The Effects of the Minimum Legal Drinking Age

The Effects of the Minimum Legal Drinking Age on Traffic Fatalities:
An International Comparison between the United States and Canada
Abstract: This paper analyzes the impact of the U.S. 1984 National Minimum Drinking
Age Act on the rate of alcohol-related traffic deaths by comparing traffic fatality rates in
the U.S and Canada. The results show that both countries experienced substantial
decreases in traffic fatalities and alcohol-related crashes since 1982. The data is broken
down by the age of the drivers involved to note any specific changes in teen and young
adult fatality rates. The similar age specific data of the U.S. and Canada implies that the
change in MLDA law can only account for a small part of the decrease in fatality rates.
Christine Cusick
Economic Honors Thesis
May 2006
Acknowledgments: I would like to thank my advisor, Professor David Lee, for his guidance on
this project and Professor Roger Craine for his help in Economics H195A. I am also grateful for
my friends and family who helped me stay motivated and focused throughout this year.
Table of Contents
I.
Introduction
II.
Historical Background and Review of the Laws
2.1
U.S. Legislation on MLDA
2.2
Canada Legislation on MLDA
2.3
Other Relevant Laws and Economic Factors
III. Literature Review on the Effects of Raising the MLDA
3.1
Raising the MLDA decreases traffic fatalities: The Age effect
3.2
Raising the MLDA shifts traffic fatalities: The Experience effect
IV.
Approach/ Strategy
V.
Data Selection
VI.
5.1
Sources
5.2
Disadvantages
Empirical Results
6.1
Overall Trends
6.2
Alcohol-Related Crashes
6.3
Teen Crashes
VII. Conclusion
VIII. Bibliography
IX.
Terms and Definitions
8.1
Terms
8.2
Definitions
1
I. Introduction
Public awareness of the consequences of driving while under the influence of
alcohol has changed public policy and culture. The past three decades have consequently
had a dramatic decease in automobile crashes. Despite the progress towards safety in
vehicle design, the problem of motor-vehicle crashes is still pertinent; the World Health
Organization estimates crashes account for the sixth world-wide leading cause of death.
In the U.S., motor vehicle crashes claimed 42,000 lives in 2002; moreover, alcohol was a
factor in over 40% of these.1 Alcohol related crashes, a subset of all motor vehicle
accidents, continue to plague society. In 2003, the 17,013 fatalities caused by alcoholrelated vehicle crashes represent an average of one alcohol-related fatality every 31
minutes (NCSA 2004).
It is estimated that drinking drivers are at least eight times more likely than sober
drivers to cause fatal crashes (Dee and Evans 2001). Clearly, drivers that heavily drink
impose negative externalities on society with their increased likelihood of crashes,
causing harms to life and property. There are two main strategies that attempt to tax
driving while intoxicated in order to make people pay the full social cost of the
consequences of alcohol. The first strategy makes use of general deterrence policies
which are used to change people’s decisions about driving while intoxicated. Examples
of these policies include strict punishments, heavy fines, education programs and mass
media campaigns. The second strategy aims not at the specific activity of driving
intoxicated, but at restricting the availability of alcohol. Alcohol control policies target
specific high-risk groups in order to deter alcohol abuse and ultimately drunk driving.
1
Carpenter 2004. According to the NCSA, A motor vehicle crash in considered to be alcohol-related if at
least one driver or non-occupant (i.e. pedestrian or cyclist) involved in the crash is determined to have had
a blood alcohol content (BAC) of 0.01 g/dl or higher
2
One policy of this later group is a minimum legal drinking age (MLDA). A lot of
drinking and driving literature focus’s on the teen driver. As a teen has less experienced
both with driving and with drinking, they are a target of many legislation policies to curb
drunk driving. Moreover, traffic fatalities are the leading cause of death among people
15-20 years old in the United States; 34% of all teenage males and 38% of all teenage
females who die annually do so because of car crashes.2 Minimum legal drinking ages
attempt to keep alcohol away from minors until they are older and hopefully more likely
to be responsible. MLDA laws work by criminalizing the sale or public possession of
alcohol to anyone under a certain age. In the United States, the MLDA is 21 years old.
This is the highest MLDA in the industrialized world, and was nationally enacted by the
National Minimum Drinking Age Act of 1984.
This paper intends to study the effects of a change in law in the United States by
comparing traffic fatalities in the U.S. and Canada to infer what might have happened in
the U.S. had there not been a uniform change in the minimum legal drinking age in 1984.
The 1984 legislation in the U.S. was discrete so we can compare fatalities in the U.S. to
values in Canada before and after the law was enacted. The expected results from this
observation are that after the law is implemented, alcohol is less accessible to teens in the
U.S. and they will have lower traffic fatality rates compared to Canadian teens.
Many cross-state studies have been done to analyze the correlation between
different MLDA policies and the amount of alcohol related crashes that occur. I review
these studies in this paper and try to apply their theories to international data. Fewer
studies have been done, however, comparing crashes internationally. I think it is
2
Dee and Evans (2000), Mayhew and Simpson (1999). Take into consideration that 15-20 years old rarely
die of natural causes and tend to be a demographic that takes more risks.
3
important to look at the international comparisons to gain a new perspective and
distinguish trends on how laws change behaviors.
Determining the effectiveness of MLDA laws in deterring drunk driving has
multiple benefits to society and individuals. Testing the effects can help policy makers
decide where to focus their resources and efforts into reducing traffic fatalities. Although
the decision to drive drunk is multifaceted, we can assume using Becker’s economic
theory of crime that the expected utility of drunk driving can be decreased if the expected
penalty and probability of being caught are high (Kenkel 2001). Enforcing MLDA laws
would hopefully deter some young people from driving drunk, because of this additional
penalty they receive from being underage.
Alcohol norms and acceptable drinking habits may vary dramatically by country.
Cultural differences are difficult to account for in a statistical study, yet it ought to be
pointed out that they serve as a potential factor to skew the results. The following
observations assume that there is not a significant difference in cultural views towards
drunk driving in the U.S. and Canada.
II. Historical Background and Review of the Laws
2.1 U.S. Legislation on MLDA
Before the National Minimum Drinking Age Act of 1984, it was up to states to
determine their own drinking age legislation. Most states had adopted a MLDA of 21
after the end to Prohibition in the 1920s. The 1970s brought social pressure to reduce the
MLDA in accordance with the trend to grant more rights to those who were deemed old
enough to be drafted to Vietnam. The Twenty-sixth Amendment lowered the federal
4
voting age to 18, so it seemed natural that 18 year olds should by extension deserve the
right to participate in other adult activities like drinking. Furthermore, the general
sentiment at the time was that alcohol legislation was ineffective in keeping alcohol away
from teens in the first place. In 1971, 12 states had an MLDA of 21 and 12 states had an
MLDA of 18.3 By 1977, however, 30 states had adopted the lower MLDA of 18.
