10
Drink Specials and the Intoxication Levels of Patrons Exiting College Bars Dennis L. Thombs, PhD, FAAHB; Virginia Dodd, PhD, MPH; Steven B. Pokorny, PhD; Morrow R. OmU, MAEd, CSAC; Ryan O'Mara, BS; Monica C. Webb, MPH; Diana M. Lacaci, BS; Chad Werch, PhD, FAAHB Objectives: To determine whether drink specials independently in- crease patrons' risk of achieving a high level of intoxication upon exit- ing drinking establishments. Meth- ods: In a campus community, data were collected from exiting pa- trons (N=291) via sidewalk inter- views and breath tests on 6 nights of 2 consecutive semesters. Re- sults: A multivariate model re- vealed that taking advantage of a drink special was associated with a fourfold increase in risk of achiev- ing a BAC > 80 mg/dl. Conclusions: These findings are the first to docu- ment that the drink discounting practices of college bars can be linked to higher intoxication levels among exiting patrons. Key words: college student alco- hol use, BAC, drink special, drink- ing establishments Am JHealth Behav. 2008;32(4):411-419 E xcessive alcohol consumption has been recognized as a serious threat to the academic mission of US col- leges and universities as well as to the health and safety of their surrounding communities.'"^ Heavy episodic drinking (defined as the consumption of 5 or more alcoholic beverages [4 for women] on at least one occasion in a 2-week period) Dennis L. Thombs, Associate Professor; Vir- ginia Dodd, Assistant Professor; Steven B. Pokorny, Assistant Professor; Morrow R. Omli, Research Associate, Florida Center for Health Promotion Research; Ryan O'Mara, Graduate Re- search Assistant; Monica C. Webb, Graduate Re- search Assistant; Diana M. Lacaci, Graduate Re- search Assistant, all from the Department of Health Education and Behavior, University of Florida, Gainesville, FL. Chad Werch, Director, Addictive & Health Behaviors Research Institute and Professor, Department of Health Education & Behavior, University of Florida, Jacksonville, FL. Address correspondence to Dr Thombs, Depart- ment of Health Education and Behavior, Univer- sity of Florida, Gainesville, FL 32611-8210. E- mail: [email protected]ß.edu has been associated with numerous prob- lems in the college student population such as sexual assault and other violent behavior, physical injury, property dam- age, high-risk sexual behavior, poor aca- demic performance, and even death.'•''•^ Despite these problems, on-premise alco- hol establishments (defined as places where patrons can obtain and consume alcoholic beverages such as bars, clubs, taverns, and restaurants) often beset col- lege students with alcohol beverage dis- counts and special promotions, typically referred to as "drink specials." For ex- ample, advertising in college newspapers touts promotions such as "Ladies drink free all night long" and "Ease your mind before final exams with our mind eraser drink special."^ Drink specials also in- clude "happy hour," "2-for-l," "shot with a beer," and similar promotions.*'^ Although several studies have documented the as- sociation between drink specials and ex- cessive alcohol consumption,*"^ method- ological limitations do not permit conclu- sions about the effects that drink specials have on patrons exiting alcohol establish- ments, including college student patrons Am J Health Behav.™ 2008;32(4):411-419 411

Drink Specials and the Intoxication Levels of Patrons Exiting College Bars

Embed Size (px)

Citation preview

Drink Specials and the Intoxication Levelsof Patrons Exiting College Bars

Dennis L. Thombs, PhD, FAAHB; Virginia Dodd, PhD, MPH; Steven B. Pokorny, PhD;Morrow R. OmU, MAEd, CSAC; Ryan O'Mara, BS; Monica C. Webb, MPH;Diana M. Lacaci, BS; Chad Werch, PhD, FAAHB

Objectives: To determine whetherdrink specials independently in-crease patrons' risk of achieving ahigh level of intoxication upon exit-ing drinking establishments. Meth-ods: In a campus community, datawere collected from exiting pa-trons (N=291) via sidewalk inter-views and breath tests on 6 nightsof 2 consecutive semesters. Re-sults: A multivariate model re-vealed that taking advantage of a

drink special was associated with afourfold increase in risk of achiev-ing a BAC > 80 mg/dl. Conclusions:These findings are the first to docu-ment that the drink discountingpractices of college bars can belinked to higher intoxication levelsamong exiting patrons.

