Johnson and Clapp 2011

JOHNSON AND CLAPP

79

Impact of Providing Drinkers With “Know Your Limit”
Information on Drinking and Driving: A Field Experiment*
MARK B. JOHNSON, PH.D.,† AND JOHN D. CLAPP, PH.D.†
Pacific Institute for Research and Evaluation, 11720 Beltsville Drive, Suite 900, Calverton, Maryland 20705

ABSTRACT. Objective: Given that most effective alcohol harm-reduction laws specify the blood alcohol concentration (BAC) that constitutes
illegal behavior (e.g., the .08% breath alcohol concentration legal limit),
interventions that allow drinkers to accurately estimate their BACs, and
thus better assess their risk, have potential importance to long-term
driving-under-the-influence prevention efforts. This study describes a
field experiment designed to test the impact on drinking of providing
“Know Your Limit” (KYL) BAC estimation cards to individuals in a
natural drinking environment. Method: We randomly sampled 1,215
U.S. residents as they entered Mexico for a night of drinking, interviewed
them, and randomly assigned them to one of six experimental conditions.
Participants were reinterviewed and breath-tested when they returned to

the United States. The experimental conditions included providing generic warnings about drinking and driving, giving out gender-specific BAC
calculator cards (KYL cards), and providing incentives to moderate their
drinking. Results: Cueing participants about the risks of drunk driving
resulted in significantly lower BACs (relative to control) for participants
who indicated that they would drive home. Providing KYL matrixes did
not reduce BACs, and, in fact, some evidence suggests that KYL cards
undermined the effect of the warning. Conclusions: KYL information
does not appear to be an effective tool for reducing drinking and driving.
Implications for prevention and future research are discussed. (J. Stud.
Alcohol Drugs, 72, 79-85, 2011)

A

difficult to target because they come from many locations
and social groups and are brought together for brief periods
in differing locations at unannounced times. Youth living
along the U.S. border who cross into Mexico or Canada to
drink provide such a target group (Clapp et al., 2001; Lange
and Voas, 2001).
Prevention programs targeting young adults have used
numerous environmental approaches, from media campaigns
to law enforcement and combinations of the two (Clapp et
al., 2006). Although broad, mass media approaches to the
reduction of drinking often provide general information on
alcohol risks, individuals may lack a method for reacting
appropriately to the message content. This tends to be true
of messages relating to laws setting an illegal BAC level,
perhaps because individuals have no direct method to assess
their own BAC levels. In response to this problem, some
countermeasure programs attempt to make risks salient by
providing drinkers with the tools to obtain more accurate
estimates of their BAC levels (Royal, 2000). State driver’s
education classes and “Alcohol 101”–style programs offered
to college students also teach BAC estimation by instructing young people how to associate alcohol symptoms to
approximate BACs (e.g., Larsen and Kozar, 2005; Sharmer,
2001; www.centurycouncil.org/see-our-work/evaluation/
alcohol101plus). This approach appears reasonable because
many effective alcohol harm-reduction laws are specified in
terms of BAC levels.
One method of providing such information is a graphic
presentation of the relationship of gender, weight, and the
number of drinks in a fixed period to the estimated BAC
level of the drinker. This information can be presented on a
wallet-size card typically called “Know Your Limit” (KYL)

LCOHOL-RELATED DRIVING FATALITIES remain
a serious public health problem. Alcohol-impaired
driving accidents accounted for 32% of all traffic fatalities,
representing more than 11,500 deaths in 2008 (National
Highway Traffic Safety Administration [NHTSA], 2008).
Young people accounted for about one third of those crashes
and are at higher risk for crashes than older adults at any
blood alcohol concentration (BAC; Zador et al., 2000). (Note
that throughout this article, BAC indicates blood alcohol
concentration as a concept, but breath alcohol concentration
[BrAC] indicates blood alcohol concentrates estimated via
breath samples and collected as part of the research.) The
data on crashes have led to a renewed concern with impaired
driving and the initiation of a NHTSA effort to promote new
state media programs tied to enforcement programs directed
at deterring drinking drivers. These media programs are most
successful when they have local and immediate relevance to
the community and where they can be delivered directly to
the target group. An important target for such messages are
young people who congregate to drink in unsupervised locations, such as fraternity houses, private homes when the parents are away, and remote deserts or beaches, where alcohol
is free or free flowing and peer support for heavy drinking is
strong (Clapp et al., 2002, 2009). Typically, these youths are