During the late 1970s and early 1980s researchers found a significant increase in
youth auto-fatality rates. They argued that increasing alcohol availability to 16-20 year
olds increased the chance of them getting into an alcohol-induced fatal crash. These
studies combined with political lobbying groups such as Mothers Against Drunk Driving
(MADD) and Students Against Drunk Driving (SADD) to put pressure on the national
government to raise the MLDA.
On July 17, 1984 Congress passed The National Minimum Drinking Age Act and
all states were thereafter required to legislate and enforce the age of 21 years as a
minimum age for purchasing or public possession of alcoholic beverages.4 Under the
Federal Aid Highway Act, states not enforcing the minimum age would be subjected to a
ten percent decrease in annual federal highway funding.5 The magnitude of this shift in
MLDA was from 18 to 21, or in some cases from 19 to 21, increasing the age by about
6%. States complied quickly, and by 1988 all of the states had complied with the new law
and had raised the MLDA to 21 (DiNardo 2001).
3
Dee and Evans (2001). States with 21 MLDA: Alaska, California, Indiana, Kentucky, Missouri, New
Mexico, Nevada, North Dakota, Oregon, Pennsylvania, Utah, and Washington. States with an MLDA of
18: Colorado, Kansas, Louisiana, Mississippi, New York, North Carolina, Ohio, South Carolina,
Tennessee, Virginia, West Virginia, and Wisconsin
4
Also known as the Uniform Drinking Age Act
5
DeJong and Hingson (1998)
5
Figure 1:
Percent of U.S. Population Age 16-20 Covered by MLDA of 21
100
90
80
70
Percent
60
50
40
30
20
10
0
1982
1983
1984
1985
Year
1986
1987
1988
Source: Hedlund, Ulmer, Preusser 2001
2.2 Canada Legislation on MLDA
Canada’s drinking age is determined by province. For over the past thirty years
the ages have remained constant with three provinces having an MLDA of 18 (Alberta,
Manitoba, and Quebec) and the other nine having a MLDA of 19.6
There has not been public support for raising the MLDA in Canada primarily
because it is seen as culturally undesirable. It is hard to justify to the public that 18-20
year olds ought to be treated like adults in all aspects except for drinking. Wagaenaar
phrases this argument best by stating,
“it may seem unfair to many observers to allow 18-20 year olds to marry, to have children, to own
cars, homes and firearms and to be financially and socially independent, and yet to be legally
prohibited from drinking a glass of wine in a restaurant, or even a glass of champagne at their own
wedding."
-Wagaenaar 1981
6
Hedlund, Ulmer, and Preusser 2001
6
Although the majority of Canadians are not interested in raising the MLDA, they
are still concerned with alcohol-related traffic fatalities. The Canadian Council of Motor
Transport Administrators (CCMTA)—a branch of Transport Canada—has put together a
comprehensive plan to reduce traffic fatalities by 2010. The annual report, “Road Safety
Visions 2010” outlines how the administration hopes to cut the percentage of crashes that
involve alcohol by 40%.7
2.3 Other Relevant Laws and Economic Factors
Although the main interest of this paper is the effect of the shift in drinking ages,
it is important to point out the other laws that were being implemented during this time
period. The 1970s and 80s had various legislation intended to increase road safety. In
1984 Congress passed a mandatory seat belt law which was found to reduce traffic
fatalities among 16 and 17 year olds by nearly 8% and 18-19 year olds by almost 10%.8
National maximum speed limits also helped reduce traffic fatalities (Dee and Evans
2001). The following Table summarizes the estimated impacts of different policies to
reduce drunk driving.9
7
Transport Canada, Road Safety Vision 2010. See their website for details:
http://www.tc.gc.ca/roadsafety/vision/menu.htm
8
Dee and Evans 2001. Assuming primary enforcement of seat belt laws.
9
All estimated impacts are for the United States
7
Figure 2: Impact of Strategies to Reduce Driving Under the Influence of Alcohol
Strategy
General Deterrence
Policies
Administrative License
Revocation
Sobriety Checkpoints
Lower per se limits
Zero Tolerance Laws
Alcohol Control Policies
Minimum Legal Drinking
Age
Increased Alcohol Excise
Tax
Impact
Reference:
9% decline in alcoholrelated fatal crashes
17% decline in alcoholrelated fatal crashes
16% decline in the
proportion of fatal crashes
involving drivers at 0.08%
BAC and higher
20% decline in the
proportion of singlevehicle, nighttime fatal
crashes among 15-20 year
old driver
Klein (1989)
10-15% decline in alcoholrelated traffic deaths among
driver under age 21
15% decline in traffic
deaths among drivers ages
18-21
Wagenaar (1981)
Lacey, Jones, Fell (1996)
Hingson, Heeren, Winter
(1996)
Hingson, Heeren, Winter
(1994)
Saffer and Grossman (1987)
Source: DeJong and Hingson 1998
Besides drinking ages, the other main attempt at controlling alcohol usage is
through alcohol excise taxes. Alcohol taxes affect everyone, so they are not as targeted
as a MLDA, but youth can evade drinking age laws whereas they have to pay taxes. A
study by Saffer and Grossman looked at the correlations between high taxes and traffic
fatality rates for certain age groups. They found that youth alcohol consumption is
negatively correlated with price and that teens substitute other leisure activities for
drinking if prices are high. Moreover, they argue that accident mortality is negatively
correlated with the price of alcohol. A similar study estimated that the tax elasticity of
8
the fatality rate is -0.10 overall and -0.20 for 18-20 year olds.10 The conclusion these
studies draw is that a small increase in alcohol taxes could create the same reductions in
fatality rates as a shift in the MLDA.
In the United States, Congress has been hesitant to increase the tax on alcohol.
The taxes have been stagnant since 1951 so inflation has decreased the value of beer and
liquor taxes by about 20% of their original cost.11 Alcohol taxes are regressive taxes,
affecting everyone regardless of income, and are politically difficult to pass despite their
proven efficiency at lowering traffic fatality rates.
Another important policy that is related to a MLDA law is the Zero Tolerance
(ZT) law. ZT laws are complimentary to MLDA laws because they criminalize young
drivers with any alcohol in their systems; the legal limit is somewhere between a BAC of
0.00-0.02 g/dl. The main economic argument for ZT laws are that they are the best
policies because they are the most specifically targeted at the type of drinking that is bad
for society—drinking while driving. Also, zero tolerance sets an easily defined limit.
Individual drinkers are often time uncertain about when they reach a BAC level of 0.08—
the adult legal limit for both the U.S and Canada. There is confusion over how many
drinks it takes and how much alcohol is in one “drink.” A zero tolerance law ends this
uncertainty because people can be certain that if they drink heavily they will be over this
new stricter limit.
In the United States, zero tolerance laws are present for all drivers under the age
of 21. This was enacted by Congress in their 1995 National Highway Systems
Designation Act, legislating that “any state not enacting a BAC limit of 0.02 g/dl or less
10
11
Cook and Moore (1994)
Saffer and Grossman (1987). There was one slight increase in alcohol taxes in 1985.