Key words: college student alco-hol use, BAC, drink special, drink-ing establishments

Am JHealth Behav. 2008;32(4):411-419

Excessive alcohol consumption hasbeen recognized as a serious threatto the academic mission of US col-

leges and universities as well as to thehealth and safety of their surroundingcommunities.'"^ Heavy episodic drinking(defined as the consumption of 5 or morealcoholic beverages [4 for women] on atleast one occasion in a 2-week period)

Dennis L. Thombs, Associate Professor; Vir-ginia Dodd, Assistant Professor; Steven B.Pokorny, Assistant Professor; Morrow R. Omli,Research Associate, Florida Center for HealthPromotion Research; Ryan O'Mara, Graduate Re-search Assistant; Monica C. Webb, Graduate Re-search Assistant; Diana M. Lacaci, Graduate Re-search Assistant, all from the Department ofHealth Education and Behavior, University ofFlorida, Gainesville, FL. Chad Werch, Director,Addictive & Health Behaviors Research Instituteand Professor, Department of Health Education& Behavior, University of Florida, Jacksonville,FL.

Address correspondence to Dr Thombs, Depart-ment of Health Education and Behavior, Univer-sity of Florida, Gainesville, FL 32611-8210. E-mail: [email protected]ß.edu

has been associated with numerous prob-lems in the college student populationsuch as sexual assault and other violentbehavior, physical injury, property dam-age, high-risk sexual behavior, poor aca-demic performance, and even death.'•''•^Despite these problems, on-premise alco-hol establishments (defined as placeswhere patrons can obtain and consumealcoholic beverages such as bars, clubs,taverns, and restaurants) often beset col-lege students with alcohol beverage dis-counts and special promotions, typicallyreferred to as "drink specials." For ex-ample, advertising in college newspaperstouts promotions such as "Ladies drinkfree all night long" and "Ease your mindbefore final exams with our mind eraserdrink special."^ Drink specials also in-clude "happy hour," "2-for-l," "shot with abeer," and similar promotions.*'^ Althoughseveral studies have documented the as-sociation between drink specials and ex-cessive alcohol consumption,*"^ method-ological limitations do not permit conclu-sions about the effects that drink specialshave on patrons exiting alcohol establish-ments, including college student patrons

Am J Health Behav.™ 2008;32(4):411-419 411

Drink Specials

in campus communities.Previous studies have examined the

effect of price on alcohol consumption. Inthe general population, as the price ofalcohol decreases, rates of alcohol con-sumption and alcohol problems in-crease.''° This effect appears especiallypronounced among young adults.^•'°''^ Evi-dence that drink discounting or drinkspecials increase alcohol consumptionalso were supported in an early laboratorystudy by Babor et al,'^ which found thatboth casual and heavy drinkers consumedmore than twice as much alcohol duringsimulated happy hours as they did whensuch drink specials were not in opera-tion. Unfortunately, concerns about eco-logical validity suggest that the findingsmay not represent the behavior of collegestudents in more naturalistic drinkingsituations.

A few studies have investigated therole of drink specials on college studentdrinking behavior. For example, Kuo et aTcompared self-report data from the 2001College Alcohol Study (CAS) and directobservational assessments of alcoholprices and promotions in on-premise al-cohol establishments surrounding cam-puses involved in the CAS study. Theinvestigators found that nearly three quar-ters of on-premise establishments offeredspecials on weekends and almost one halfprovided at least one type of beer promo-tion. The analysis indicated that low saleprices and frequent promotions at on-premise establishments were associatedwith higher self-reported heavy episodicdrinking among the students of the cam-pus communities. Other studies haveassessed students' self-perceptions of al-cohol promotions and advertising and havefound that they are positively correlatedwith drinking intentions and drinkingbehavior.^''''^ A weakness of these andthe Kuo et al study is the sole reliance onself-report measures. The validity of self-report measures, particularly those as-sessing risk behavior such as substanceuse and abuse, may be limited.'^ In col-lege drinking studies, error in self-reportmay result from (1) an inability to accu-rately remember the frequency and quan-tity of drinking, (2) an impaired ability torecall the quantity of consumed drinksduring heavy drinking episodes, (3) a lackof proficiency in estimating a standardalcoholic drink, and (4) an intentionalmisrepresentation. In addition, self-re-

port surveys usually do not account forindividual differences in alcohol metabo-lism, body weight, food consumption, andduration of alcohol consumption, all ofwhich can affect intoxication levels. Thus,it is advisable to assess objective mea-sures, such as breath alcohol concentra-tions (BAC), whenever possible in researchon college student drinking.'^