Received: March 4, 2010. Revision: June 22, 2010.
*This study was supported by National Institute on Alcohol Abuse and
Alcoholism grant R21 AA014633 awarded to Mark B. Johnson.
†Correspondence may be sent to Mark B. Johnson at the above address
or via email at: mjohnson@pire.org. John D. Clapp is with the Center for
Alcohol and Drug Studies, San Diego State University, San Diego, CA.

79

80

JOURNAL OF STUDIES ON ALCOHOL AND DRUGS / JANUARY 2011

cards. KYL cards have been widely used for 40 years, correlating with the use of BAC tests to enforce impaired-driving
laws. Surveys have demonstrated that the public has a poor
understanding of the number of drinks required to reach
illegal BAC levels and that those with high BACs tend to
underestimate their BAC levels (e.g., Beirness et al., 1993).
It follows that BAC estimation tools (such as KYL cards)
could be useful in helping motorists regulate their drinking.
The use of these tools, however, has been controversial;
they might be misread and, consequently, mislead drinkers
into consuming more rather than less alcohol before driving. Persons who otherwise might think fewer drinks were
required to reach illegal BAC levels might be encouraged to
drink more (Johnson and Voas, 2004; Johnson et al., 2008).
Method
Data for this study were collected using the portal survey
method (Voas et al., 2006) at the San Ysidro border crossing between San Diego County, CA, and Tijuana, Mexico.
Our previous research used the portal survey method at the
same location to examine the drinking behaviors of young
U.S. residents who cross into Tijuana to drink (Lange et al.,
2006a) and to test experimentally several interventions designed to encourage the appropriate use of designated drivers
(Lange et al., 2006b).
The study examined four research aims: (a) to determine
the impact of warning drinkers about nearby police enforcement activity on alcohol consumption, (b) to determine the
impact on alcohol consumption of adding a KYL card to the
warning so that drinkers have a means of estimating their
BACs, (c) to determine whether the effectiveness of providing drinkers with KYL cards is greater when participants are
provided with instructions and asked to use the card during
the evening, and (d) to determine whether KYL cards allow
persons to better moderate their drinking under conditions
where they are motivated to maintain a relatively low BAC.
Recruitment
The portal survey method involved a pseudorandom
process of selecting and attempting to recruit naturally occurring peer groups as they approached the U.S.–Mexico
border (from the U.S. side). In this study, two teams of three
survey staff each worked at the border crossing. Whenever a
survey team was not actively interviewing participants, one
team member would approach the first individual to cross a
preselected mark on the sidewalk as he or she headed south
toward the border. This individual, along with each person in
his or her peer group, was invited to take part in a research
study on drinking and safety at the border.
Potential participants were offered $10 each for participating in the research, although in some conditions, participants
could earn up to $20. Entire groups were solicited because