9
for young drivers would lose 5% of its federal highway funds for that year and 10% each
subsequent year.” 12 Once a driver turns 21, however, they are then subject to the adult
BAC limit of 0.08 g/dl. In Canada, seven of the twelve provinces have a ZT law in place,
with one province (Manitoba) in the process of making a new ZT law (NHTSA 2000).
An econometric study done on U.S data was conducted by Christopher Carpenter
who studied the effects of ZT laws on self-reported alcohol use and driving drunk using
data from the Behavioral Risk Factor Surveillance System (BRFSS) 1984 to 2001.
Carpenter concluded that “the (ZT) laws reduced heavy episodic drinking (5 or more
drinks at one sitting) among underage males by 13%.” Furthermore, the implementation
of ZT laws increased the likelihood of being a light drinker as well as decreased the
overall number of drinks consumed by young males. There weren’t any significant
effects on the drinking behaviors of young women that he could find. His results argue
that ZT laws can change drinking behavior of some young adults and thus can reduce the
quantity of drunken drivers.13
The combinations of all of these laws make it difficult to distinguish the sole
effects of MLDA on traffic fatalities. Moreover, other potential factors exist that effect
alcohol use and drunk driving that have yet to be accounted for. Any factor that affects
alcohol use may consequently affect alcohol-related traffic fatalities. For example,
according to one study, state unemployment rates are negatively correlated with alcohol
use (DiNardo and Lemieux 2001). Unemployment and other economic conditions may
interact with trends in rates of fatal car crashes.
12
13
Dejong and Hingson 1998. All the states complied with this law by July of 1998.
Carpenter 2004. This study was conducted under the University Of Michigan School Of Public Health.
10
III. Literature Review on the Effects of Raising the MLDA
3.1 The Age Effect. Raising MLDA and Reducing Alcohol-Related Auto Fatalities
The intent behind a MLDA is to decrease the supply of alcohol to teenagers in
order to prevent a range of negative consequences that arise with alcohol abuse.
According to a study by a collaboration of doctors at the Boston University of Public
Health, adults who start drinking at the age of 14 were three times more likely to self
report driving drunk and four times more likely to be in an alcohol related car crash than
their peers who started drinking at the age of 21 (Hingson 2002). Delaying access to
alcohol has clear personal health and societal benefits.
Unfortunately, a MLDA is not completely enforceable, as some teenagers’
acquire alcohol from older friends, strangers, or from using fake IDs. Although some
evasion occurs, there is still evidence that raising the MLDA decreases alcohol use and
heavy teen drinking. According to a study by Thomas Dee, raising the MLDA from 18 to
21 reduced heavy teen drinking by MTF respondents by 8.4%.14
As a MLDA makes alcohol less accessible, studies have found that teens
substitute marijuana for alcohol. One study estimated that changing the MLDA from 18
to 21 decreases alcohol prevalence by 4.5 percentage points but also increases marijuana
prevalence by 2.4 percentage points (DiNardo and Lemieux 2001).
If teens are consuming less alcohol when faced with a higher MLDA, it is logical
to assume they will get in less alcohol-induced traffic accidents. Various studies have
proven this effect. Data taken from the national high school survey, “Monitoring the
Future,” shows that raising the MLDA to 21 reduces traffic fatalities by at least 9% (Dee
1999). Likewise, a study on the effects of raising Michigan’s MLDA from 18 to 21
14
Dee 1999. MTF respondents are high school seniors who took the “Monitoring the Future” survey.
11
estimated that single-vehicle nighttime crashes for 18-20 year olds were 16% lower then
the expected level (Waganaar 1986). As a final point, the NHTSA estimates that the
movement to a higher MLDA has saved over 23,000 lives since 1984 (NCSA 2004). The
results from these studies were the driving force behind the movement to raise the
national MLDA in the 1980s.
3.2 The Experience Effect: Raising MLDA and shifting fatalities to older ages.
One critique of the previous studies correlation between higher MLDA and lower
teen traffic fatalities is that the studies do not account for the bigger picture of overall
traffic fatalities.
Teens have high rates of accidents because they are inexperienced driving and
inexperienced drinking. Understandably this creates two added risks to teen driving and
they consequently have a higher accident rate then adults. If becoming a new drinker
creates an extra risk for getting into car crashes, we would expect to see a shift in
fatalities when the MLDA is attained no matter what this age is.15 This implies that the
national increase of MLDA from 18 to 21 may have shifted some of the fatality risks
from teens (16-20 year olds) to young adults (21-25 year olds).
Some research has been done that supports this theory. A 2001 study found that,
“among 23 and 22 year olds, those who faced an MLDA of 19 at age 18 had,
respectively, 8.9 and 7.5 percent fewer passenger-vehicle fatalities than those who could
legally drink at 21” (Dee and Evans 2001).
Asch and Levy call this effect the “Experience Effect” because it recognizes the
learning-by-doing aspect of alcohol consumption. This theory argues that increasing the
15
Asch and Levy 1990
12
MLDA from 18 to 21 may reduce 18-20 year olds fatalities but will also increase
fatalities for people 21 and over. Arguably a net life-saving effect can occur if the
reductions are greater then the increases. If this experience effect is significant, then the
gains to increasing the MLDA may be overstated in studies that correlate drinking ages to
traffic fatality rates. For example, a 1986 study by the National Institute of Alcohol
Abuse and Alcoholism calculated a 19.5% reduction in alcohol related crashes from an
increased drinking age by adding up a 6% decrease on 18-20 year old crashes and a
further 13% estimated decrease.16 This estimated decrease was found by comparing the
difference between 18-20 year olds crash involvement in states where the MLDA was 21
to the crash involvement of people in that same state who were 21 and over. The report
states that:
“Drinking drivers 21 and over experienced a significant 13.4% increase in their rate of
involvement in injury producing crashes in the early 1980s. Thus, the raised drinking age
apparently prevented 18-20 year olds from experiencing the same 13.4% increase in the rate of
HBD (had-been-drinking) crash involvement that occurred among those 21 and over.”
-Wagaenaar 1986
Clearly, adding up the 6% decrease from 18-20 year olds accident rate and the 13%
increase in 21 and over accident rate does not capture the true effect of the MLDA law if
the law caused some of those increases in the 21 and over accident rates.
Comparing age-specific fatality rates between the U.S. and Canada can help
determine how “natural” the increase in the fatality rates for young adults was after the
implementation of the higher MLDA. If Canada’s 21 and over age cohort did not see
similar increases in traffic fatality rates in the 1980s, then it is possible that the MLDA
life saving effect was less then previously estimated in the United States.