The purpose of this pilot study was todetermine whether drink specials are in-dependently associated with an increasedrisk of patrons exiting on-premise alcoholestablishments in a highly intoxicatedstate. In this study, highly intoxicated wasdefmed as having a BAC greater than orequal to 80 mg/dl of blood, which is thepresumptive legal limit for driving underthe influence of alcohol in the UnitedStates.'^ We hypothesized that taking ad-vantage of a drink special would signifi-cantly increase the risk of being highlyintoxicated when exiting a drinking estab-lishment, after controlling for the effects ofpatron gender, legal drinking age status,money spent on alcohol and cover charges,hours spent drinking, number of consumeddrinks, and length of time between lastdrink and the breath test.

METHODSSiteThis IRB-approved field study was con-

ducted in a campus community in thesoutheastern United States during 3nights in December 2006 and 3 nights inMay 2007. These 2 data collection periodswere consecutive Wednesday, Thursday,and Friday nights of the last week ofclasses during fall and spring semesters.The community is the location of 2 insti-tutions of higher education. One school isa public research university enrollingapproximately 50,000 undergraduate andgraduate students. The other school is apublic community college where about16,000 undergraduates attend classes.The community has a large number of on-premise alcohol establishments that ca-ter to college students. Although state lawprohibits persons under the age of 21 frompurchasing and consuming alcohol, nostate or local ordinances forbid minorsfrom patronizing the alcohol establish-ments in the community that served asthe site of this study.

ProcedureIn this natural observation study, a

412

Thombs et al

representative sample of 15 on-premiseestablishments in the community wasselected based on 5 criteria: location inthe city, customer capacity, type of pa-trons, reliance on drink specials, and foodservice status. Each establishment as-sessment (or visit) consisted of an obser-vational assessment inside the estab-lishment (not discussed in this study) andpatron interviews outside the establish-ment. The time sequence of these as-sessments was randomized prior to col-lecting data each night. The sidewalkinterviews were conducted in an unob-trusive location outside the establish-ment, where the research team couldinterview exiting patrons. The establish-ments were surveyed twice (once between7:00 PM and 10:00 PM and once between11:00 PM. and 2:00 AM) on 3 consecutivenights (Wednesday, Thursday, and Fri-day) at the end of both fall semester andspring semester. Each assessment (orvisit to an establishment) took about 15 to30 minutes. The purpose for 2 nighttimeassessments was to sample changingconditions in the bar environment overthe course of a night, eg, varying drinkspecials and clientele. The owners andmanagers of the establishments did notparticipate in planning or conducting thestudy and, as intended, did not appearaware of the presence of the researchteam.

The sidewalk interview. Two or 3trained team members were located out-side the establishment, at least 50 feetaway from the entrance, to conduct thesidewalk interviews and collect breathsamples. Each of the sidewalk interview-ers had on a white T-shirt with largeblack letters reading "ALCOHOL RE-SEARCH." As patrons exited the drinkingestablishment a member of the researchteam would approach them and invitethem to participate in the study.

One of the first questions asked of theparticipants was "How long has it beensince your last drink?" A procedure devel-oped by Wright and Cameron'^ was used toreduce the possibility of obtaining artifi-cially high BAC measurements caused byresidual mouth alcohol. In this proce-dure, data collectors provided water tothose patrons who reported that their lastdrink was consumed within the previous15 minutes, so that they could rinse theirmouth. These 185 participants (or 63.6%)were instructed to expectorate (not swal-

low) the water. The patrons who reportedthat it had been 15 minutes or more sincetheir last drink were not given water.

After determining whether the patronneeded to use the water-rinse procedure,they participated in a 3-minute, 20-ques-tion interview. Examples of questions in-cluded the following:

When did you start drinking today?How many drinks have you had today?Did you take advantage of any drink

specials tonight at this establishment?Did you take advantage of any drink

specials at other bars tonight?How much money have you spent on just

cover charges tonight?Excluding cover charges, how much money

have you spent on alcohol tonight?

Upon completion of the interview, theparticipant was asked to blow into a hand-held, breath-testing device to determinetheir BAC. The Alco-Sensor IV (manufac-tured by Intoximeters, Inc) is approved bythe US Department of Transportation forevidential use. The device meets andexceeds the federal model specificationfor traffic enforcement and the OmnibusBreath Alcohol Testing requirements.