our experience indicated that if group members were not
invited to participate, they were unlikely to wait around for
their friend. The recruited individual was therefore less likely
to participate in the research. Not all persons in a group,
however, were required to participate for some group members to take part in the study. Entire participant groups were
assigned to one of six experimental conditions.
Entry into Mexico
After agreeing to participate in the research, all participants were given a clipboard containing the survey instrument and a pen. The survey instrument was administered on
a custom-formatted optically scannable form that eliminated
the need to hand enter survey responses. The survey instrument contained questions related to demographics (e.g.,
sex, age, race and ethnicity, student status), recent drinking
history (e.g., on how many days in the past 4 weeks did they
drink, number of times consuming five or more drinks in
the past 2 weeks), transportation to the border (e.g., were
they a driver or a passenger), and their drinking plans for
the evening (e.g., did they plan to not drink, to get buzzed,
to get drunk, or to get very drunk, and how important was
it for them to reach their drinking goal). Young people who
visit Tijuana bars and clubs rarely drive into Mexico to park.
Rather, most park on the U.S. side of the border, where there
is ample parking, and walk a quarter mile to the main strip
of bars and clubs in Tijuana.
After participants completed the survey form, we gave
them hospital-style identification bracelets so we could identify them on return and so that their entry data and exit data
could later be linked. We asked participants to meet with the
survey team upon their return between 1 A.M. and 5 A.M. to
answer a few additional questions.
Finally, before crossing into Mexico, we requested an
anonymous breath sample from all participants. Breath
samples were collected and analyzed with calibrated Intoxilyzer SD400 preliminary breath-test units (CMI, Inc.,
Owensboro, KY) and used to produce estimates of blood
alcohol concentration (BrAC). These breath-test units are of
evidentiary quality and were programmed to record BrAC
readings internally but not to display the BrACs. We used
anonymous breath-testing procedures to help protect the
rights of participants (who might, for example, be drinking
although younger than age 21). The following day, the stored
BAC data were downloaded and merged with the survey
data.
Experimental instructions
Participants received instructions and KYL information
according to which of the six experimental conditions their
group was assigned. Figure 1 depicts the procedures that
distinguish the six conditions.

JOHNSON AND CLAPP

FIGURE 1.

81

Model of experimental procedures. BAC = blood alcohol concentration; KYL = “Know Your Limit.”

(1) Control condition. Participants in groups assigned to
the control condition did not receive any interventions but
were given the entry and exit surveys.
(2) BAC warning condition. Except for those assigned
to the control condition, all participants received information about the drunk-driving enforcement operations being
conducted in the San Diego area. The information cards
warned participants that they could be arrested if they drove
with BACs of .08% or higher and that consumption of any
amount of alcohol would increase their impairment and risk
of crash while driving. No additional information or KYL
cards were provided.
(3) KYL condition. Participants in three of the treatment
conditions, including the KYL condition (see Figure 1), were
given information cards (see above) with the KYL matrix
(gender specific) printed on the back of the card. These
participants received the KYL cards without explicit instructions or training on how to use them.
(4) KYL + instruction condition. Each participant received the KYL cards, as well as explicit instructions on how
to use them, and was instructed to use the card at least twice
during the evening to estimate his or her BAC.
(5) Motivation condition. These participants received the
warning about police activity but no KYL information (per
Condition 3). However, these participants were also told that
if they could limit their drinking while in Tijuana and return
to the United States with a BAC of less than .05% (to be
determined with a follow-up breath test), they would receive
an additional $10 ($20 total). The purpose of this additional

monetary incentive was to motivate drinkers to moderate
their drinking behavior.
(6) KYL + motivation condition. In the sixth experimental condition, participants were offered the additional $10
incentive to limit their drinking and return with a BAC of
less than .05%. These participants were also given a KYL
matrix as well as the generic warning. Thus, participants in
Condition 6 were given both a motive to control their drinking and a tool (the KYL matrix on the card) to help them
estimate their BACs. To the extent that KYL matrices are
beneficial for persons who are motivated to moderate their
BACs to achieve their goal, we expected lower BACs (or a
significantly higher proportion of BACs less than .05%) in
this experimental condition.
Exit procedure
Participants were instructed to locate the survey staff
stationed in the U.S. border crossing control facility between 1 A.M. and 5 A.M. Returning participants were given
an oral interview, and their responses were recorded by the
interviewer. The exit interview contained questions related
to demographics; drinking behavior (e.g., number of drinks
consumed and number of bars visited); transportation home
(e.g., whether the participant would be a driver); and for
Conditions 3, 4, and 6, whether they used the KYL card
while in Tijuana.
After completing the interview, participants were asked
to provide an exit BrAC sample. As in the entry survey,

