Content uploaded by Michael Scherer
Author content
All content in this area was uploaded by Michael Scherer on May 16, 2019
Content may be subject to copyright.
A comprehensive examination of U.S. laws enacted to reduce
alcohol-related crashes among underage drivers
Eduardo Romano, ⁎Michael Scherer, James Fell, Eileen Taylor
Pacific Institute for Research and Evaluation (PIRE), 11720 Beltsville Dr., Suite 900, Calverton, MD 20705, United States
abstractarticle info
Article history:
Received 15 January 2015
Received in revised form 18 May 2015
Accepted 13 August 2015
Available online 28 August2015
Keywords:
Young drivers
Alcohol
Policies
Communities
Fatal crashes
Introduction: To effectively address concerns associated with alcohol-related traffic laws, communities must
apply comprehensive and well-coordinated interventions that account for as many factors as possible. The goal
of the current research article is to examine and evaluate the simultaneous contribution of 20 underage drinking
laws and 3 general driving safety laws, while accounting for demographic, economic, and environmental vari-
ables. Methods: Annual fatal crash data (1982 to 2010), policies, and demographic, economic, and environmental
information were collected and applied to each of the 51 jurisdictions (50 states and the District of Columbia). A
structural equation model was fit to estimate the relative contribution of the variables of interest to alcohol-
related crashes. Results: As expected, economic factors (e.g., unemployment rate, cost of alcohol) and alcohol
outlet density were found highly relevant to the amount of alcohol teens consume and therefore to teens' im-
paired driving. Policies such as those regulating the age of bartenders, sellers, or servers; social host civil liability
laws; dram shop laws; internal possession of alcohol laws; and fake identification laws do not appear to have the
same impact on teens' alcohol-related crash ratios asother types of policies such as those regulating alcohol con-
sumption or alcohol outlet density. Conclusions: This effort illustrates the need for comprehensive models of
teens' impaired driving. Aftersimultaneouslyaccounting for as many factorsas possible, we found that in general
(for most communities) further reductions in alcohol-related crashes among teens might be more rapidly
achieved from efforts focused on reducing teens' drinking rather than on reducing teens' driving. Future efforts
should be made to develop models that represent specific communities. Practical applications: Based on this
and community-specific models, simulation programs can be developed to help communities understand and
visualize the impact of various policy alternatives.
© 2015 National Safety Council and Elsevier Ltd. All rights reserved.
1. Introduction
To reduce the prevalence of impaired driving and other alcohol-
related problems among underage Americans, states have passed a bat-
tery of laws, such as minimum legal drinking age (MLDA), graduated
driver licensing (GDL), and zero tolerance laws. Evidence shows that
these efforts have greatly reduced the involvement of underage drivers
in alcohol-related fatal crashes (Chen, Gruenewald, & Remer, 2009; Fell,
Fisher, Voas, Blackman, & Tippetts, 2008; Shults et al., 2001; Toomey,
Rosenfeld, & Wagenaar, 1996; Voas, Torres, Romano, & Lacey, 2012;
Wagenaar & Toomey, 2002). Despite these efforts, motor vehicle
crashes remain the leading cause of death for young people aged 16 to
20 years in the United States, accounting for approximately 28% of
deaths in that age group (Subramanian, 2012). Young drivers aged 15
to 20 years make up between 8% and 9% of the U.S. population but
only about 6.4% of licensed drivers. However, they are involved in 18%
of the fatalities resulting from traffic crashes each year (National
Highway Traffic Safety Administration [NHTSA], 2014). Drivers aged
16 years have crash rates that are three times greater than those for
drivers aged 17 years, five times greater than drivers aged 18 years,
and even two times greater than drivers aged 85 years (McCartt,
Shabanova, & Leaf, 2003). Explanations for why such a devastating
problem persists are varied, including concerns about alcohol laws no
longer being as effective as they were, or could be (Ferguson, Fields, &
Voas, 2000).
However, evaluating law effectiveness is not straightforward. There
is a complex and interrelated array of legal, demographic, and environ-
mental factors shaping the effectiveness of alcohol laws (Nelson et al.,
2013). Conceptually identical laws tend to vary from jurisdiction to
jurisdiction both regarding the number and type of provisions and
exemptions they contain (denoted in this document as the strength of
the law) as well as the way these laws are implemented (Fell,
Romano, & Voas, 2013; Voas & Fell, 2013). Alcohol-related traffic laws
may vary on effectiveness across population groups. For instance, recent
research indicates that although GDL laws reduced crash rates among
teenagers aged 16 to 17 years (Hartling et al., 2004; Shope & Molnar,
2003; Simpson, 2003), they have increased crash rates for drivers
Journal of Safety Research 55 (2015) 213–221
⁎Corresponding author at: Paci fic Institute for Research and Evaluation, 11720
Beltsville Dr., Suite 900, Calverton, MD 20705-3111, United States. Tel.: +1 301 755
2724; fax: +1 301 755 2799.
E-mail address: romano@pire.org (E. Romano).
http://dx.doi.org/10.1016/j.jsr.2015.08.001
0022-4375/© 2015 National Safety Council and Elsevier Ltd. All rights reserved.
Contents lists available at ScienceDirect
Journal of Safety Research
journal homepage: www.elsevier.com/locate/jsr
aged 18 years (Fell et al., 2013; Masten, Foss,& Marshall, 2011)andhave
demonstrated reduced effectiveness among Latinos when compared
with other teens (Romano, Fell, & Voas, 2011). Drivers' socioeconomic
status (Hasselberg & Laflamme, 2004; Laflamme & Diderichsen, 2000)
also impacts the effectiveness of alcohol-related traffic laws. Changes
in the prevailing economic conditions, from unemployment rates to
inflation rates (in particular, changes in the price of alcohol and gas),
further impact the effectiveness of trafficlaws(Bezruchka, 2009;
Buziarsist, 2009; Chi et al., 2011). Furthermore, the effectiveness of
alcohol-related traffic laws also depends on the effectiveness of laws
and regulations not specifictotraffic, such as those limiting the
availability of alcohol to youth (Gruenewald, Ponicki, & Holder, 1993;
Holder et al., 2000).
To effectively address the concerns associated with alcohol-related
traffic laws, communities must apply comprehensive and well-
coordinated interventions that account for as many factors as possible
(Holder, 1993; Holder, 2000; Holder et al., 2000; Holder, Saltz, Treno,
Grube, & Voas, 1997; Shults et al., 2009; Voas, 1997; Voas, Holder, &
Gruenewald, 1997). However, the interconnected factors contributing
to teens' drinking and driving and other alcohol-related problems
make the evaluation of laws difficult. Without a clear understanding of
which policies work better under different environments, communities
with a need for such policies and programs, and the means by which to
implement them, will find it difficult to decide howto prioritize their re-
source allocation to ensure optimal results. In the past, researchers have
attempted to provide help to these communities by developing
computer-based simulation programs that simultaneously account for
a variety of factors, which could assist them in their policy decision-
making (Holder & Blose, 1987; Kibel & Holder, 1994). However, those
modeling attempts were both limited in their scope and—without the
necessary maintenance—lost much of their relevance. For these com-
munities, a simulation model could help them: (a) more fully address
the underage (i.e., younger than 21 years) impaired driving problem;
(b) evaluate the expected impact of the alternative policy changes;
and (c) inform policymakers and community stakeholders about the
likely impact of allocated resources.
Funded by the Office of Juvenile Justice and Delinquency Prevention
(OJJDP), the goal of this research effort is to begin developing the scien-
tific framework that will be needed for such a tool. More specifically, the
goal of the current research article is to examine and evaluate the simul-
taneous contribution of 20 underage drinking laws and 3 general
driving safety laws—administrative license revocation/suspension
(ALR/ALS), blood alcohol concentration (BAC) .08 per se, and seat belt
laws—on the alcohol-related crash rates of underage drivers in the
United States. To address the complex environment in which laws
operate, we simultaneously control for the unique impact of variables
previously demonstrated to impact underage alcohol-related crashes.