Once the breath sample was collected,the participant was given a "walk-away"card with information on transportationservices and local sources of help for analcohol problem, as well as the principalinvestigator's contact information and thatof the university's institutional reviewboard. Each respondent also was assignedan anonymous identifier, which was writ-ten on his or her card. The identifierallowed participants to anonymously callthe number on the card the following day toobtain their BAC from the previous evening.This next-day callback method was usedbecause there is reason to believe thatsome individuals may return to drinking ifthey appraise their BAC to be too low.̂ "

RESULTSDuring the course of the study, the

research team approached individualsand groups that included approximately600 patrons who had been in a drinkingestablishment. A convenience sample of291 of these patrons (61.4% male and86.2% college students) agreed to partici-pate in the study. Among the college stu-dent subsample, the largest proportionidentified themselves as seniors (45.2%),

Am J Health Behav.™ 2008;32(4):411-419 413

Drink Specials

Table 1Gender Differences in Taking Advantage of Drink Specials

Proportion Reporting "Yes"Men Women

Measures %(n) %(n)

Did you come to this establishment because of a drink special?" 23.6(42) 42.0(47)

Did you take advantage of any drink specials tonight at this establishment? 39.0(69) 47.3(53)

Did you take advantage of a drink special at another bar tonight? 33.3(59) 27.7(41)

Do you plan to look for more drink specials at other bars tonight? 31.6(56) 29.7(33)

Note.N = 291 patrons exiting drinking establishments between 7:00 PM and 2:00 AMa Gender difference was statistically significant (chi-square = 10.70, df = 1, P<.001)

followed by juniors (25.2%), graduate stu-dents (13.6%), sophomores (8.8%), andfreshmen (7.2%). The majority reportedtheir age to be 21 years of age or older(84.0%). The median interview time was12:30 AM (the es tabl ishments were re-quired by law to close at 2:00 AM). Therewere no statistically significant semes-ter differences (P > .05) on any of thealcohol measures. Furthermore, night ofweek was not significantly associatedwith BAC, seeking a drink special, ortaking advantage of a drink special.

Effectiveness of the Water-rinseProcedureData analyses produced evidence sup-

porting the use of the water-rinse proce-dure (described in Methods). The bivari-ate correlation between number of min-utes from last drink to breath test andBAC was not statistically significEint (r = -.03, P = .59, N = 278). In addition, numberof minutes from last drink to breath testwas entered in a logistic regression modeland was found to not be a significantpredictor of BAC when divided into 2 lev-els (Table 2). These findings are consis-tent with previous research^''^^ that hasfound that the influence of residual mouthalcohol on measured BAC decays rapidlyafter just 4-6 minutes with or without awater rinse.

Gender DifferencesThe median time male patrons re-

ported that they began to drink that day

was 8:30 PM compared to 9:30 PM for femalepatrons. On average, males reported theyhad been drinking that day for 4.7 hours(sd = 3.2), compared to 3.5 hours (sd = 2.3)for females. Based on self-report, the meannumber of drinks consumed that day was8.8 (sd = 2.3) for men and 5.3 (sd = 3.3) forwomen. A somewhat larger proportion ofmen (73.9%) drank before coming to theassessment establishment than women(62.5%). The mean number of drinks con-sumed at the assessment establishmentwas reported by men to be 4.2 (sd = 3.9)versus 2.9 (sd = 2.7) for women. Mostpatrons were quite intoxicated upon exit-ing the establishment; eg, 58.9% of themen and 56.3% of the women had BACs >.80 mg/dl. The mean BAC was 97 mg/dl(sd = 57) for men and 83 mg/dl (sd = 49) forwomen, with a range of 0 to 281 mg/dl formales and 0 to 200 mg/dl for females.