82

JOURNAL OF STUDIES ON ALCOHOL AND DRUGS / JANUARY 2011

participants’ BrACs were not displayed but, rather, were
stored internally in the breath-test units. In Conditions 5 and
6, however, participants were informed that they could earn
an additional $10 if they returned with a BrAC less than
.05%. For these conditions, the survey staff used an SD400
breath-test unit that recorded the BrAC internally but also
displayed (yes or no) whether the BAC was .05% or higher.
After providing a breath sample, each participant was given
an incentive (a money order for either $10 or $20, depending
on condition).
Sampling strategy and random assignment
Portal survey data collection occurred on 13 weekends
(26 nights) between February 2005 and February 2006. In
all, 387 participant groups took part in the research. A total
of 1,125 participants completed both entry and exit portions
of the survey, although only 1,075 provided valid BrAC
readings at both entry and exit.
On a given night, only two of the six possible conditions
were administered (because of the complexity of conducting
experimental field research), and all members in a participant
group were assigned at random to the same experimental
condition. An attempt was made to balance the assignment
of condition pairs to data-collection nights. Each experimental condition was administered on seven to nine survey
nights.
Participant characteristics
Of the 1,215 participants, 51.8% were male, and the
median age was 20, with 79.2% being younger than age
21. The sample consisted of 27.2% White (non-Hispanic),
9.3% Asian, 14.5% Black, 40.8% Hispanic, and 8.0% other.
Slightly more than one quarter (29.2%) indicated that they
were drivers.
Research hypotheses
We analyzed the data to test four specific hypotheses
about the efficacy of KYL information as a tool for reducing drinking and driving. We hypothesized that providing
a warning about the risks of drinking and driving would
reduce BACs (particularly of drivers) relative to individuals
who did not receive an explicit warning (Hypothesis 1). We
hypothesized that providing KYL cards along with risk information should significantly reduce drinking (particularly
for drivers) more so than providing risk information alone
(Hypothesis 2). We hypothesized that providing instruction
to use the KYL matrix should increase its use and reduce
BACs (particularly for drivers) relative to those who receive
KYL cards but are not instructed or trained on how to use
them (Hypothesis 3). Finally, we hypothesized that providing drinkers (particularly drivers) with KYL cards (relative
to not having KYL cards) would motivate participants to

maintain relatively low BACs that would be reflected in their
having significantly lower BACs on return (Hypothesis 4).
Analytic approach
Our research design involved complex sampling whereby
individual participants served as our primary unit of analysis, but participant groups were our primary sampling unit.
All analyses were conducted using generalized linear mixed
modeling (PROC GLIMMIX in SAS Version 9.1 [SAS Institute Inc., Cary, NC]) with participant groups modeled as
a random variable to accommodate this design. Our previous analyses of portal survey data (e.g., Lange et al., 2006a,
2006b) included participant groups as a random-effect variable in analyses.
Each set of analyses included participant demographics
and drinking variables (gender, race, age, entry BrAC) as
statistical controls; our primary variables of interest were
driver status, an a priori contrast between two (or more)
experimental conditions, and the interaction between driver
status and the planned contrast. We did not conduct omnibus
tests of the six-level experimental condition variable. Exit
BrAC (controlling for entry BrAC) served as the primary
dependent measure.
Results
A preliminary random-effects analysis determined that
BrACs measured upon exit varied significantly as a function
of participant group (Z = 7.1, p < .01, intraclass correlation
coefficient [ICC] = .31), thereby necessitating the inclusion
of participant group as a random variable in subsequent
analyses.
Analyses of covariates
We conducted an initial series of analyses to predict exit
BrAC from entry BrAC, gender, race, age, and self-reported
driver status (returning home), and all two-way interactions
involving these variables. Participant group was modeled as
a random variable. Neither the two-way interactions nor age
significantly predicted exit BrAC. However, gender, race,
entry BAC, and driver status all predicted exit BrACs (all
ps < .01). In general, men consumed more alcohol than did
women (estimated mean exit BrACs of .047% and .040%,
respectively); drivers consumed less than nondrivers (.038%
vs. .050%); and Black participants consumed less than
Asian, White, and Hispanic participants (.028% vs. .050%,
.056%, and .045%, respectively, all contrast ps < .01). Entry
BrACs were positively associated with exit BrACs.
Omnibus analyses
Before conducting tests of the four specific research
hypotheses, in order to present a broader picture of the