These covariates include drivers' age and gender, unemployment
rates, vehicle miles traveled (VMT), cost of gasoline, cost of alcohol,
sobriety checkpoints (to account for law enforcement), alcohol outlet
density, and teen alcohol consumption.
2. Materials and methods
Annual fatal crash data (1982 to 2010), policies, and demographic,
economic, and environmental information were collected and applied
to each of the 51 jurisdictions (50 states and the District of Columbia).
Our analytical approach was based on setting the state and year as the
unit of analysis. The model we used, as well as the information and
the mechanisms used for analysis, are described below.
2.1. Model
Fig. 1 shows the model used in this effort. The outcome measure
appears on the rightmost side of the figure. Our model assumes that
for teenage drivers, the rates of alcohol-related fatal crashes in a certain
state and year depend directly on the teens' amount of driving and
alcohol (beer) consumption, the alcohol outlet density in that state
and year, and the MLDA-21 policies under study. The model assumes
that the number of miles driven by teenagers is influenced by the cost
of gasoline and the unemployment rate (a broad measure of the
economic environment). It assumes that the amount of alcohol they
consume varies with the cost of alcohol (Chaloupka, 2009; A.C.
Wagenaar, M.J. Salois, & K.A. Komro, 2009), the unemployment rate,
and the sex of the driver (Wilsnack, Wilsnack, Kristjanson, Vogeltanz-
Holm, & Gmel, 2009). It also assumes that, for teenagers, the rate of
alcohol-related crashes in a certain state and year is influenced by the
implementation of traffic safety laws not specifically targeted to young
drivers (.08 BAC per se, ALR/ALS, and seat belt laws) (Shults et al.,
2001; Voas, Tippetts, & Fell, 2000). The model in Fig. 1 also assumes
that the number of sobriety checkpoints conducted in a certain year (a
proxy for law enforcement intensity) influences the rate of alcohol-
related crashes (as reported by Shults et al., 2001). Finally, the model as-
sumes that the20 underage drinking lawshave a direct impact on teens'
crash rates, as well as an indirect impact through their influence on
alcohol consumption. To increase legibility, the laws appear collapsed
in Fig. 1.
2.2. Data and measures
2.2.1. Underage drinking laws
Drawing from legal data gathered by the Alcohol Policy Information
System (APIS)
1
and the STOP Act Report to Congress on the Prevention
and Reduction of Underage Drinking
2
, we obtained the effective dates of
statutes for 20 types of underage drinking laws for each of the 51 juris-
dictions in the United States comprised of the 50 states and the District
of Columbia. A summary descriptionof the 20 laws with their provisions
and exemptions appears in Table 1. Based on the provisions and exemp-
tions, Fell et al. (2015) developed a measure of the strength of the law in
each jurisdiction. For a detailed description of each type of law, its com-
ponents and provisions, the scoring matrix, and the strength of the law
in each state, see Fell et al. (2015).
For modeling purposes, we followed Fell et al. (2015,2009,2014),
who operationalized the existence and strength of each type of law as
follows. First, we coded years in whichthe law took effect from January
through December as “1”and years in which the law was not present at
any time as “0.”Years in which the effective dates of laws were after Jan-
uary 1st were coded as a proportion that indicated the fraction of the
year in which the law became effective. For example, a law that became
effective in October of any given year would only have been relevant for
the last quarter of the year and as such,would be coded as “.25”while all
years prior to the effective date would be coded as “0”and all years fol-
lowing the effective date would be coded as “1.”However, simply
employing a dichotomous classification for each law as being either
enacted or not enacted would not capture the state-by-state nuance of
each policy area. States differ from each other in the components of a
policy area by including provisions and exemptions that vary from
state to state. To remedy this, in the current research we utilize the scor-
ing system developed by Fell et al. (2015), which assigns a numerical
value based on provisions and exemptions for each law on a state-by-
state basis. The resulting score is what we refer to as the strength of
the law. Importantly, provisions and exemptions for each law that
were deemed to have a greater impact on the overall effectiveness of
the law were given more weight in the law strength score than weaker
ones. This allowed us to take into account the unique impact of individ-
ual provisions and exemptions for each law in each state.
Because laws vary in the maximum number of provisions they can
accommodate, we standardized—for modeling purposes—the number
1
http://alcoholpolicy.niaaa.nih.gov/.
2
https://www.s topalcoholabu se.gov/media/ReportToCongress/2013/r eport_main/
report_to_congress_2013.pdf.
214 E. Romano et al. / Journal of Safety Research 55 (2015) 213–221
of provisions a law contains in each jurisdiction on a scale from 0 to 1 by
dividing the number of provisions implemented in a state by the total
number of possible provisions for that law. For example, after examin-
ing the number of provisions a possession law has in each state, our
previous research determined the total score of different provisions
was 11. Then using this scoring system, we determined that a state
had exemptions to the possession law that resulted in a score reduction
from 11 to 8. The state then had a score of 8 out of a possible 11. In this
case, a state with a score of 8 would have a standardized possession
score of 8/11 = .73. The final measure of the strength of a law in a cer-
tain state and year was obtained by multiplying the standardized imple-
mentation score by the proportion of time in which the law was
implemented in that state. Following the example above, the variable
possession law in that state would have a value of “0”(.73 × 0 = 0) in
years before the law was in place, a score of “.18”in the year it was
implemented (.73 × .25 = .18) if it was implemented in October and
only active for one quarter of that year, and a score of “.73”for every
year after (.73 × 1 = .73).
2.2.2. BAC .08 per se law
The .08 BAC per se law targets adults by establishing a BAC limit for
drivers aged 21 years and older (when implemented, it reduced the BAC
limit in states from .10 to .08). Drivers at or above this BAC per se limit
are considered to be DWI; no other evidence is necessary. Despite not
being directly applicable to underage drivers (for which a BAC N.00
applies), the implementation of the adult-targeted per se law has been
shown to have a trickle-down impact on the drinking driver fatal
crash rates of drivers younger than 21 years (Fell et al., 2009; Voas,
2003). We used the NHTSA Digest of Impaired Driving and Selected
Beverage Control Laws, DOT HS 811–796, July 2013 to determine in
which year a state passed a BAC .08 per se law (NHTSA, 2013a).
2.2.3. Administrative license revocation or suspension
The ALR or ALS law mandates that drivers arrested for driving while
intoxicated have their driver's licenses automatically suspended with-
out the intervention of the court. Because the evidence supporting the
benefits of ALR/ALS to curb alcohol-related crashes is documented in
several studies (Klein, 1989; Voas, 2003; Voas et al., 2000; Wagenaar,
Maldonado-Molina, Ma, Tobler, & Komro, 2007), we decided to include
it in our model. We used data drawn from the Insurance Institute for
Highway Safety (IIHS, 2012) to determine the years a jurisdiction has
had an ALR/ALS law in place.
2.2.4. Seat belt laws
For seat belt laws, however, the evidence is strong that primary seat
belt laws—laws that allow officers to ticket for a seat belt violation per
se—are more effective in curbing alcohol-related crash rates than
secondary laws—laws that only allow officers to ticket for a seat belt vio-
lation after the driver is stopped for another suspected traffic violation
(such as speeding). These data were obtained from NHTSA's Summary
of Vehicle Occupant Protection Laws (2006) to determine for each
jurisdiction and year in the database whether the jurisdiction has laws
dictating mandatory seat belt use (primary, secondary, or none). The
-.322
-.055
-.241
-.104
-.122
-.022
-.075
-.038
-.023
-.003
-.517
-.888
-.043
-.023
-.152
-1.243
Cost of Gasoline
Cost of Alcohol
Unemployment
Female to Male
Ratio
.08 BAC Laws
MLDA-21 Laws1,2
Sobriety
Checkpoints
Seat Belt
Laws
ALR/ALS Laws
Vehicle Miles
Travelled
Alcohol Outlet
Density3
Beer
Consumption
Alcohol/Non-Alcohol
Ratio of Fatal Crashes
Fig. 1. Complete structural model. MLDA-21 lawsconsist of the followinglaws: possession, purchase, consumption, internal possession,use and lose, fake identification—minor, zero tol-
erance, graduated driver's license with nighttime restrictions, furni shing, age of serve rs, age of bartenders, age of sellers, keg re gistration, ret ail beverage service training, fake
identification—retail, social host prohibition, dram shop, social host civil liability, fake identification—production, and state control of alcohol. The regression weights are removed from
the current model for purposes of clarity. To see regression weights for these variables, see Table 2. Data for alcohol outlet density were only available for the State of California and
were run as a separate regression equation.