The male patrons reported spending anaverage of $3.28 (sd = 6.18) on establish-ment cover charges, whereas female pa-trons spent an average of $2.92 (sd = 4.52).On alcohol expenditures, the gender dif-ference was much more pronounced: menspent an average of $15.80 (sd = 18.91),compared to $8.12 (sd = 12.58) amongwomen. However, in both sexes, the me-dian amounts spent on alcohol were con-siderably less than the mean values($10.00 in men and $3.00 in women) dueto a handful of participants who reportedspending very large sums of money ondrinks that night. The mode was $0.00 forboth men and women, with 27.8% of the

414

Thombs et al

men and 45.5% of the women claimingthey spent no money at all on alcohol. Ofthose patrons who reported drinking atthe establishment they just exited, 24.2%of the men claimed they spent no moneyon alcohol, compared to 37.8% of females.The association between spending (covercharges plus alcohol costs) and intoxica-tion was slightly weaker in men (r = .28,P< .001) than in women (r= .44, F< .001).In these correlation analyses, covercharges and the amount of money spenton alcohol were summed together be-cause at many establishments covercharges essentially represented the ex-pense of drinking during certain timeperiods (eg, "Ladies' night" specials). Table1 summarizes gender differences in tak-ing advantage of drink specials.

Service to Underage PatronsThe sidewalk interviews were used to

estimate the extent to which alcohol bev-erage service was being provided to un-derage patrons at the 15 establishments.A majority of the exiting patrons reportedthat they believed persons under the ageof 21 were drinking in the bar that night(65.9% ofthe males at 13 different estab-lishments and 72.3% ofthe females at 12different establishments). Of those pa-trons who reported drinking at the estab-lishment they just exited (n = 235), 12.3%acknowledged that they were youngerthan 21 years of age. In this underagegroup of patrons (n = 29), 82.8% had BACs> 80 mg/dl. There was no statisticallysignificant difference (P > .05) betweenunderage and legal-age patrons on thetotal amount of money spent on alcoholand cover charges that night. Thoughsubjective and based on self-report, thesefindings strongly suggest that patronsunder the age of 21 were consumingalcohol in many of the drinking establish-ments.

Logistic Regression Analyses of 2Intoxication Levels Among PatronsA logistic regression analysis, testing

7 predictor variables, was conducted todistinguish the highly intoxicated patronsfrom those with BACs below 80 mg/dl. Inaddition to "took advantage of a drinkspecial that night," 6 other variables wereentered in the logistic model to accountfor alternative explanations of BAC levelat the time of leaving an establishment.Among the 267 patrons who exited on-

premise alcohol establishments and pro-vided complete data, 105 (or 39.3%) hadBAC readings less than the 80 mg/dlcutoff. Peduzzi et al̂ ^ recommend in thesmaller of the 2 groups in logistic regres-sion analysis, there be at least 10 casesper variable to reduce the possibility of aninvalid model. Thus, the case-to-variableratio of 15 in the logistic model (105/7)was substantially greater than thatneeded to establish statistical conclusionvalidity.

As shown in Table 2, the predictor setaccounted for a significant amount ofvariance in intoxication level defmed bythe 80 mg/dl cutting score (model chi-square = 79.44, df = 7, P < .001). TheNagelkerke pseudo-R^ statistic indicatedthat an estimated 34.9% of the variancein intoxication level could be explained bythe predictor set. The variable that bestdistinguished the highly intoxicated pa-trons (BACs > 80 mg/dl) from those withlower BACs was "took advantage of a drinkspecial that night" (OR = 4.38), followed bylegal drinking age status (OR = 2.92),hours spent drinking that day (OR = 1.15),number of drinks consumed that day (OR= 1.13), and total amount of money spenton cover charges ajid alcohol (OR = 1.03).Gender and number of minutes betweenlast drink and breath test had odds ratiosnot significantly different from 1.0. Thus,after accounting for the effects of theother variables in the multivariate model,taking advantage of a drink special pro-duced a fourfold increase in risk of achiev-ing a high BAC (> 80 mg/dl), and beingunder the age of 21 increased the samerisk almost 3 times. Though statisticallysignificant, the effects of hours spentdrinking that day, number of drinks con-sumed that day, and total expenditureswere relatively modest.

An additional logistic regression analy-sis, using the same predictor set, wasconducted in which the intoxication cut-ting score were changed to 100 mg/dl.Taking advantage of a drink special re-mained the predictor that most clearlyincreased the odds of achieving the higherintoxication level with a multivariate oddsratio of 4.25 (95% CI = 2.28 - 7.93) followedby legal drinking age status [OR = 1.73,(95% CI = 0.78 - 3.86)], suggesting thatdrink specials may contribute to boostingpatrons' intoxication to even higher lev-els as they exit establishments (ie, than80 mg/dl).