JOHNSON AND CLAPP
TABLE 1. Model estimated breath alcohol concentration (BrAC) by condition and driver status
Variable
Control condition
Warning-only condition
KYL condition

Drivers
BrAC, %

Nondrivers
BrAC, %

.050
.029
.039

.044
.050
.050

Note: KYL = “Know Your Limit.”

study results, we conducted an analysis that combined the
six experimental conditions into three: a control condition
(Condition 1), a BAC warning condition (Conditions 2 and
5), and a KYL condition (Conditions 3, 4, and 6). Our statistical model included participant group as a random variable,
and participant demographics (along with entry BrAC) were
included as covariates. We modeled driver status, the threelevel experimental condition variable, and the Driver Status
× Condition interaction on returning BrACs. The interaction
was statistically significant, F(2, 701) = 5.62, p < .01. The
patterns of estimated means, presented in Table 1, were subsequently tested.
Subsequent analyses revealed no statistically significant
differences in returning BACs among the three experimental
conditions for nondrivers (all ps > .21). For drivers, however,
BrACs in the BAC warning condition were significantly
lower than for those in the control condition (p < .01), although only marginally lower than the KYL condition (p
= .09). There was no difference in the BrACs of drivers
and nondrivers in the control condition (p = .35); however,
BrACs for drivers were significantly lower than for nondrivers in both the BAC warning and the KYL conditions (both
ps < .01).
Hypothesis 1
The first specific hypothesis concerned the effect on exit
BrACs of providing participants with a generic warning
about the risks of drinking and driving. We tested the main
effect of driver status, the planned contrast between the
control condition and BAC warning conditions (Condition
1 vs. 2), and the Driver Status × Contrast interaction. Our
model included participant group as a random variable and
entry BrAC, gender, race, and driver status as covariates.
The contrast main effect was not statistically significant, but
the Driver Status × Contrast interaction was statistically significant, F(1, 695) = 11.0, p < .01. In the control condition,
the BrACs for drivers and nondrivers (.050% and .044%,
respectively) did not differ significantly (p = .36), but in the
warning condition, exit BrACs were significantly lower for
drivers than nondrivers (.032% vs. .059%, respectively, p
< .01). The BrACs of drivers significantly decreased in the
BAC warning condition, whereas the BrACs of nondrivers
significantly increased in the BAC warning condition.

83

Hypothesis 2
The second hypothesis concerned the effect on exit
BrACs of providing KYL information. The test of this hypothesis examined driver status, the contrast between the
BAC warning condition and the warning + KYL condition
(Condition 2 vs. 3), and the Driver Status × Contrast interaction. Participant group was included as a random variable,
and participant demographics and entry BrAC were included
as covariates. Neither the contrast main effect nor the Driver
Status × Contrast interaction (p = .08) was statistically significant. The trends of this marginal effect were not consistent with our predictions.
Hypothesis 3
The third hypothesis concerned the effect of providing
participants with instruction and training on how to use the
KYL information. This hypothesis involved multiple tests.
First, we wanted to see whether participants in the KYL
+ instruction condition used the KYL card more than participants in the KYL condition did (based on the self-report
measures at exit). The analysis examined driver status, the
contrast between Conditions 3 and 4, and the Driver Status
× Contrast interaction. The dependent measure was binary
(self-reported use vs. no use of the KYL card), and the analyses controlled for participant demographics, entry BrAC,
and participant group. This analysis revealed a statistically
significant Condition × Driver Status interaction, F(1, 236)
= 5.08, p < .01, whereby drivers who received instruction to
use the KYL matrix were significantly more likely to have
done so (estimated proportion = .80) than those who were
given KYL cards but did not receive instructions to use them
during the night (proportion = .45), p < .01. The difference
in proportion of nondrivers who used the KYL cards was not
statistically significant (.56 vs. .47, respectively), p = .24.
We then tested the effect of providing the KYL cards with
instruction to (a) the BAC warning condition (no KYL information) and (b) the KYL condition (with no instruction).
Thus, one series of analyses modeled the interaction between
driver status and contrast between Conditions 2 and 4, and a
second series modeled the interaction between driver status
and Conditions 3 and 4 (see Figure 1 for a diagram of experimental conditions). Neither the contrast main effect nor
the Driver Status × Contrast interaction approached statistical significance.
Not all participants reported using the KYL cards (even
those in the KYL + instruction condition), which might
explain why we failed to detect differences in exit BrACs
between the two experimental conditions. To more directly
test the effect of providing drinkers with KYL information,
we recoded participants in the KYL and KYL + instruction
conditions in two groups: “Did Use the KYL matrix” and
“Did Not Use the KYL matrix.” The contrast between these