16
Wagenaar 1986
13
The policy implications from the Experience Effect are seen by focusing efforts not
on MLDA laws but on graduated license programs. Graduated licenses capture the
benefits of learning-by-doing by restricting access to risky driving situations. Young
drivers are granted the privilege to drive in a series of stages. Often under these
programs licenses are granted with restrictions for the first 6 months to 2 years to help
new drivers become experienced driving in safe environments before they drive in more
risky environments. For example, California’s current law allows newly licensed 16 year
olds to drive only during the hours of 6am to midnight for the first six months of their
license.17 Many studies have shown that alcohol involvement in fatal accidents is higher
during the nighttime, so restricting access to both the alcohol (MLDA) and the driving at
night (graduated licensing) would be the best approach to decreasing nighttime crashes.18
IV. Approach/ Strategy
To interpret the effect of raising the MLDA in the United States I am comparing
the relationship between an input (minimum age legislation) and the output (auto
fatalities). This approach does not specify or prove the causal mechanism which links
them; however many studies have hypothesized different theories for how MLDA’s
affect auto fatalities. As explained above, previous studies have found mixed results on
the magnitude and effects of a shift in the MLDA. I interpret how this new data fits into
previous theories on the relationship between MLDA and auto fatalities. Specifically, I
want to see if comparing the U.S. data to Canada supports the standard assumption that
17
24 States including California have a graduated license system in place. Likewise, 10 of 12 provinces in
Canada have them as well.
18
Dee 1998. A NHTSA repot in 1998 found that during 1982-1992 alcohol was involved in traffic
fatalities 68-75% of 18-20 year old driver crashes between midnight and 5am whereas its only 10-20%
from 7am to 3pm.
14
MLDA lowered fatality rates (the age effect) or if it shifted some of these deaths to young
adulthood (the experience effect).
The advantage of comparing the United States and Canada lies in their
similarities. Figure 3 compares different variables in the U.S. and Canada.19
Figure 3: Comparing the United States and Canada 2004
Road Safety Factors
BAC limit
Passenger Vehicles per 1000 People
(1998 levels)
Paved Roads, as Percent of Total Roads
(1995)
Economic Factors
GDP per capita (constant 2000 US $)
Public Spending on Education (% of
GDP) (2000)
Hospital Beds per 1,000 People (1999)
Unemployment Rate (% of total Labor
Force)
Cultural/ Demographic Factors
Per Capita Consumption of Absolute
Alcohol in Liters20
Crude Birth Rate per 1,000 People
Crude Death Rate per 1,000 People
Population aged 15-64 (as percent of
total)
United States
Canada
0.08 g/dl
485.7
0.08 g/dl
458
60.7
35.3
$36,789.81
5.748
$24,711.85
5.248
3.6
6
3.9
7
7.6
7.9
13.8
8.5
66.8
10.5
7.2
69.2
The fact that the U.S. and Canada are similar on some variables does not indicate
that they are similar in most comparisons. It simply points out that Canada is a good
candidate to compare to the U.S. when using natural experiments where we cannot alter
the treatment and control groups to be random.
19
Data taken from the World Development Indicators, part of the United Nations Common Database and
the World Bank Group, from 2004 unless otherwise noted.
20
NHTSA 2000, estimating 1995 levels
15
We treat the absence of a higher MLDA in Canada as a natural experiment on the
effects of MLDA on auto fatality rates and view the results in comparison to the U.S.
This allows another dimension of policy analysis as we can see which theory holds up
when you expand beyond U.S. data. Canada is a good comparison because they are in
proximity to the U.S. and have similar roads, economies and cultures. Furthermore,
Canada’s MLDA has been steady throughout the changes in U.S. law.
V. Data Section
5.1 Sources of Data
Data for the United States is taken from the National Highway Traffic Safety
Administration (NHSTA). NHSTA maintains a census of all motor vehicle traffic
accidents that involve a fatality. The Fatal Accident Reporting System (FARS) includes
detailed information on the characteristics of the involved vehicles, drivers, occupants
and non-occupants.
Data for Canada is taken primarily from the Traffic Injury Research Foundation
(TIRF), which maintains a database of results from tests for alcohol in the fatally injured
drivers in seven provinces (British Columbia, Alberta, Saskatchewan, Manitoba, Ontario,
New Brunswick, and Prince Edward Island) from 1973-1987. After 1987 the Fatality
Database was expanded to include all ten provinces and the two territories.
Supplementary Canadian data is taken from Transport Canada and their affiliated
publications.
The TIRF data differs from the FARS data in a few important ways. First, the
Canadian data records driver fatalities instead of driver involvement in fatal crashes. This
16
limits comparisons slightly, as we can only compare driver fatalities over both countries
instead of overall driver involvement per age group. The next major difference between
the countries is in how they categorize and organize the crash data. In the United States
we categorize age groups into those who are less then 16, 16-20, 21-24, and so on.
Canada, however, groups the ages into those less then 15, 15-19, 20-24, and so on. The
age groups in each country are similar after age 25. The young adult data in Canada is
aggregated into groups that are a year off from the U.S. data. At first I ignore the
difference in data categories in hopes they do not affect overall comparisons. I then try to
adjust for the year difference by re-arranging the age groups to match up.
The last major difference in the Canadian data is that the TIRF does not estimate
alcohol presence for a driver without a BAC test. Fortunately, most drivers are tested:
76% of the age 15-19 driver fatalities in 1982 and 93% in 1997.21 These differences
should not affect the data when looking at overall trends of traffic fatalities, yet it should
be noted that the comparisons are not exact.
There has been some attempt at international comparison of this sort of data. The
OECD’s International Road Traffic Accident Database (IRTAD) and studies complied by
the UN offer a good resource for contrasting data across different countries, and some of
their papers are quoted in this study.
5.2 Disadvantages with the Approach
There are a few disadvantages to comparing cross-country statistics. For example,
the treatment and control groups (U.S and Canada) might differ in important ways, either
before or during the study period which could bias the results. Some of these differences
21
Hedlund, Ulmer, Preusser 2001
17
can be accounted for, such as population levels, whereas others are less concrete and hard
to measure.
Another potential omitted variable is the effect of some similar exogenous law
that aims to decrease total automobile fatalities. These laws are examined in the historical
background section of this paper yet it is important to point out that it is difficult to
distinguish and calculate the effects of each law separately.
The last major disadvantage with this comparison is that I am focusing on broad
indicators, such as total fatality rate per 100,000 people, and not specific behavioral
changes, like using a designated driver. The reason for choosing fatality rates as a
dependent variable is that it is well documented in both countries. Specific behavioral
changes are harder to measure even though they most likely have significant impacts on
decisions to drive drunk.
In previous studies on the affects of MLDA, additional control variables are often
accounted for. Examples of these variables are the natural log of the population for each
age group; the unemployment rate; real personal income per capita; and year fixed
effects. I do not include these additional controls, although I do comment on their
potential impact on the results.
VI. Empirical
6.1 Overall Trends
In the United States, and most other developed countries, there has been a notable
decrease in traffic fatality rates since the 1970s. According to the National Highway
Traffic Safety Administration, (NHTSA), the occupant fatality rate per 100,000 people
18
declined by 24.6% from 1975 to 2004.22 Internationally, a World Bank Study found that
“the mean road death rate of the OECD countries has fallen by half since 1970, from 34
deaths/ 100,000 persons in 1970 to approximately 12 deaths/100,000 persons in 2000”
(Kopits and Cropper 2005). There have been declines in the fatality rates per 100 million
vehicle miles traveled since the 1970s, when much of the legislation aimed at preventing
drunk driving became enacted.