Am J Health Behav.™ 2008;32(4):411-419 415

Drink Specials

Table 2Logistic Regression Analysis: Association of

Characteristics With BAC >80mg/dl

Variables in the Model

Took advantage of a drink special that night

Legal drinking age status

Hours spent drinking that day

Self-reported number of drinks consumed that day

Money spent on cover charges and alcohol that night

Gender

Number of minutes between last drink and breath test

OR(95%CI) Wald

4.38(2.39-8.01) 22.94

2.92(1.18-7.20) 5.39

1.15(1.01-1.31) 4.40

1.13(1.03-1.24) 7.21

1.03(1.01-1.06) 8.47

1.79(0.94-3.41) 3.12

0.99(0.98-1.00) 2.04

P

.0001

.02 I

.04

.007

.004

ns

ns

Note.Model chi-square = 79.44, df=7, P<.001. Total N=267 patrons exitingcampus community on a Wednesday, Thursday, and Friday night. Amissing values resulted in 24 cases being excluded from the analysi

<80mg/dl(n=10S)

Yes=29.3%

Jnder 21=31.8%

m=3.13(sd=2.69)

m = 5.07(sd=4.66)

m=$ 10.30(sd=$ 13.28)

--

--

Patron

BAC

vs.

vs.

vs.

vs.

vs.

>80mg/dl(n=162)

Yes=70.7%

Under 21=68.2%

m =5.00(sd=2.94)

m=9.15(sd=5.35)

m=$ 19.75(sd=$20.60)

--

--

drinking establishments in alistwise deletion of cases with

s.

DISCUSSIONThe current investigation makes a sig-

nificant contribution to the research onalcohol use in campus communities. First,it is among a handful of studies that havecollected field data from patrons exiting on-premise alcohol establishments and amongvery few that have done so in a campuscommunity with a large number of estab-lishments that cater to college students.Second, the data were not limited to self-report measures of alcohol consumptionand included the collection of breathsamples. Thus, the findings presented heremay have greater ecological validity thanthose obtained from studies that rely ex-clusively on self-report surveys of sobercollege students conducted in nondrinkingsettings^''*'^ or from protocols devised forlaboratory settings.'^

The primary purpose of the study was toexamine the relationship between drinkspecials and alcohol intoxication at thetime of exiting a drinking establishment.We hypothesized that patrons who tookadvantage of drink specials would be morelikely to have BACs greater than or equalto 80 mg/dl after accounting for relevantpatron characteristics. In support of this

hypothesis, the findings indicated thatpatrons exiting alcohol establishmentswho had taken advantage of a drink spe-cial were 4.38 times more likely to havea BAC >̂ 80 mg/dl than were those who didnot take advantage of a drink special.This is the first investigation to show thatdrink specials can be linked to elevatedlevels of intoxication in individuals exit-ing on-premise establishments. These"micro-level" results are consistent withprevious economic research on the ef-fects of price on alcohol consumption inlarge populations.'° Although the data werecollected in just one campus community,these results suggest that drink specialsshould be a priority focus of policy develop-ment and regulation aimed at improvingthe serving practices of on-premise alco-hol establishments; deterring alcohol-impaired driving; and generally reducingall sources of harm associated with drink-ing at bars, taverns, clubs, and restau-rants.

Consistent with the voluminous lit-erature on college drinking,'•^'' we foundsubstantial gender differences in self-reported quantity of alcohol consumption(in this case after exiting an establish-

416

Thombs et al

ment). At the same time, however, thegender difference in intoxication wasnegligible (men mean BAC = 97 mg/dl vs83 mg/dl in women). These findings areconsistent with previous field researchassessing the drinking behavior of youngadults^ '̂̂ ^ and provide additional evidencethat the sexes achieve similar levels ofintoxication even though college mengenerally consume more alcoholic drinksthan do college women. This growing bodyof field research questions the prioritiesidentified in Healthy People 2010 Objec-tive 26-11, which relies on drink countmeasures (5 or more) rather than intoxi-cation levels and calls for particular at-tention directed to the problem of bingedrinking in men.^

The results of this study suggest thatpatrons under the age of 21 not onlyconsume alcohol, but also have higherlevels of intoxication when exiting theestablishment than those of legal-agepatrons. These findings highlight the needfor tighter regulation of the minimum barentry age. Such regulations probably needto be implemented on a statewide basis toavoid creating conditions that promptunderage youth to travel to municipali-ties with lower entry-age requirements.^^