84

JOURNAL OF STUDIES ON ALCOHOL AND DRUGS / JANUARY 2011

two groups, however, did not significantly interact with
driver status in predicting BACs (p = .26), nor was the main
effect of this contrast statistically significant (p = .36).
Hypothesis 4
The fourth hypothesis concerned whether KYL information could facilitate the drinkers’ ability to control their
alcohol consumption under conditions where they had been
motivated to do so. This required a series of analyses. First,
we tested the effect of our motivation manipulation (e.g.,
offering a $10 incentive to return with a low BAC) by contrasting mean BACs of participants in motivation condition
(Condition 5) to the BAC warning condition. Here, we found
BACs of nondrivers were significantly lower in the motivation condition than the BAC warning condition (.044% vs.
.059%, p < .05) but observed no significant differences in
BACs between the two groups of drivers (.032 vs. .027, p =
.58). Presumably, drivers already are motivated, on average,
to maintain a relatively low BAC and thus are less affected
by the motivational manipulation. These results suggest that,
at least for nondrivers, the additional $10 incentive served
as an adequate motivation for participants to regulate their
drinking.
To test the efficacy of providing KYL information to persons who are motivated to maintain low BACs, we contrasted
the BrACs in the motivation condition with those in the KYL
+ motivation condition (Condition 5 vs. 6). Relatively lower
BrACs in the KYL + motivation condition would suggest
that the KYL information helped motivate participants to
regulate their drinking. However, neither the interaction
contrast (p = .40) nor the main effect contrast (p = .38) was
statistically significant.
We then replicated the previous analysis (using the same
model) but this time predicting the likelihood (dichotomous)
of participants returning with BrACs of .05% or higher
(given that the incentive instruction was specifically linked to
that level). We limited the analyses to only those participants
(in the KYL + motivation condition) who reported actually
using the KYL card. Under incentive ($10) to maintain a
relatively low BrAC but without receiving KYL information, drivers were significantly less likely to return with high
BrACs (.05% or higher) than were nondrivers (proportions
= .18 vs. .41, respectively), F(1, 626) = 7.88, p < .01. No
significant difference was observed when individuals were
given an incentive plus KYL information (proportions =
.40 vs. .50, p = .33). Importantly, the proportion of drivers returning with relatively high BrACs in the motivation
condition was significantly lower than the proportion in the
motivation + KYL condition, F(1, 626) = 4.47, p < .05.
Discussion
This experimental study examined the ability of KYL information to impact drinking behavior among young adults.