215E. Romano et al. / Journal of Safety Research 55 (2015) 213–221
Table 1
Laws examined in this effort.
Law Description Provisions
a
Possession Prohibits the possession of alcoholic beverages by those younger than 21 years. There are three location exemptions possible, including private locations, private residence,
and parents' home. Additionally, there is an exemption for parental/spousal presence.
Purchase Illegal for minors to purchase alcoholic beverages. There is a provision that allows youth to purchase alcohol for law enforcement purposes.
Consumption Prohibits the observed consumption of alcohol by minors. There are three location exemptions possible, including private locations, private residence,
and parents' home. Additionally, there is an exemption for parental/spousal presence.
Internal possession Illegal for minors to have evidence of alcohol in their body (i.e., by breath test, urine, etc.). There are three location exemptions possible, including private locations, private residence,
and parents' home. Additionally, there is an exemption for parental/spousal presence.
Use and lose License sanctions against minors found drinking, purchasing, or in possession of alcoholic
beverages.
Provisions if the law extends to purchase and possession laws, and if it is mandatory or
voluntary. There are additional provisions increasing the length of suspension and an
exemption placing the upper age limit at 21 years.
Use of fake identification among minors The use of false identification by a minor. Provisions for whether there are administrative procedures, judicial procedures, or both.
Zero tolerance Illegal for minors to drive with any measurable level of alcohol in their systems. Provisions for whether administrative and criminal sanctions are mandatory or discretionary
as well as the length of the sanction for each.
Graduated driver's license (GDL) A system in which beginning drivers are required to go through three stages of limited
driving privileges.
Implementation and length of nighttime and/or passenger restrictions.
Furnishing/selling Illegal to furnish alcoholic beverages to minors. There are three location exemptions possible, including private locations, private residence,
and parents' home. Additionally, there is an exemption for parental/spousal presence.
Age of on-premise alcohol sellers/servers Prohibition of those younger than 21 years to serve alcoholic beverages. Minimum service age for all three beverage types (beer, wine, spirits) and the presence of a
manager when alcohol is being sold.
Age of bartenders Prohibition of those younger than 21 years from bartending. Minimum service age for all three beverage types (beer, wine, spirits) and the presence of a
manager when alcohol is being sold.
Age of off-premise alcohol sellers/servers Prohibition of those younger than 21 years to sell alcoholic beverages. Minimum selling age for all three beverage types (beer, wine, spirits) and the presence of a
manager when alcohol is being sold.
Keg registration Prohibits sale or at least requires wholesalers or retailers to attach an identification number
to their kegs and collect identifying information from the keg purchaser. Extension of the information required—including the purchaser's identification and the
address where the keg will be consumed—the type of warning issued to a purchaser, and
whether a deposit is required. Additionally, the state of Utah prohibits keg use entirely.
Responsible beverage service training Requirements for retail alcohol outlets to participate in programs aimed to prevent alcohol
sales and service to minors and intoxicated persons, and train managers and servers/clerks to
implement policies and procedures effectively.
Requirement of whether the program is mandatory or voluntary, who is trained by the
program, incentives for having a program, and type of establishments and licensees covered.
Retailer support for fake identification State provisions to assist retailers in avoiding sales to potential buyers who present false
identification (e.g., issuing distinctive driver licenses to minors).
Variations in provisions and/or extension of sanctions.
Social host Prohibitions against hosting underage drinking parties. Provisions include general or specific statutes, type of actions triggering a violation, type of
property covered, and knowledge standard for a violation. Additionally, there is also an
exemption for preventive actions.
Dram shop liability The availability of private action against commercial alcohol providers. Type of law. Exemptions for who may be sued and how standards of proof necessary for a
violation.
Social host civil liability Private cause of action against a non-commercial alcohol provider for injuries or damages by
an intoxicated guest.
Type of law. Exemptions for who may be sued and how standards of proof necessary for a
violation.
Transfer/production of fake identification Prohibits the production of false identification and/or the lending or transferring of
identification to another person.
Variations in whether the law is criminalized.
State control of alcohol The use of state-run retail distribution systems. Variations on which kind of beverage is state run (beer, wine, spirits).
a
See Fell et al. (2015) for a detailed description of the laws, their strength coding, and the provisions/exemptions considered for each.
216 E. Romano et al. / Journal of Safety Research 55 (2015) 213–221
information from 2006 was carried up to 2010, as no change has occurred
in seat belt laws since then.
2.2.5. Sobriety checkpoints
Sobriety checkpoints were derived from previous research conducted
by Fell and colleagues (2003), who examined the frequency with which
sobriety checkpoints were conducted in each of the 51 jurisdictions, cod-
ing them as “0”if checkpoints were illegal or otherwise not conducted, “1”
if they were conducted infrequently (e.g., only during holidays), and “2”if
they were conducted frequently (e.g., on a monthly or weekly basis). As
we were not aware of any publication examining sobriety checkpoint fre-
quency in the same manner, the checkpoint strengths listed in Fell et al.
(2003) were considered constant through 2010.
2.2.6. Fatal crashes
Annual traffic fatality data were derived from NHTSA's Fatality
Analysis Reporting System (FARS) (NHTSA, 2013b). FARS is a continu-
ous census of vehicular crashes that resulted in the death of an individ-
ual within 30 days following the incident. The database also contains
information on the drivers' BAC, either collected from blood or imputed
when missing (Subramanian, 2002). We gathered data from 1982 to
2010, which allowed for the inclusion of all the measures needed for
analysis. Based on that information, for each state and year (our unit
of analysis), we computed the ratio of the number of alcohol-related
crashes among drivers aged 15 to 20 years divided by the number of
non-alcohol-related crashes among drivers of the same age group.
This ratio, called the crash incidence ratio (CIR), was used instead of
the number of crashes as an outcome measure to reduce bias in the
estimates as it allows for a better control of crash exposure over other
standardizing variables (Voas, Tippetts, Romano, Fisher, & Kelley-
Baker, 2007). The CIR took into account that drivers in the same age
group tend to face similar driving exposure (therefore equally affecting
both the CIR numerator and denominator), leaving alcohol (also
impacting the numerator) as the sole determinant of variation in the
CIR (Voas et al., 2007).
2.2.7. Sex of the driver
Previous research indicated that the sex of a teenage driver was a
strong indicator in both alcohol consumption (Wilsnack et al., 2009)
and fatal crash rates (Fell, 1977; Fell, Tippetts, & Voas, 2010). As such,
we deemed it necessary to include a measure of driver sex in the
model. We computed the ratio of females to males from data obtained
from the U.S. Census Bureau for each year of the current study
(1982–2010).
2.2.8. Alcohol use (beer)
Prior research indicated that when it comes to consuming alcohol,
though they consumed wine and spirits, underage drivers were more
likely to consume beer (Siegel et al., 2013). As such, we used beer
consumption in each state for those aged 15 years and older. Although
it would have been ideal to identify alcohol consumption only among
underage drivers, these data werenot available and statewide percapita
beer consumption was used as a proxy for teen beer consumption. Per
capita beer consumption rates were obtained for individuals aged
15 years and older by year and state from the annual publication of
the National Institute on Alcohol Abuse and Alcoholism's Alcohol
Epidemiologic Data System (2012).