These trends are similar in Canada. Traffic fatalities in 2001 are half of what they
were in 1975.23 Alcohol-related traffic fatality rates have decreased as well—a one-third
decrease since 1980. Despite gains to traffic safety, driving while intoxicated continues
to be a major problem for Canada; compared to Canada’s national homicide rate, drunk
driving kills two times the number of Canadians murdered each year.24 The following
graphs show the traffic fatality rates in the U.S. and Canada since 1970.
Figure 4: Traffic fatalities since 1970 per 100,000 populations
Source: IRTAD 30/200
22
NHTSA, Traffic Safety Facts 2004
Mayhew, Brown and Simpson 2002
24
SADD Canada website, referencing Statistics Canada
23
19
From the graphs it appears that overall trends of decreased traffic fatalities are
rather similar for both countries. During the 1990s, however, the U.S. fatality rates
stabilize while Canada’s continue to fall. The following graph shows road fatalities per
million people in each country in the past decade.
Figure 5:
Road Fatalities per Million Population in the United States and Canada
200
185
Fatalities per Million Population
170
155
140
United States
Canada
125
110
95
80
65
50
1990
1991
1992
1993
1994
1995
1996
Year
1997
1998
1999
2000
2001
2002
Source: OECD Factbook 2005
It is unclear why the 1990s yield so little change in the U.S. fatality rates.
Canada’s levels also stabilize somewhat, compared to the sharp decreases of the 1980s.
One can speculate that the reason for the stabilization of fatality rates may be due to the
lack of public attention traffic safety issues have received in the 1990s. Drunk driving
initiatives were a higher political priority in the 1980s and the attention it received may
be the driving force behind some of the decreases from that time period.
20
Other factors to consider when viewing the overall trend include: improved
technology in car safety features such as anti-lock breaks and air bags, social
improvements in health care and health services, and social norms changing considering
wearing a seat belt and driving under the influence.25 Also, with continuing social
pressure to crack down on intoxicated drivers, police arrests and DUI tickets have
increased since the 1970s, which would also decrease the overall amount of crashes.
However, this effect has not curbed the problem of drunk driving when in 2002, one out
of every 130 licensed drivers—or 1.5 million drivers—in the U.S were arrested for
driving while intoxicated (NCSA 2003).
When comparing crash statistics internationally one has to account for the
differences in population. Canada is a smaller country then the U.S. and will naturally
have fewer crashes. That’s why we take the fatality per 100,000 or million population
instead of the absolute number of crashes when comparing the U.S. and Canada. The
Canadian trends tend to fluctuate more then the U.S. trends since the absolute number of
crashes is less in Canada. Overall, however, the two countries both show decreasing crash
rates throughout the 1980s.
6.2 Alcohol-Related Crashes
The overall trends in alcohol-related crashes also appear to be rather similar for
the U.S. and Canada. Both countries had around 60% alcohol involvement of their fatal
crashes in 1982. By 2003, however, alcohol is a factor in only 40% of U.S. fatal crashes
and around 37% of Canadian crashes.
25
Tessmer 1999. Studies show seat belts increase the probability of crash survival by 50-60%. Other
factors that improve car safety include the use of daytime running lamps, improved roads, etc.
21
Figure 6:
Alcohol Involvement in Fatal Car Crashes 1982-2003
70
Percent Alcohol-Related Traffic Fatalities
65
60
55
50
United States
Canada
45
40
35
30
25
02
01
00
99
03
20
20
20
20
97
96
95
98
19
19
19
19
94
Year
19
92
91
90
93
19
19
19
19
88
87
86
85
84
83
89
19
19
19
19
19
19
19
19
19
82
20
Source: NHSTA and Transport Canada 2004
The U.S. and Canadian curves do not diverge from each other in the expected
direction once the 1984 legislation is passed. Even by 1988, when all U.S. states had
adopted the higher MLDA of 21, the U.S. did not gain extra reductions in alcohol-related
traffic fatalities compared to the subsequent decreases that were going on in Canada at
the time. In fact, around 1988 the graph indicates that Canada actually had greater
reductions in fatalities. This indicates that overall, the MLDA change was not the most
significant factor at achieving substantial reductions in alcohol-involved traffic fatalities.
The next graph adds in linear regression estimations of each countries traffic
fatalities over time. The linear regressions point out that in the period from 1982-2003,
Canada has been slightly more effective then the U.S. at reducing the percentage of
alcohol-related traffic fatalities.
22
Figure 7:
Alcohol Involvement in Fatal Car Crashes 1982-2003
70
Percent Alcohol-Related Traffic Fatalities
65
60
55
United States
Canada
Linear (Canada)
Linear (United States)
50
45
40
35
30
25
2
1
0
9
8
7
6
5
3
20
0
20
0
20
0
20
0
19
9
19
9
19
9
19
9
4
Year
19
9
2
1
0
3
19
9
19
9
19
9
19
9
19
9
8
7
6
5
4
3
9
19
8
19
8
19
8
19
8
19
8
19
8
19
8
19
8
2
20
Source: NHSTA and Transport Canada 2004
Even if Canada has had a higher percentage drop in fatalities, the overall trends
for both countries are similar. There does not appear to be a clear divergence once the
1984 change in MLDA occurred.
6.3 Teen Crashes
As the main focus of MLDA targets teenagers, we should look at teenage driving
to single out the effect of the law. Teen crashes in the U.S. have fallen substantially in
the past few decades. During 1979-2001 teen traffic fatality rates fell by 37%, with the
bulk of this drop (25%) occurring from 1986-1992 (Dee and Evans 2001). In 1982
drivers under the age of 21 had 4,393 fatal crashes while in 1999, they were only
involved in 1,687 (Hedlund, McCartt 2002). These reductions are particularly impressive
considering the increased amount of vehicle miles traveled (VMT) by teens since the
23
1980s. VMT by teens aged 16-19 doubled from 1983-1995. Taking this into account, teen
fatality rates have fallen by 50% since 1983 if denominated by VMT (Dee and Evans
2001).
Alcohol-related car crashes have also decreased for teens. The NHTSA estimates
that alcohol-related teen fatality rates decreased by 61% from 1982-1998 (Hedlund,
Ulmer, Preusser 2001). The following graph shows the trends of alcohol involvement in
car crashes for people aged 16-20, 21-24 and 25-34 years old over the time 1982-2004.
Figure 8
Percent of American Driver Fatalities by Age Group with Positive BAC
60
Drivers 21-24
Drivers 16-20
Drivers 25-34
50
Percent
40
30
20
10
04
03
01
02
20
20
20
00
99
20
20
97
96
95
98
19
19
19
19
94
Year
19
92
91
90
89
88
93
19
19
19
19
19
19
19
86
85
84
83
87
19
19
19
19
19
19
82
0
Source: Traffic Safety Facts 2004, NHSTA
Figure 8 shows that teen drivers have had the most reductions in alcohol-related
fatal car crashes. During the late 1990s, however, all three age groups see little change.
