinking is the drinking age itself. In Canada, the drinking age is 18 in three provinces (Alberta, Manitoba, and Quebec) and 19 in the other seven. These legal ages have not changed in the past 20 years. The Canadian reduction in youth drinking and driving must have been caused entirely by other factors.

Eight of the ten Canadian provinces have zero tolerance laws, frequently as part of a graduated licensing system; Alberta and Manitoba do not. Aside from Prince Edward Island's law, which became effective in 1990, the laws were introduced from 1994 to 1999. Figures 31 and 33 show that virtually all the reduction in youth drinking and driving, as measured by traffic fatality data, had occurred by 1994 when these laws began to be introduced. So other factors must have caused the reduction.

In Canada, as in the United States, many educational and motivational programs in the past 20 years addressed youth drinking and driving after drinking. There was a strong student movement through organizations such as SADD, Teens Against Drunk Driving (TADD), and Ontario Students Against Impaired Driving (OSAID). Health Canada's primary program target during this period was youth. However, as in the United States, there is virtually no evaluation evidence on the effectiveness of these efforts. Mayhew and Simpson (1999) briefly review the programs and the limited evidence.

Discussion and Conclusions

Canadian reductions in youth drinking and driving, measured both by fatal crash data and by surveys, followed virtually the same pattern as in the United States. But the Canadian reduction was not due to laws directed at youth: the drinking age did not change during this time, and zero tolerance laws were implemented after the reduction had occurred. This means that the changes must have resulted from some combination of the difficult-to-assess educational and motivational programs and from other factors outside of traffic safety. This conclusion suggests that a substantial portion of the reduction in the United States also resulted from these same causes.