2.2.9. Alcohol outlet density
Alcohol outlet density was deemed an important addition to the
current model as previous literature found that an increase in alcohol
outlet density allows underage populations to accessalcohol more read-
ily through commercial outlets, family members, and social networks
(Chen et al., 2009), resulting in significant increases in the rates of pur-
chasing and consuming alcohol (Rowland, Toumbourou, & Livingston,
2014) and, subsequently, motor vehicle crashes (Gruenewald &
Johnson, 2010). Information on alcohol outlet density was obtained
only for California from the California Department of Alcohol Beverage
Control. Because California state-specific data did not differ significantly
in terms of laws and outcome measures, it was deemed appropriate to
merge alcohol outlet density information into the national model.
2.2.10. Vehicle miles traveled
VMT were derived from Federal Highway Administration (FHWA)
data. The FHWA produces an annual estimate of total VMT by state
and year. Because the FHWA only offers annual data at the state level,
stratification by sex was not possible. As such, VMT for all drivers was
used in the current analyses.
2.2.11. Gasoline taxes
As the cost of gasoline was found to be too volatile for meaningful sta-
tistical comparison, gasoline tax rateswereusedastheyweredeemedto
be more stable and consistent indicators of market fluctuations, signifi-
cant enough to impact VMT and traffic fatalities (Grabowski & Morrisey,
2004). Gasoline tax rates were obtained from the Tax Foundation
(2014). Gasoline prices were adjusted for rate of inflation (Consumer
Price Index) and expressed in 2,010 dollars.
2.2.12. Alcohol taxes
Alcohol consumption has been directly associated with the cost of
alcohol, among youth in particular (Chaloupka, 2009; Chaloupka,
Saffer, & Grossman, 1993; Wagenaar, Salois, & Komro, 2009). Rather
than using alcohol prices to account for the cost of alcohol, many
researchers have preferred using alcohol excise taxes, which are more
stable and less dependent on local and sudden fluctuations (Wagenaar
et al., 2009). In this effort, we used alcohol excise taxes to control for
the cost of alcohol on alcohol consumption. Alcohol excise taxes were
derived from the Tax Foundation (2014). Alcohol prices were adjusted
for rate inflation and expressed in 2,010 dollars.
2.2.13. Unemployment rates
Unemployment statistics were derivedfrom the U.S. Bureau of Labor
Statistics, which publishes monthly employment statistics by state (U.S.
Bureau of Labor Statistics, 2014).
2.3. Data analysis
Based on the model in Fig. 1, we computed the direct and indirect ef-
fect the variables had on the outcome of interest (teens' alcohol-related
crash rates), as well as on the intermediate variables (VMT, alcohol
[beer] use). The data were analyzed using structural equation modeling
(SEM) techniques with Analysis of Moment-Based Structures (AMOS
v.21), an SPSS-based package (IBM SPSS Inc., Chicago, IL). SEM is a sta-
tistical technique frequently used to estimate causal relationships
based on qualitative assumptions represented in a path diagram.
3. Results
3.1. Direct effects
All regression estimates and p-values for the current model are
displayed in Table 2.Significant regression estimates indicate that a sin-
gle unit increase in the predictor variable would result in a change in the
outcome variable. For example, the predictor variable gas tax is signifi-
cant (p= .048) and results in an estimated −.322 change in the
outcome variable VMT. This means that for every unit increase in gas
tax, there is a 32.2% corresponding decrease in VMT. Direct effects are
also indicated in Fig. 1. Ideally, fit statistics for a structural model
would yield a nonsignificant χ
2
statistic (Barrett, 2007), a comparative
fitindex(CFI)andnormedfit index (NFI) of 0.95 or greater (Hu &
Bentler, 1999), and a root mean square error of approximation
(RMSEA) between 0.05 and 0.10 (MacCallum, Browne, & Sugawara,
217E. Romano et al. / Journal of Safety Research 55 (2015) 213–221
1996). Based on these guidelines, the current model displayed relatively
low fit—model fit statistics: χ
2
/df = 24.92***, CFI = 0.352, NFI = 0.348,
RMSEA = 0.190. However, fit statistics are merely considered guide-
lines and should be considered acceptable if they are comparable to fit
statistics previously established in the field (Bollen, 1989). We found
the fit statistics of the current model comparable to those established
by Fell and colleagues (2015;2009) in previous research and, as such,
retained the model.
3.1.1. Intermediate variable: vehicle miles traveled
Table 2 shows that the amount of teens' driving was highly
influenced by the cost of gasoline, the health of the economy (measured
by the unemployment rate), and the ratio in the number of female to
male teens in the state–year unit.
3.1.2. Intermediate variable: alcohol use
Table 2 shows that per capita beer consumption was significantly af-
fected, decreasing with an increase in the cost of alcohol, but increasing
with an increase in the density of alcohol outlets in a jurisdiction and the
number of female teens relative to that of males. Alcohol use was also
affected (reduced) by BAC .08 per se laws and fake identification laws
(minors and production).
3.1.3. Outcome variable: fatal crashes
Increases in alcohol-related crash ratios among teens were signifi-
cantly associated with a direct increase in teens' alcohol (beer)
consumption, VMT, and an increase in the ratio of female to male
teens in the state. Decreases in teens' alcohol-related crash ratios were
associated with sobriety checkpoints, seat belt laws, ALR/ALS laws, .08
BAC per se laws, and the following MLDA-21 laws: fake identification
(minors and retail), possession, purchase, social host civil liability, use
and lose, and zero tolerance. Conversely, fatal alcohol-related crashes
increased when keg registration laws increased, an unexpected finding
also reported by Fell et al. (2009).
3.2. Total (direct and indirect) effects
Direct effects such as those shown in Fig. 1 and Table 2 only tell part
of the story. Factors may also indirectly contribute to crash risk through
their impact on intermediate factors. After taking all direct and indirect
effects also into account, Table 3 shows the total effect (i.e., direct and
indirect effects) of all factors under consideration on the ratio of
alcohol-related teen fatal crashes. As shown in Table 3, the three factors
with the largest impact on the alcohol crash ratio were the ratio of female
to male teens in the state and the fake identification (retail) law (reducing
crash rates), and teens' alcohol (beer) consumption (increasing crash
rates). Other important contributing factors (reducing crash rates) were
the level of alcohol outlet density and the fake identification (minor),
possession, purchase, use and lose, and zero tolerance laws.
4. Discussion and conclusions
Similar to Fell et al. (2009), we found that possession and purchase
laws significantly predicted a decrease in underage drinking to non-
drinking drivers in the FARS dataset. However, Fell et al. detected a nota-
ble 16% decrease in FARS ratios for possession and purchase composite
scores, while we found only a 4.1% and a 4.5% decrease, respectively.
This difference is likely due to the methodology involved in the measuring
and analysis of these two distinct laws. Primarily, Fell et al. (2009) used a
dichotomous classification of laws in the original study, and as possession
Table 2
Regression weights and significance level for direct effects on alcohol-related crash ratios and teen alcohol use.