In general, providing drinkers with KYL information—from
which they could estimate their BACs—did not play a role
in reducing drinking behavior. For nondrivers, pairing KYL
information with a general warning about drinking and
driving risks clearly failed to reduce BrACs more than did
the general warning alone, and marginal trends suggest that
the KYL information might have increased consumption.
This potential increase in drinking due to KYL information
deserves further investigation.
When we examined whether providing KYL information
helped drinkers regulate their consumption under conditions
where they were incentivized to drink moderately, a significantly larger proportion of drivers returned with BrACs
higher than .05% when they were given a generic warning
plus KYL information relative to when they were given a
generic warning alone. Although further research is necessary to corroborate these findings, our study found that under
these experimental conditions, providing drivers with KYL
information hindered their ability to regulate their alcohol
consumption.
These findings are contradictory to what we predicted
and appeared inconsistent with the goal of reducing drinking
by drivers. Why would giving drinkers BrAC information
contribute to higher drinking—particularly among drivers
who are motivated, presumably, to maintain ostensibly safe
drinking levels? Although research (e.g., Beirness et al.,
1993) suggests that drinkers are poor at estimating their BAC
levels, this research shows that drinkers still understand the
relative relationship between alcohol consumption and BAC.
Drinkers who plan to drive and who want to reduce their
risks know that simply drinking less can facilitate this goal.
Indeed, given a clear incentive to moderate their drinking, a
large proportion (estimated 81%) of drivers returned with
BrACs lower than .05% without being provided with any
tools for BAC estimation. For many drinkers, motivation
alone was sufficient to achieve the goal.
Previous research suggests that some BAC estimation
tools (e.g., saliva-based alcohol test strips) reduce drinkers’
ability to accurately assess their alcohol impairment (e.g.,
Johnson et al., 2008), perhaps by interfering with their attention to their own physiological cues. In this study, where
we actually examined the effect on drinking behavior, it is
plausible that drinkers in the KYL conditions used the KYL
cards to maximize their drinking up to some ostensibly
“safe” level. Drinkers who did not have KYL information,
rather than targeting their drinking to a specific BAC, simply
relied on a strategy of gross reductions in drinking. Potentially, then, providing drinkers with BAC estimation information enabled them to attempt to target their drinking toward
a specific BAC, yet the lack of precision of the estimates led
some to overshoot their target.
Why, then, did the KYL information affect the drinking
behavior of only drivers and not significantly influence the
drinking of nondrivers? The warning provided to participants

JOHNSON AND CLAPP
along with the KYL information pertained exclusively to
drinking-and-driving risk. Nothing increased the salience
of risks associated with alcohol consumption in general.
Without BAC-specific risks for nondrivers, there was little
incentive for them to use the KYL information.
One final noteworthy finding was the result that simply
warning individuals about the risks of drinking and driving
was sufficient to reduce BrACs of drivers. This warning,
however, also appeared to increase the BrACs of nondrivers.
Presumably, the warning motivated the participant groups to
clarify designated-driver roles, which may have “liberated”
nondriving drinkers to consume more. This finding replicates
results from a prior portal survey on drinking and driving
(Lange et al., 2006b).
Field research on two different approaches of BAC estimation (the current report and Johnson et al., 2008) suggests at least some iatrogenic effects associated with using
personal BAC estimation products as drinking-and-driving
countermeasures. It is unclear whether personal alcohol
tests are widely used, but intervention strategists should be
cautious about implementing programs that revolve around
these tools. The research also calls into question the utility
of teaching students how to associate symptoms to BACs.
Although there is no evidence that this information, typically
presented in didactic format, is harmful, the evidence suggests there is a greater likelihood of it increasing drinkingand-driving risk rather than reducing it.
At face value, the argument seems reasonable that providing drinkers with methods for estimating BACs—via education or personal testing tools—makes them more responsible
and facilitates their making more informed decisions about
driving. Yet this argument assumes that the drinkers are appropriately motivated to use this information to reduce their
risks. However, if potential drivers are strongly motivated to
avoid the risks associated with drinking and driving, BAC
estimation tools are not necessary. Our research shows that
motivated drinkers can and do reduce their alcohol consumption, even when they lack BAC information. It appears that
BAC estimation tools may be most useful for individuals
who wish not simply to minimize their risk but who try to
maintain their drinking while keeping BACs under some
prescribed level.
Although teaching drivers to estimate BAC levels prior to
driving may be ineffective at best, legal BAC limits for driving remain an effective public health and safety mechanism.
The research presented here suggests that brief field interventions that remind drivers of enforcement activities and
legal risks associated with driving under the influence can
be effective. These are complementary approaches. Further
research might address how such warnings can be coupled