The period of sharpest decline occurred during 1991-1994. Some of this decline is due to
there being fewer teens in general. The population of teens aged 16-20 in the U.S.
24
dropped 6% during 1982-1998 while the population of adults aged 25-54 rose 31%
(Hedlund and McCartt 2002).
The trends are similar in Canada. Teens have decreased both their overall number
of traffic fatalities and their percentage of alcohol-related crashes.
Figure 9: Percent of Canadian Driver Fatalities, Age 16-19, with Positive BAC
Source: Hedlund, Ulmer, Preusser 2001
The trends are similar for Canada even though their absolute percentages are higher
then the U.S. For example, the 1982 Canadian level is 67% whereas the 1982 U.S. level
is 43%. One way to compare the trends in each country is to set 1982 as the base year and
then graph the percent change from that level in subsequent years. In this format, the
decreases in alcohol-related fatal crashes appear similar in the U.S and Canada.
25
Figure 10: 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)
Source: Hedlund, Ulmer, Preusser 2001
Another way to break down the data is to look at specific years and see the age
distribution on the drivers involved in fatal crashes. The most recent data is for 2004
which is used to make the graph below.
Figure 11:
Drivers of Fatal Car Crashes by Age Group 2004
25
Percent of All Drivers in Fatal Crashes
20
15
United States
Canada
10
5
0
0-15
16-20
21-24
25-34
35-44
Driving Age (Years)
45-54
55-64
65 +
Unknown
Sources: NCSA Traffic Safety Facts 2004 and Canadian
Motor Vehicle Collision Statistics 2004
26
The results are that the U.S. has more 16-20 year old drivers involved in fatal crashes
and less 21-25 year old drivers then Canada. I had expected it to be the opposite because
both the age and experience effects. One important note about this data is that the
comparisons are slightly off. 16-20 age cohorts in the U.S. are compared with 15-19 year
olds in Canada. Also 21-24 in the U.S. is compared with 20-24 in Canada. To check if the
year difference is significant I re-arranged the data to line up the age groups. I estimated
the rates per individual age year as a fraction of the original age group it was in. Figure
11.2 shows the results of rearranging the data to deal with the year difference in age
groupings.
Figure 11.2
Drivers of Fatal Car Crashes by Age Group (Adjusted Data)
25
Percent of All Drivers in Fatal Crashes
20
15
United States
Canada
10
5
0
0-15
16-20
21-24
25-34
35-44
Driving Age (Years)
45-54
55-64
65 +
Unknown
Sources: NCSA Traffic Safety Facts 2004 and Canadian
Motor Vehicle Collision Statistics 2004
In Figure 11, U.S. 16-20 year olds had a much higher involvement in fatal crashes
then their 15-19 year old Canadian peers and U.S. 21-24 year olds had less crashes then
27
the 20-24 year old Canadians. When adjusting the data, the magnitudes of the differences
between the U.S. and Canadian young adult levels are reduced, as seen in Figure 11.2
These results are surprising because as U.S. teens face a higher MLDA, we would
expect them to have less crashes then Canadian teens. The 16-20 year old group includes
legal drinkers in Canada yet they still have less crashes then American teens that are all
legally underage.
The next graph shows fatal crashes by age group and by alcohol involvement. Again,
there is the complication with the ages not being directly comparable. Drivers less then
16 years old were omitted because they constituted a negligible amount of alcohol-related
fatalities.26
Figure 12
Percent of All Fatally Injured Drinking Drivers by Age Group 2004
30
25
20
Percent
United States
Canada
15
10
5
0
16-20
21-24
25-34
35-44
Driver Age (Years)
45-54
>55
Source: NCSA Traffic Safety Facts 2004 and The AlcoholCrash Problem in Canada 2003
26
NHSTA estimated drivers less then 16 were responsible for around 0.3% of all alcohol-related fatal
crashes.
28
The results for alcohol-related crashes are similar to the results for all fatal crashes. In
2004, U.S. teenagers ages 16-20 had a higher percentage of alcohol-related fatalities then
Canadian teenagers. The opposite is true for American 21-24 year olds. However, when
adjusting the data using the same method as before, the results change.
Figure 12.2
Percent of all Fatally Injured Drinking Drivers by Age Group 2004 (Adjusted Data)
30
25
Percent
20
United States
Canada
15
10
5
0
16-20
21-24
25-34
35-44
Driver Age (Years)
45-54
>55
Source: NCSA Traffic Safety Facts 2004 and The
Alcohol-Crash Problem in Canada 2003
Using the adjusted data, American 21-24 year olds are involved in a higher
percent of all alcohol-related fatal crashes then their 21-24 year old Canadian peers. This
result supports previous theories on the Experience Effect as we see an age group that is
new to alcohol participating in more alcohol-related crashes. Another difference in the
adjusted data is that 16-20 year olds constitute a higher percentage of alcohol-related fatal
crashes in Canada then in the U.S. This result supports the Age Effect theory of the
29
MLDA, as 16-20 year olds in the U.S. have less access to alcohol then Canadian 16-20
year olds, so they consequently get in less alcohol-related fatal car crashes.
These findings may not hold a lot of statistical ground because it is unlikely that my
estimates are accurate. I’m assuming that 20, 21, 22, 23, 24, and 25 year olds all
constituted an equal amount of traffic fatalities, and then switching the 20 year olds
portion of fatalities (1/5 of the age cohort 20-25) to the younger age cohort (16-20).
Similarly, I assume all 16, 17, 18, and 19 year olds each account for ¼ the crashes in their
age group of 16-19. If we had the age specific data for each year, we would assume the
rates to differ between each year of age.
The following table shows data from Canada that 19 year olds have a much higher
percentage of involvement with fatal crashes then 16 year olds (50% compared to 33%)
in 1998.
Figure 13: Percent of Drinking and Legal Impairment by Age
in Canada
Fatally Injured Drivers
Age
Had-BeenLegally Impaired
Drinking
16
30.8%
30.8%
17
36.2%
25.8%
18
33.3%
22.2%
19
50%
40.3%
Source: Mayhew and Simpson 1999
Although the 2004 data can present one story, the overall trends of Canadian and
U.S. teen fatality rates are extremely similar. The 2004 data could be an example of a
fluctuation and not represent the overall trend. To test this, I look back to the trends of
teen crashes and young adult crashes in Canada and the U.S.
30
Figure 14:
Percent of Drivers in Fatal Car Crashes that are Legally Intoxicated (BAC>0.08) by Age Group
1987-2001
60
50
40
U.S. 16-20
Percent
CAN <19
U.S 21-24
30
CAN 20-25
20
10
0
1987
1988
1989
1990
1991
1992
1993
1994
Year
1995
1996
1997
1998
1999
2000
2001
Source: Traffic Safety Facts 2004 and Road Safety in Canada
2001
According to Figure 14, U.S. teens have been less likely then Canadian teens to
be legally intoxicated when involved in a fatal car crash. Likewise, U.S. young adults
have a lower rate of intoxicated drivers in fatal crashes then Canadian young adults.