Vehicle miles traveled Beer consumption Teen alcohol crash ratios
Predictor Estimate p-value Estimate p-value Estimate p-value
Gas tax rate −.322⁎.048⁎––––
Alcohol tax rate ––−.055⁎.037⁎––
Unemployment rate −.517⁎b.001⁎
−.023 .546 ––
Ratio of female to male drivers −.888⁎b.001⁎
−.241⁎b.001⁎
−1.243⁎b.001⁎
BAC .08 laws ––−.104⁎b.001⁎
−.023⁎b.001⁎
Sobriety checkpoints ––––−.022⁎b.001⁎
ALR/ALS laws ––––−.075⁎b.001⁎
Seat belt laws ––––−.038⁎b.001⁎
Underage drinking laws
Age of bartender laws ––−.050 .351 −.007 .457
Age of seller laws ––.054 .402 −.018 .106
Age of server laws ––.021 .807 .029 .052
Consumption laws ––.017 .771 −.007 .445
Dram shop laws ––−.033 .573 ––
Fake identification—minor laws ––−.028⁎.043⁎
−.067⁎b.001⁎
Fake identification—Production laws ––.069 .271 −.008 .443
Fake identification—retail laws ––−.189⁎.049⁎
−.078⁎b.001⁎
Furnishing laws ––−.117 .416 ––
GDL nighttime laws ––––−.004 .779
Internal possession laws ––.048 .701 −.030 .154
Keg registration laws ––−.162 .122 .149⁎b.001⁎
Possession laws ––−.124 .107 −.032⁎.043⁎
Purchase laws ––−.047 .525 −.040⁎.040⁎
Responsible beverage service laws ––.059 .578 ––
Social host civil laws ––.021 .704 −.047⁎b.001⁎
Social host prohibition laws ––−.036 .708 .004 .825
State control of alcohol laws ––−.043 .827 .006 .836
Use and lose laws ––−.051 .688 −.085⁎b.001⁎
Zero tolerance laws ––−.007 .917 −.072⁎b.001⁎
Intermediate variables
Beer consumption ––––.152⁎b.001⁎
Alcohol outlet density
a
––.122⁎.002⁎.003 .233
VMT ––––.043⁎.004⁎
a
Data for alcohol outlet density was only available for the State of California and was run as a separate regression equation.
⁎Indicates statistical significance (pb.05).
218 E. Romano et al. / Journal of Safety Research 55 (2015) 213–221
and purchase laws were implemented simultaneously, the collinearity
between these two laws made separate examination impossible. In the
current research, however, we incorporated the strengths of laws. The
variability inherent in the law strengths allowed us to overcome the ob-
stacle of collinearity and measure these laws separately resulting in a
change in variations attributable to each law.
Overall, social host civil liability was also found to predict a nonsig-
nificant 1.4% decrease in underage drinking ratios, a finding consistent
with Fell and colleagues (2014). Fake identification—minor and
retail—resulted in a 14.1% and 9.3% overall reduction in underage
alcohol-related fatal crash ratios. This finding was somewhat surprising
as Fell and colleagues (2014) reported neither fake identification
minor nor retail to be a significant predictor of FARS ratios. This was
likely the result of the culmination of several design factors. First
and foremost, the structural model proposed in the current study
differed from those used by Fell et al. (2014) in that ours included
several additional control factors, including economic considerations
(i.e., alcohol and gasoline taxation), sex of drivers, and alcohol outlet
densities. This resulted in a substantially altered model, which would
necessarily require a redistribution of variance among individual pre-
dictors. Further, Fell et al. (2014) used a non-standardized law scoring
component, which may have limited the predictable impact of each
law. In the current research, we used standardized law strengths,
which could result in better fitting data and increased predictability of
the laws on FARS ratios.
Similar to previous research on the impact of MLDA-21 laws on
alcohol-related fatal crash ratios among adolescents, we found that
use and lose laws and zero tolerance laws both demonstrated a signifi-
cant impacton underage alcohol-related fatal crash ratios, evenafter ac-
counting for all other co-factors. Specifically, we found that use and lose
laws predicted a 6.9% overall decrease in crash ratios, while zero toler-
ance laws predicted a 3.3% decrease. This is similar to previous research,
which founda 5% decrease for both use and lose laws and zero tolerance
laws (Fell et al., 2009).
Consistent with findings by Fell et al. (2009), keg registration laws
demonstrated a 2.4% overall increase in underage alcohol-related fatal
crash ratios over the study period. Fell et al. theorized that by making
it more difficult to acquire kegs—and thereby more difficult to acquire
beer—underage populations may be prone to consuming spirits instead
of beer. As spirits have higher alcohol content, intoxication may be more
likely and an increase in drunk driving and fatal crashes may be
expected.
The implementation of GDL nighttime restriction laws was found to
cause a 4.5% decrease in underage alcohol-related fatal crash ratios. This
seems in line with previous research findings on the benefits of those
nighttime restrictions (Fell, Todd, & Voas, 2011; Hartling et al., 2004;
Shope & Molnar, 2003; Simpson, 2003). Also as expected, alcohol outlet
density was found to be an importantpredictor of alcohol use among in-
dividuals younger than 21 years (Chen et al., 2009; Rowland et al.,
2014). Althoughalcohol outlets were not directlyrelated to crash ratios,
their overall total impact on alcohol-related crash ratios was still found
to be significantly positive due to an indirect effect through alcohol use.
This was consistent with previous work by Gruenewald and Johnson
(2010), who found that a 10% increase in alcohol outlet density was as-
sociated with up to a 150% increase in single-vehicle nighttime crashes.
Interestingly, the economic variables that had been so influential in
shaping the intermediate variables (e.g., VMT, alcohol [beer] consump-
tion) did notappear to have a significant overall contribution to alcohol-
related crash rates.
The main focus of the current research was to examine the impact of
underage alcohol policies on beer consumption and underage fatal
crash ratios of drinking to nondrinking drivers after accounting for envi-
ronmental, population, economic, and enforcement factors. As expect-
ed, many of the findings of this effort have been previously reported
in the literature. New to this effort is the comprehensive and simulta-
neous evaluation of these contributors to teens' alcohol-related crash
risk.
This effort illustrates the need for comprehensive models of teens'
impaired driving. In examining the overall contribution of teens' drink-
ing and driving to teens' alcohol-related crash rates, we found that al-
though a reduction in teens' drinking rates translated into reductions
in teen's rates of impaired driving fatal crashes, reducing the amount
of miles traveled by teens apparently will not have such effect. We inter-
pret this finding as indicative that the problem of drinking and driving
among teens is related more to the relative risk of the alcohol they
consume than to their driving. Compared with older drivers, teens are
inexperienced, risky drivers, and current policies (e.g., GDL with night-
time restrictions) have been effective in reducing such risks. Therefore,
after simultaneously accounting for as many factors as possible, we
found evidence suggesting that in general (for most communities)
further reductions in alcohol-related crashes among teens might be
more rapidly achieved from efforts focused on reducing teens' drinking
rather than on reducing teens' driving.
To this regard and as expected, economic factors (unemployment
rate, cost of alcohol) and the environment (alcohol outlet density) are
highly relevant to the amount of alcohol teens consume. There is
ample evidence showing that increasing the cost of alcohol and/or reduc-
ing alcohol availability reduces teens' alcohol consumption. Although
economic factors do have a direct effect on alcohol consumption, and al-
cohol consumption is one of the largest contributors to teens' crash risk,
economic factors do not show a significant overall contribution to
alcohol-related crash rates. However, alcohol outlet density remained
closely associated with alcohol-related crash risks. Again, we interpret
this finding as indicative of the effect economic environment has on
teens' crash risk, mainly through teens' alcohol consumption.
Also of interest are the set of laws and policies that after accounting
for all other factors do not seem to have a significant impact on teens'
alcohol-related crashes. Policies such as those regulating the age of bar-
tenders, sellers, or servers; social host civil liability laws; dram shop
laws; internal possession laws; and fake identification (production)
Table 3
Regression weights and significance level for total effects on fatal crash ratios.⁎
Predictor Total effect size
Age of bartender laws −0.1%
Age of seller laws −0.9%
Age of server laws +3.2%
ALR/ALS laws −7.0%⁎
Consumption laws −0.5%
Dram shop laws −0.5%
Fake identification—minor laws −7.2%⁎
Fake identification—production laws −1.0%
Fake identification—retail laws −10.6%⁎
Furnishing laws −11.7%
GDL nighttime laws −0.4%
Internal possession laws −2.3%
Keg registration laws +12.4%⁎
Possession laws −6.4%⁎
Purchase laws −4.9%⁎
Responsible beverage service laws +0.9%
Social host civil laws −4.4%⁎
Social host prohibition laws −0.2%
State control of alcohol laws +0.4%
Use and lose laws −8.5%⁎
Zero tolerance laws −7.3%⁎
.08 BAC per se law −8.8%⁎
Seat belt laws −3.8%⁎
Beer consumption +15.2%⁎
VMT 0.0%
Ratio of female to male drivers −124.1%⁎
Sobriety checkpoints −2.2%⁎
Alcohol tax rate −1.0%
Alcohol outlet density +3.4%⁎
Gas tax rate −1.5%
Unemployment rate −0.2%
⁎Indicates statistical significance (pb.05).