85

with designated-driver interventions and larger environmental prevention campaigns in a comprehensive approach to
driving under the influence.
References
Beirness, D. J., Foss, R. D., & Voas, R. B. (1993). Drinking drivers’ estimates of their own blood alcohol concentration. Journal of Traffic
Medicine, 21 (2), 73-78.
Clapp, J. D., Reed, M. B., Holmes, M. R., Lange, J. E., & Voas, R. B.
(2006). Drunk in public, drunk in private: The relationship between
college students, drinking environments and alcohol consumption.
American Journal of Drug and Alcohol Abuse, 32, 275-285.
Clapp, J. D., Reed, M. B., Min, J. W., Shillington, A. M., Croff, J. M.,
Holmes, M. R., & Trim, R. S. (2009). Blood alcohol concentrations
among bar patrons: A multi-level study of drinking behavior. Drug and
Alcohol Dependence, 102, 41-48.
Clapp, J. D., Voas, R. B., & Lange, J. E. (2001). Cross-border college drinking. Journal of Safety Research, 32, 299-307.
Clapp, J. D., Whitney, M., & Shillington, A. M. (2002). The reliability of
environmental measures of the college alcohol environment. Journal of
Drug Education, 32, 287-301.
Johnson, M. B., & Voas, R. B. (2004). Potential risks of providing drinking drivers with BAC information. Traffic Injury Prevention, 5, 42-49.
Johnson, M. B., Voas, R. B., Kelley-Baker, T., & Furr-Holden, C. D. (2008).
The consequences of providing drinkers with blood alcohol concentration information on assessments of alcohol impairment and drunkdriving risk. Journal of Studies on Alcohol and Drugs, 69, 539-549.
Lange, J. E., Johnson, M. B., & Reed, M. B. (2006a). Drivers within natural
drinking groups: An exploration of role selection, motivation, and group
influence on driver sobriety. American Journal of Drug and Alcohol
Abuse, 32, 261-274.
Lange, J. E., Reed, M. B., Johnson, M. B., & Voas, R. B. (2006b). The efficacy of experimental interventions designed to reduce drinking among
designated drivers. Journal of Studies on Alcohol, 67, 261-268.
Lange, J. E., & Voas, R. B. (2001). Defining binge-drinking quantities
through resulting blood alcohol concentrations. Psychology of Addictive
Behaviors, 15, 310-316.
Larsen, J. D., & Kozar, B. (2005). Evaluation of a computer administered
alcohol education program for college student. Journal of Alcohol and
Drug Education, 49 (4), 69-83.
National Highway Traffic Safety Administration. (2008). Traffic Safety Facts
2007 Data: Alcohol-Impaired Driving (DOT HS 810 985). Washington,
DC: National Highway Traffic Safety Administration.
Royal, D. (2000). National survey of drinking and driving: Attitudes and
behavior: 1999, Findings, Vol. 1 (DOT HS 809 190). Washington, DC:
National Highway Traffic Safety Administration. Retrieved from http://
ntl.bts.gov/lib/26000/26000/26009/DOT-HS-809-190.pdf
Sharmer, L. (2001). Evaluation of alcohol education programs on attitude,
knowledge, and self-reported behavior of college students. Evaluation
and The Health Professions, 24, 336-357.
Voas, R. B., Furr-Holden, C. D. M., Lauer, E., Bright, C., Johnson, M. B.,
& Miller, B. (2006). Portal surveys of timeout drinking locations: A tool
for studying binge drinking and AOD use. Evaluation Review, 30, 44-65.
Zador, P. L., Krawchuk, S. A., & Voas, R. B. (2000). Alcohol-related relative
risk of driver fatalities and driver involvement in fatal crashes in relation
to driver age and gender: An update using 1996 data. Journal of Studies
on Alcohol, 61, 387-395.