Notice, however, that the age groupings are different for each country. American young
adults are grouped as ages 21-24 whereas Canadian young adults contain everyone in the
ages 20-25. I next adjust the data so that we can compare U.S. and Canadian young
adults within the exact same age grouping.
31
Figure 15
Drivers in Fatal Car Crashes who are Legally Intoxicated 1987-2001 (adjusted data)
45
40
U.S 21-24
35
CAN 21-24
Percent
30
25
20
15
10
1987
1988
1989
1990
1991
1992
1993
1994
Year
1995
1996
1997
1998
1999
2000
2001
Source: Traffic Safety Facts 2004, Road Safety in
Canada 2001
Figure 15 shows that between 21-24 year olds, Americans actually have a higher
percentage of drivers in fatal car crashes that are legally intoxicated. This result is
consistent with the adjusted data results from earlier. These trends support the theories
on the age effect and the experience effect.
VII. Conclusion
There have been substantial decreases in youth drinking and driving since the
early 1980s. There is not a specific reason why alcohol-related fatality rates have
decreased, but we can infer that a combination of policies helped create this drop.
Changing the MLDA might have decreased teenage drinking and driving and may
have also helped promote a cultural taboo against driving drunk. The change in the
32
MLDA did not, however, cause the majority of the decreased auto-fatality rates due to the
fact that similar decreases occurred in Canada without a change in drinking age. The age
specific data showed that both countries experienced substantial decreased fatality rates
over the past few decades—namely in the 1980s. Other factors are working to decrease
the fatality rates that may or may not be measurable.
Examples of other factors that occur in the U.S. and Canada are active citizen
groups such as MADD and SADD, various drunk driving legislation such as ZT laws or
punishments for DUIs, and vigorous public education campaigns. Future studies could
incorporate these factors into models estimating the impact of the U.S.’s higher MLDA.
The result from comparing age specific data in the U.S. to Canada does support
previous research on the effects of the MLDA. When adjusting the data to create
comparable age groups, I found that Canadian teens (16-20) are involved in more
alcohol-related fatal crashes then American teens. This is presumably a result of
Canadians teens’ legal access to alcohol. Moreover, Canadian young adults (21-25)
appear to constitute a slightly less overall percent of alcohol-related fatal crashes then
American young adults. This result arguably is a result of American young adult’s lack of
experience with alcohol because of the higher MLDA, creating an added risk of alcoholrelated car crashes when first exposed to alcohol.
The statistical methods used to obtain these results, however, are not precise. The
issue of measurement error weakens my results. Besides the potential measurement
error, the problem with my analysis is that it covers too broad a topic. A change in
MLDA may affect traffic fatality rates to some degree, yet this effect is hard to
distinguish in the aggregate data.
33
MLDAs ought not to be discounted as ineffective. Other studies have shown that
a MLDA has positive externalities on society such as increasing the probability of
graduating college by 4.2%.27 A MLDA also sends out a strong social message about
cultural expectations not to drive drunk.
There is plenty of room for future research to expand upon the relationship
between MLDA and traffic fatalities. After taking the overall trends into account, studies
can look at other variables as well and try and see which has had the biggest impact on
fatality rates. Examples of such variables have been mentioned earlier in this paper and
include other U.S. legislation; general economic variables to capture the state of the
economy in both the U.S. and Canada in the 80s; drunk driving education programs; and
youth behavior changes.
Another aspect to include in future research is to include the effects of alcohol
taxes. Taxes are another form of an alcohol control policy and have been found to be
effective at reducing the demand for alcohol and consequently reducing drunk driving.
Looking at Canadian and U.S. alcohol tax rates may explain why Canada was able to
decrease their fatality rates without a change in MLDA.
This study could be enhanced with access to more data. It would be interesting to
look at trends of crash participation for every age, instead of age groupings. Ideally, the
data would be gathered by a common source, so that the technique is the same for both
Canada and the U.S., eliminating potential differences in data collection and recording
methods.
27
Cook and Moore (1994). Calculated effects of changing the MLDA from 18-21 using data between
states.
34
VIII. Bibliography
Asch P and David Levy (1990) “Young Driver Fatalities: The Roles of Drinking Age and
Drinking Experience,” Southern Economic Journal, Vol. 57, No. 2. pp 512-520
Bjerre, Bo (2003) “An Evaluation of the Swedish Ignition Interlock Program” Taylor and
Francis, Volume 4, #2, pages 98-104
Carpenter, Christopher (2004) “How do Zero Tolerance Drunk Driving Laws Work?”
Journal of Health Economics Volume 23, Issue 1, Pages 61-83
Cook P and George Tauchen (1984) “The Effect of Minimum Drinking Age Legislation
on Youthful Auto Fatalities 1970-1977,” Journal of Legal Studies, Vol. 13, No. 1. pp
169-190.
Cook P and Michael Moore (1994) “This Tax’s for you: the Case for Higher Beer
Taxes,” National Tax Journal Vol. 47, No. 3, pp. 559-73
Dee, Thomas (1999) “State Alcohol Policies, Teen Drinking and Traffic Fatalities,”
Journal of Public Economics 72: 289-315
Dee, Thomas and William Evans (2001) “Behavioral Polices and Teen Traffic Safety”
The American Economic Review, Vol. 91, No. 2, Papers and Proceedings of the Hundred
Thirteenth Annual Meeting of the American Economic Association, pp 91-96
DeJong W and Ralph Hingson (1998) “Strategies to Reduce Driving Under the Influence
of Alcohol,” Annu. Rev. Public Health. 19:359-78
DeYoung, David J. (2002) An Evaluation of the Implementation of Ignition Interlock in
California, California Department of Motor Vehicles, Sacramento, CA
DiNardo J and Thomas Lemieux (2001) “Alcohol, Marijuana, and American Youth: the
Unintended Consequences of Government Regulation,” Journal of Health Economics 20:
991-1010
Hedland J and Anne T. McCartt (2002) Drunk Driving: Seeking Additional Solutions,
Preusser Research Group, Inc. Trumbull, CT
Hedland, Ulmer, and Preusser (2001), Determine Why There Are Fewer Young Alcohol
Impaired Drivers, U.S. DOT NHTSA, Final Report, DOT HS 809 348, Preusser Research
Group, Inc. Trumbull, CT
Hingson R, Heeren T, Winter M, (1994) “Lower Legal Blood Alcohol Limits for Young
Drivers,” Public Health Rep. 109, pages 739-44
35
Hingson R, Heeren T, Winter M, (1996) “Lowering the State Legal Blood Alcohol Limits
to 0.08 percent: The Effect on Fatal Motor Vehicle Crashes,” American Journal of Public
Health 86; 1297-99
Hingson, Ralph et al. (2002) “Age of Drinking Onset, Driving After Drinking, and
Involvement in Alcohol-Related Motor Vehicle Crashes” Department of Transportation,
HS 809, Springfield, VA
Holder H and Alexander Waganaar (1994) “Mandating Server Training and Reduced
Alcohol-Involved Traffic Crashes: A Time-series Analysis of the Oregon Experience,”
Accid. Anal. Prev. 26:89-97
Hutt, Katherine (2001) Setting Limits, Saving Lives: The Case for .08 BAC Laws, U.S.