219E. Romano et al. / Journal of Safety Research 55 (2015) 213–221
laws do not have a significant impact on teens' overall alcohol-related
crash ratios. Nevertheless, their lack of significance should not be
interpreted as irrelevant. Rather than examining the impact of individ-
ual laws on reducing teens' involvement in alcohol-related crashes,
we should examine the impact of the entire body of legislation on
such crashes. Although some individual laws may appear to be nonsig-
nificant contributors to risk reduction, these laws may complement
others andcontribute to the collective effectiveness of these laws. In ad-
dition, while our outcome measure was drinking driver rates in fatal
crashes, some of these MLDA-21 laws may impact underage alcohol
consumption, binge drinking, and alcohol-related suicides and/or homi-
cides, or may delay theonset age of drinking by youth, which would not
be adequately represented in our analyses.
Having said that, it is important to acknowledge that laws and policies
differ in their individual contributions to the abatement of teens' impaired
driving. Such a difference must be taken into account by communities
willing to address the problem of teens' impaired driving. To do so, com-
munities must decide on the set of policies they need to enact based on
the set of policies they already have in place and the resources available.
Communities need to choose and enact the policies that would allow
them to efficiently achieve the desired result. It is with this regard that
this effort becomes most useful—by providing a broad framework for
states and communities to: (a) examine the potential impact of alcohol-
related policies on teen drinking and driving, and (b) determine the utility
of devoting limited resources to implementing new policies and/or
strengthening existing policies to curtail underaged drinking and driving.
More specific models should be developed to target specificcommunities.
Based on these models, simulation programs could be developed to help
communities understand and visualize the impact of policy alternatives. A
simulation model similar to the one described here, which allows users to
predict how individual changes in law strengths could impact underage
drinking and driving, would help inform the decision-making process.
The current research is meant to serve as the foundation for such a
simulation. The structural model proposed and tested in the current en-
deavor examined the unique contributions of 20 underage drinking
laws, administrative license revocation/suspension, .08 per se BAC limits,
seat belt laws, economic factors, environmental factors, and driver factors
on the alcohol-related crash rates of underage drivers in the United States.
Each of these components was selected because previous research indi-
cated its potential impact on underage drinking and driving and/or fatal
crashes. Future research will examine how slight alterations of each of
the variables proposed in the current model impact predicted FARS ratios
for underage drivers. By doing so, we will begin to develop a tool that
could be used to inform researchers, advocates, policymakers, and com-
munity groups in their efforts to understand how implementing new or
altering existing policies may impact underage drinking and driving and
maximize gains made by limited resources.
This effort is not free of limitations. The impact of some of the laws
may be dampened by the use of ratios in our outcome measure and by
the relatively low number of states that have enacted those particular
laws (i.e., internal possession, age of servers, and state control of
alcohol). Alcohol outlet density data were only available in the State of
California, while all other data were available on a national level.
Although this was a necessary limitation given the extreme variation
in recordsof alcohol outlets from state to state and by year, it neverthe-
less may affect the impact of this variable in the model. Additionally, the
current research examines the impact of these laws on alcohol-related
outcomes only (i.e., per capita beer consumption and underaged
alcohol-related FARS ratios). This may to some extent obscure the effec-
tiveness of these laws in preventing adverse non-alcohol related out-
comes, which may be evident in some of the more general laws
(e.g., GDL nighttime restrictions, seatbelt safety). Although this would
have been beyond the scope of the current study, in considering the ef-
fectiveness of all laws, additional outcomes may be necessary. Perhaps
the most important limitation of this study is that despite its relative
comprehensiveness, it was not possible to incorporate all the factors
that have been shown to contribute to the underage drinking and driv-
ing problem into the model. Such incorporation would have caused the
model to become intractable. Subsequently, it is possible that by ignor-
ing some of these potentially contributing factors, we somehow biased
the results. To this regard, a possible way to address this limitation in fu-
ture research efforts is to circumscribe the comprehensive analyses of
underage drinking problems to individual communities or regions,
each with specificidentified problems and risk factors.
Acknowledgements
The research and preparation of this manuscript were conducted
under a grant from the Office of Juvenile Justice and Delinquency
Prevention, U.S. Department of Justice (2012-AH-FX-0005). Points of
view or opinions in this document are those of the authors and do
not necessarily represent the official position or policies of the U.S.
Department of Justice.
References
Barrett, P. (2007). Structural equation modeling: Adjudging model fit. Personality and
Individual Differences,42,815–824.
Bezruchka, S. (2009). The effect of economic recession on population health. Canadian
Medical Association Journal,181(5), 281–285.
Bollen, K. A. (1989). Structural equations with latent variables. New York, NY: John Wiley.
Buziarsist, J. (2009). Economic crisis on public health. Archives of Public Health,67(3), 97.
Chaloupka, F. J. (2009). Alcoholic beverage taxes, prices and drinking. Addiction,104(2),
191–192 [Commentary].
Chaloupka, F. J., Saffer, H., & Grossman, M. (1993).Alcohol control policies and motor vehicle
fatalities: National Bureau of Economic Research.
Chen, M. -J., Gruenewald, P. J., & Remer, L. G. (2009). Does alcohol outlet density affect
youth access to alcohol? Journal of Adolescent Health,44(6), 582–589.
Chi, G., Zhou, X., McClure, T. E., Gilbert, P.A., Cosby, A.G., Zhang, L., ... Levinson, D. (2011).
Gasoline prices and their relationship to drunk-driving crashes. Accident Analysis and
Prevention,43(1), 194–203.
Fell, J. C. (1977). Aprofile of fatal accidents in volving alcohol. Washington, DC: U.S.
Department of Transportation, National Highway Traffic Safety Administration.
Fell, J. C., Ferguson, S. A.,Williams, A. F., & Fields, M. (2003). Whyare sobriety checkpoints
not widely adopte d as an enforceme nt strategy in the U nited States? Accident
Analysis and Prevention,35(6), 897–902.
Fell, J. C., Fisher, D. A., Voas, R. B., Blackman, K.,& Tippetts, A. S. (2008). The relationship of
underage drinki ng laws to reductions in drinking drivers in fatal crashes in the
United States. Accident Analysis and Prevention,40, 1430–1440.
Fell, J. C., Fisher, D. A., Voas, R. B., Blackman, K., & Tippetts, A. S. (2009). The impact of un-
derage drinking laws on alcohol-related fatal crashes of young drivers. Alcoholism:
Clinical and Experimental Research,33(7), 1208–1219.
Fell, J. C., Romano, E., & Voas, R. B. (2013). A national evaluation of graduated driver li-
censing laws in the United States. 20th International Council on Alcohol, Drugs and
Traffic Safety (Vol. August 25 to 28, 2013). Brisbane, QLD, Australia: International
Council on Alcohol, Drugs and TrafficSafety(ICADTS).
Fell, J. C., Scherer, M., Thomas, S., & Voas, R. B. (2014). Effectiveness of social host and fake
identification laws on reducing underage drinking driver fatal crashes. TrafficInjury
Prevention,15(Sup 1), S64–S73 (Special Issue). doi: 10.1080/15389588.2014.928929.