Department of Transportation, National Highway Traffic Safety Administration, Page 12
Jacobs, G, and A. Aeron-Thomas (2000) “Estimating Global Road Fatalities” Transport
Research Laboratory, Department for International Development, London
Jones, Ralph and John Lacey (2000) “Final Report, State of Knowledge of AlcoholImpaired Driving: Research on Repeat DWI Offenders,” U.S. Department of
Transportation; National Highway Traffic Safety Administration., Washington DC
Kenkel, Donald (1993) “Drinking, Driving and Deterrence: The Effectiveness and Social
Costs of Alternative Policies,” Journal of Law and Economics, Vol. 36, No. 2. pp 877913
Kenkel, Donald and Steven Koch (2001) “Deterrence and Knowledge of the Law: The
Case of Drunk Driving,” Applied Economics, Volume 33, Pages 845-854
Klein T (1989) Changes in Alcohol Involving Fatal Crashes Associated with Tougher
State Alcohol Legislation (DTHN-22-88-C-7045). Washington, DC: US Dep. Transport.
National Highway Traffic Safety Administration
Kopits, Elizabeth and Maureen Cropper (2003) “Traffic Fatalities and Economic
Growth” The World Bank Development Research Group, Policy Research Working
Paper 3035, Washington DC
Lacey J, Jones R, Fell J (1996) “The effectiveness of the Checkpoint Tennessee
Program,” Proc. Assoc. Adv. Automotive Med., 40th, Vancouver, pp 275-82. Des Plaines,
IL
Loxley, Wendy et al. (1990) “Drinkers and their Driving: Compliance with DrinkDriving Legislation in Four Australian States” Canberra: Australian Institute of
Criminology, (AIC conference proceedings; no1) pages 83-94.
36
Mayhew DR and HM Simpson (1999), “Youth and Road Crashes: Reducing the Risks
from Inexperience, Immaturity and Alcohol,” Traffic Injury Research Foundation of
Canada, Ottawa, Ontario
Mayhew DR, Brown SW and HM Simpson (2002) The Alcohol-Crash Problem in
Canada: 2000, The Traffic Injury Research Foundation of Canada, Transport Canada, (TP
11759 E) Ottawa, Ontario
Mayhew DR, Brown SW and HM Simpson (2005) The Alcohol-Crash Problem in
Canada: 2003, The Traffic Injury Research Foundation of Canada, Transport Canada,
Ottawa, Ontario
National Center for Statistics and Analysis (2004) Traffic Safety Facts 2004: A
Compilation of Motor Vehicle Crash Data from the Fatality Analysis Reporting System
and the General Estimates System, NHTSA, DOT HS 809 919, Washington DC
National Highway Traffic Safety Administration (2000) “DWI in Other Countries”
Department of Transportation, Washington DC
Page, Yves (2001) “A Statistical Model to Compare Road Mortality in OECD
Countries,” Accident Analysis and Prevention, Volume 33, pages 371-385
Road Safety and Motor Vehicle Regulation Directorate (2004) Road Safety in Canada
2001, CCMTA, Standing Committee on Road Safety Research and Policy, TP 13951E,
Transport Canada,
Ruhm, Christopher (1994) “Economic Conditions and Alcohol Problems” National
Bureau of Economic Research, Working Paper No. 4914
Saffer H, Grossman M (1987) “Beer Taxes, the Legal Drinking Age, and Youth Motor
Vehicle Fatalities,” Journal of Legal Studies 16:351-74
Tessmer, Joseph (1999) “Comparing International Crash Statistics” Journal of
Transportation and Statistics, Volume 2, #2
United Nations (2002), Statistics of Road Traffic Accidents in Europe and North
America, United Nations Economic Commission for Europe, Geneva, Vol. 47
Wagenaar, Alexander (1981) “Effects of an Increase in the Legal Minimum Drinking
Age,” Journal of Health Policies. 2:206-25
Wagenaar, Alexander (1986) “Preventing Highway Crashes by Raising the Legal
Minimum Age for Drinking: The Michigan Experience 6 Years Later,” Journal of Safety
Research. Vol.17. pp 101-109
37
IX. Definitions and Abbreviations
9.1 Terms
BAC—Blood Alcohol Content
CCMTA—Canadian Council of Motor Transport Administrators
DUI—Driving Under the Influence
FARS—Fatal Accident Reporting System (U.S.)
MADD—Mothers Against Drunk Driving
MLDA—Minimum Legal Drinking Age
NCSA—National Center for Statistics and Analysis (U.S.)
NHSTA—National Highway Safety Traffic Administration (U.S.)
SADD—Students Against Drunk Driving, also known as Students Against
Destructive Decisions
TIRF—Traffic Injury Research Foundation (Canada)
VMT—Vehicle Miles Traveled
ZT—Zero Tolerance law
9.2 Definitions:
Traffic Accident:
Canada: A traffic collision which occurs on a public roadway that involves at
least one motorized vehicle and results in death or injury to at least one involved person,
or property damage exceeding a monetary amount (specified by law in each province or
territory). (IRTAD, Special Report, Definitions and Data Availability, Compilation and
Evaluation of A-Level Roads and Hospitalized Victims in OECD Countries, Accident
38
and Injury Definitions, OECD-RTR BASt, Road Transport Germany, Research Program
June 1998)
United States: An event that produces injury and/or damage, involves a motor
vehicle in transport, and occurs on a traffic way or while the vehicle is still in motion
after running off the traffic way. "Within the U.S. DOT/NHTSA the word "accident" will
no longer be used ...Continuous use of the word "accident" promotes the concept that
these events are outside of human influence or control. ... "Crash", "collision",
"incident"... are more appropriate terms and should be encouraged as substitute for the
word "accident". (NHTSA, and above IRTAD Special Report)
Injury Accident:
Canada: A traffic collision which occurs on a public roadway that involves at
least one motorized vehicle and results in non-fatal injuries to one or more involved
persons.
United States: A traffic accident that results in at least one person receiving an
injury as reported by the police as killed, incapacitating, non-incapacitating, but evident,
possible injury, or other injury, severity unknown.
Fatal Injury
Most of the countries, including Canada and the U.S. use the standard definition
given by the United Nations: "Any person who was killed outright or who died within 30
days as a result of the accident". A few countries have a different standard. In 1997 only
France, Italy, Portugal, Turkey and Korea do not adhere to the “30 days” definition.
39
(Statistics of Road Traffic Accidents in Europe and North America, Annex I, United
Nations, Economic Commission for Europe, Geneva, 1995)
Seriously Injured
Canada: A person involved in a traffic collision that occurs on a public roadway,
who suffers non-fatal injuries that result in hospitalization, including for observation
only, for a period of at least 24 hours.
United States: A police-reported incapacitating injury.
40