Fell, J. C.,Tippetts, A. S., & Voas,R. B. (2010). Drinkingcharacteristics of drivers arrested for
driving while intoxicated in two police jurisdictions. TrafficInjuryPrevention,11(5),
443–452. http://dx.doi.org/10.1080/15389588.2010.488274.
Fell, J. C.,Todd, M., & Voas, R. (2011). A nationalevaluation of the nighttime and passenger
restriction components of graduated driverlicensing. Journalof Safety Research,42(4),
283–290. http://dx.doi.org/10.1016/j.jsr.2011.06.001.
Fell, J., Thomas, S., Scherer, M., & Fisher, D. A. (2015). Scoring the strengths and weak-
nesses of underage dri nking laws in the Unit ed States. World Medical a nd Health
Policy (in press).
Ferguson, S. A., Fields, M., & Voas, R. B. (2000). Enforcement of zero tolerance laws in the
US —Prevention section. In H. Laurell, & F. Schlyter (Eds.), Alcohol, drugs and traffic
safety —T 2000: Proceedings of the 15th International Conference on Alcohol, Drugs
and Traffic Safety, May 22–26, 2000. 2.(pp.713–718). Stockholm, Sweden: ICADTS.
Grabowski, D. C., & Morrisey, M. A. (2004). Gasoline prices and motor vehicle fatalities.
Journal of Policy Analysis and Management,23(3), 575–593.
Gruenewald, P. J., & Johnson, F. W. (2010). Drinking, driving, and crashing: A traffic-flow
model of alcohol-related motor vehicle accidents. Journal of Studies on Alcohol and
Drugs,71(2), 237–248.
Gruenewald, P. J., Ponicki, W. R., & Holder, H. (1993). The relationship of outlet densities
to alcohol consumption: A time series cross-sectional analysis. Alcoholism: Clinical
and Experimental Research,17(1), 38–47.
Hartling, L., Wiebe, N., Russell, K., Petruk, J., Spinola, C., & Klassen, T. P. (2004). Graduated
driver licensing for reducing motor vehicle crashes among young drivers. Cochrane
Database of Systematic Reviews,2Art. No.: CD003300.pub003302.
Hasselberg, M., & Laflamme, L. (2004). Children at risk in traffic: Improvement potentials
in the Swedish context. Acta Paediatrica,93(1), 113–119.
220 E. Romano et al. / Journal of Safety Research 55 (2015) 213–221
Holder, H. D. (1993 ). Prevention of alcohol-related accidents in the community. Addiction,
88(7), 1003–1012.
Holder, H. D. (2000). Community prevention of alcohol problems. Addictive Behaviors,
25(6), 843–859.
Holder, H. D., & Blose, J. O. (1987). The reduction of community alcohol problems: Computer
simulation experiments in three counties. Journal of Studies on Alcohol,48(2), 124–135.
Holder,H.D.,Gruenewald,P.J.,Ponicki,W.R.,Treno,A.J.,Grube,J.W.,Saltz,R.F.,...Roeper,P.
(2000). Effect of community-based interventions on high-risk drinking and alcohol-
related injuries. Journal of the American Medical Association,284(18), 2341–2347.
Holder, H. D., Saltz, R. F., Treno, A. J., Grube, J. W., & Voas, R. B. (1997). Evaluation design for a
community prevention trial: An environmental approach to reduce alcohol-involved
trauma. Evaluation Review,21(2), 140–165.
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Con ventional cr iteria versus new alternati ves. Structural Equation
Modeling,6,56–83.
Insurance Institute for Highway Safety (2012). Alcohol-impaired driving state laws. Re-
trieved December 10, 2012, from http://www.i ihs.org/iihs/ topics/laws/ dui?
topicName=alcohol-impaired-driving
Kibel, B. M., & Holder, H. D. (1994). Dynamic systems-modeling as a means to estimate
community-based prevention effects. In L. M. Collins, & L. A. Seitz (Eds.), Advances
in data analysis for prevention intervention research (National Institute on Drug Abuse
Monograph 142) (pp. 404–446). Washington, DC: Superi ntendent of Documents,
U.S. Government Printing Office.
Klein, T. M. (1989). Changes in alcohol-involved fatal crashes associated with tougher state
alcohol legislation. Washington, DC: National Highway Traffic Safety Administration.
Laflamme, L., & Diderichsen, F. (2000). Social differences in traffic injury risks in childhood
and youth—A literature review and a research agenda. Injury Prevention,6(4), 293–298.
MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determina-
tion of sample size for covariance structure modeling. Psychological Methods,1(2),
130–149.
Masten, S. V., Fo ss, R. D., & Marsha ll, S. W. (2011). Graduated driver licensing and
fatal crashes involving 16- to 19-year-old drivers. Journal of the Ame rican Medical
Association,306(10), 1098–1103.
McCartt, A. T., Shabanova, V. I., & Leaf, W. A. (2003). Driving experience, crashes and traf-
fic citations of te enage beginnin g drivers. Accident Analysis and P revention,35,
311–320.
National Highway Traffic Safety Administrat ion (2006). Summary of vehicle occupant
protection laws: Seventh edition (current as of January 1, 2006). Washington, DC: U.S.
Department of Transportation, National Highway Traffic Safety Administration.
National Highway Traffic Safety Administration (2013a). Digest of impaired drivingand se-
lected beverage control laws. (DOT HS 811 796) Washington, DC: U.S. Department of
Transportation, National Highway Traffic Safety Administration.
National Highway T raffic Safety Administration (2013b). Fatality analysis reporting sys-
tem data files, 1982–2010. Retrieved May 3, 2013, from http://www.nhtsa.gov/FARS
National Highway Tr affic Safety Adminis tration (2014). traffic safety facts 2012 data:
Young drivers. (DOT HS 812 019) Washington, DC: US Department of Transportation,
National Highway Traffic Safety Administration.
National Institute on Alcohol Abuse and Alcoholism (2012). Apparent per capita alcohol
consumption: National, state, and regionaltrends, 1977–2012. Washington, DC: Nation-
al Institute of Health, National Institute on Alcohol Abuse and Alcoholism.
Nelson,T.F.,Xuan,Z.,Babor,T.F.,Brewer,R.D.,Chaloupka,F.J.,Gruenewald,P.J.,...Naimi,T.
S. (2013). Efficacy and the strength of evidence of U.S. alcohol control policies. American
Journal of Preventive Medicine,45(1), 19–28. http://dx.doi.org/10.1016/j.amepre.2013.
03.008.
Romano, E., Fell, J., & Voas, R. B. (2011). Theroleofraceandethnicityontheeffectofgrad-
uated driver licensing laws in the United States. Annals of Advances in Automotive
Medicine,55,51–61.
Rowland, B., Toumbourou, J. W., & Livingston, M. (2014). The association of alcohol outlet
density with illegal underage adolescent purchasing of alcohol. Journal of Adolescent
Healthhttp://dx.doi.org/10.1016/j.jadohealth.2014.08.005.
Shope, J. T., & Molnar, L. J. (2003). Graduated driver licensing in the United States: Evalu-
ation of results from early programs. Journal of Safety Research,34(1), 63–69.
Shults, R. A., Elder, R. W., Nichols, J. L., Sleet, D. A., Compton, R., & Chattopadhyay, S. K. (2009).
Effectiveness of multicomponent programs with community mobilization for reducing
alcohol-impaired driving. American Journal of Preventive Medicine,37(4), 360–371.
Shults, R. A., Elder, R. W., Sleet, D. A., Nichols, J. L., Alao, M. O., Carande-Kulis, V. G., ...
Thompson, R. S. (2001). Reviews of evidence regarding interventions to reduce
alcohol-impaired driving. American Journal of Preventive Medicine,21(4 Suppl), 66–88.
Siegel, M., DeJong, W., Naimi, T. S., Fortunato, E. K., Albers, A. B., Heeren, T., ... Rodkin, S.
(2013). Brand-s pecific consumption of alcohol amo ng underage youth in the
United States. Alcoholism: Clinical and Experimental Research,37(7), 1195–1203.
Simpson, H. M. (2003). The evolution and effectiveness of graduated licensing. Journal of
Safety Research,34,63–69.
Subramanian, R. (2002). transitioning to multiple imputation —A new method to estimate
missing blood alcohol concentration (BAC) v alues in FARS. ( DOT HS 809 403).
Washington, DC: Mathematical Analysis Division, National Center for Statistics and
Analysis, National Highway Traffic Safety Administration, U.S. Department of Trans-
portation Retrieved from http://www-nrd.nhtsa.dot.gov/Pubs/809-403.PDF.
Subramanian, R. (2012). Motor vehicle traffic crashes as a leading cau se of death in the
United States, 2008 and 2009. (DOT HS 811 620) Washington, DC: U.S. Department
of Transportation, National Highway of Traffic Safety Administration Retrieved from
http://www-nrd.nhtsa.dot.gov/Pubs/811620.pdf.
Tax Foundation (2014). State sales, gasoline, cigarette, and alcohol tax rates by state. Re-
trieved October 30, 2014, from http://taxfoundation.org/article/state-sales-gasoline-
cigarette-and-alcohol-tax-rates
Toomey, T. L., Rosenfeld, C., & Wagenaar, A. C. (1996). The minimum legal drinking age:
History, effectiveness, and ongoing debate. Alcohol Health and Research World,20(4),
213–218.
U.S. Bureau of Labor Statistics (2014). Regional and state unemployment (Annual). Re-
trieved October 30, 2014, from http://www.bls.gov/schedule/archives/all_nr.htm
#SRGUNE
Voas, R. B. (1997). Drinking and driving prevention in the community: Program planning
and implementat ion. Addiction,92(Supplement 2), S201–S219.
Voas, R. B. (2003). Robert F. Borkenstein: An appreciation. Addiction,98(3), 371–372.
Voas, R. B., & Fell, J. C. (2013). Strengthening impaired-driving enforcement in the United
States. Traffic Injury Prevention,14(7), 661–670. http://dx.doi.org/10.1080/15389588.
2012.754095.
Voas, R. B.,Holder, H.D., & Gruenewald, P. J. (1997). The effect of drinking and driving in-
terventions on a lcohol-involve d traffic crashes within a comprehensive community
trial. Addiction,92(Suppl. 2), S221–S236.
Voas, R. B., Tippetts, A. S., & Fell, J. C. (2000). The relationship of alcohol safety laws to
drinking drivers in fatal crashes. Accident Analysis and Prevention,32(4), 483–492.
Voas, R. B., Tippetts, A. S., Romano, E., Fisher, D. A., & Kelley-Baker, T. (2007). Alcohol in-
volvement in fatal crashes under thr ee crash exposure measures. TrafficInjury
Prevention,8(2), 107–114.
Voas, R. B., Torres, P., Romano, E., & Lacey, J. H. (2012). Alcohol-related risk of driver fatalities:
An update using 2007 data. Journal of Studies on Alcohol and Drugs,73(3), 341–350.
Wagenaar, A. C., & Toomey, T. L. (2002). Effects of minimum drinking age laws: Review
and analysesof the literature from 1960 to 2000. Journal of Studies on Alcohol,Supple-
ment No. 14,206–225.
Wagenaar, A. C., Maldonado-Molina, M., Ma, L., Tobler, A., & Komro, K. (2007). Effects of
legal BAC limits on fatal crash involvement: Analyses of 28 states from 1976 through
2002. Journal of Safety Research,38,493–499.
Wagenaar, A. C., Salois, M. J., & Komro, K. A. (2009). Effects of beverage alcohol price and
tax levels on drinking: A meta-analysis of 1003 estimates from 112 studies. Addiction,
104,179–190.
Wilsnack, R. W., Wilsna ck, S. C., Kristjanson, A. F., Vogeltanz-Holm , N. D., & Gmel, G.
(2009). Gender and alcohol consumption: Patterns from the multinational GENACIS
project. Addiction,104(9), 1487–1500.
Eduardo Romano is a senior research scientist at PIRE in Calverton, MD. His past work in-
volved estimating the incidence and cost of national and state intentional and unintentional
injuries, and the evaluation of Mexican policies aimed to deter binge drinking by young
American visitors in Tijuana (Mexico). As a principal investigator (PI), he has participated
in efforts funded by the National Institutes of Health (NIH) to evaluate the involvement in
crash-risk situations of women and different minority groups; as well in a project funded
by the National Highway Traffic Safety Administration to study the involvement in trafficvi-
olations of recent immigrants to the United States.He is currently the PI in a NIH-funded pro-
ject looking at estimating alcohol-related and drug-related relative risks, and on an Office of
Juvenile Justice and Delinquency Prevention project to study the impact of alcohol-related
laws and policies on teens' impaired driving. Dr. Romano holds a Ph.D. in Agricultural and Ap-
plied Economics from the Virginia Polytechnic Institute and State University.
Michael Scherer is an associate research scientist and statistician at PIRE in Calverton, MD.
His research has focused primarily on the adverse impact of substance use and misuse on
outcomes such as motor vehicle crashes or DUI/DWI convictions. He is also interested in
the role of social networks, cognitive abilities and the personality factors and the role they
play in risk-taking behavior. His previous research focused on developing, implementing
and evaluating innovative methods in addressing substance use and the adverse impact it
has on people's lives. He has an M.S. in Rehabilitation Counseling from the Medical College
of Virginia and a Ph.D. in Counseling Psychology from Virginia Commonwealth University.
James C. Fell is a senior research scientist with the Pacific Institute for Research and Evalua-
tion (PIRE) in Calverton, MD. He recently completed research on the effectiveness of gradu-
ated driver licensing laws under a grant from the National Institute of Child Health and
Human Development (NICHHD), on enforcement intensity measures and impaired driving
on the roads for the National Institute on Alcohol Abuse and Alcoholism (NIAAA) and studies
on responsible beverage service, alcohol ignition interlock laws, high visibility enforcement
and alcohol monitoring devices on impaired driving of fenders for the National Highway Traf-
fic Safety Administration (NHTSA). He is currently working on three grants from the NIAAA
studying the effects of various underage drinking laws on underage drinking and driving fatal
crashes, the length of administrative license suspens ion on impaired driving recidivism, and a
meta-analysis of the effectiveness of lowering blood alcohol concentration (BAC) limits for
driving to .08 and to .05. In addition, he is co-principal investigator on a large demonstration
project on high visibility enforcement of impaired driving for NHTSA.
Mr. Fell formerly worked at NHTSA from 1969 to 1999 and has 47 years of trafficsafetyand
alcohol research experience. He has both a Bachelor's and Master's degree in Human Factors
Engineering from the State University of New York at Buffalo. Mr. Fell is a member and cur-
rently Secretary-Elect of the International Council on Alcohol, Drugs, and Traffic Safety
(ICADTS), and a member of the Association for the Advancement of Automotive Medicine
(AAAM), the Research Society on Alcoholism (RSA), the Society for Prevention Research
(SPR) and the Human Factors and Ergonomics Society (HFES). He received the Widmark
Award from ICADTS in 2013 for outstanding, sustained and meritorious contribution to the
field of alcohol, drugs and trafficsafety.
Eileen Taylor is a program director at PIRE in Calverton, MD. A former probation/parole
officer, Ms. Taylor has managed traffic safety research, evaluation and field projects for
PIRE since 1988.Ms. Taylor holds a Master of Science Degree in Criminal Justice Adminis-
tration from American University, as well as a Bachelor's degree from the University of
Michigan–Dearborn.
221E. Romano et al. / Journal of Safety Research 55 (2015) 213–221