Strengths and Limitations of theMethodsThe strengths of this investigation in-

clude the direct observational assess-ment of establishment serving practicesand, most important, the collection ofbreath samples in a naturalistic fieldsetting. To date, the vast majority of col-lege drinking studies have relied solelyon self-report measures in nondrinkingsettings to assess perceptions, attitudes,and behavior. The findings reported here,as well as those from previous field inves-tigations that have collected breathsamples,'̂ '̂ '̂̂ ^ point out the weakness ofusing retrospective self-reports of alcoholconsumption quantity in college drinkingresearch. One early field study found thatBAC and self-reported drink quantity wereonly moderately correlated (r = .37).̂ ^ Like-wise, in this study, it was found that self-reported number of drinks had a rela-tively modest ability to distinguish thosepatrons with BACs > 80 mg/dl from thosebelow this threshold.

The most notable limitation of thisfield study involves questions aboutsample representation. These questions

frequently arise in the conduct of fieldresearch.^° The investigation sampledpatrons exiting on-premise alcohol estab-lishments in a single campus commu-nity on 6 nights. Moreover, the environ-ment in which the data were collected isnot an orderly one. At night, many patronscould be found entering and exiting estab-lishments on an almost continuous ba-sis. Outside some establishments, stu-dents congregated to socialize. At othersites, they were waiting to enter an es-tablishment. At closing time, large num-bers of patrons exited rapidly and usuallyin groups. These conditions precludedsystematic random sampling of individu-als. Thus, we sampled establishmentsusing a randomized schedule. Althoughthere is some uncertainty about thesample's ability to represent patron char-acteristics that exist in the broad range ofcampus communities, we believe the ob-tained sample represents the populationof patrons in the study community.

There are other questions that can beraised about the validity of the data collec-tion method used in this investigation.For example, it is possible that the mosthighly intoxicated exiting patrons werenot adequately represented in the sample.However, it was the observation of theresearch team that the patrons most likelyto decline to be interviewed were thosewho indicated that they had not con-sumed any alcohol that night (ie, therequest for participation was perceived tobe irrelevant). Another validity concernmight be the relatively high rate ofnonparticipation among exiting patrons.Indeed, most did decline to be interviewedas they were leaving the establishment.However, it was not possible to calculatean actual refusal rate because of thespeed of the interview activity, the clus-tering and movement of patron groups,and multiple exiting doors (a number ofthe establishments had front and backdoors). Also, some measures such as "tak-ing advantage of a drink special" and"age" were based solely on self-report.Finally, there is a small possibility that apatron could have been interviewed onmore than one occasion. However, thiswas not likely due to the large number ofexiting patrons and the randomized sched-ule of visiting establishments.

Suggestions for Future ResearchFuture field research on college drink-

Am J Health Behav.™ 2008;32(4):411-419 417

Drink Specials

ing should consider exploring the follow-ing possibilities. First, improved night-time protocols are needed for enhancingthe recruitment and participation of pa-trons as they exit on-premise alcoholestablishments. Second, it is importantto note that the observed BACs representa level of intoxication at a single point intime; ie, exiting a drinking establish-ment. They do not provide informationabout peak BAC for the night. Futurecollege drinking studies should seek toincorporate breath testing into assess-ment batteries where possible. Consider-ation should be given to designs thatmight plot students' changing intoxica-tion levels over the course of a night innaturalistic settings.

Third, partnerships should be estab-lished with undergraduates and under-graduate organizations to engage them inthe research process, including develop-ing field research questions, collectingand analyzing data, and implementingchange strategies to reduce alcohol-re-lated harm in campus communities.^^ Littleis known about the extent to which stu-dents can be engaged in participatory fieldresearch. Even greater uncertainty sur-rounds the prospect of engaging alcoholretailers in the research process. Fourth,there is a need to develop measures thatdistinguish between types of drink spe-cials to determine whether specific dis-counting practices (eg, "2 for 1 happy hour,""Ladies drink free") are more likely thanothers to boost patron intoxication levels.Such an assessment could generate infor-mation that would be useful to the develop-ment of sound alcohol control policy.

AcknowledgmentsThe authors wish to acknowledge the

following students and staff for their con-tributions to this research project: AshleyArthur, Katie Bello, Oksana Chelnokov,Anthony Delisle, Katie Dendy, KristenDevick, Feliz Naz Erenguc, MeganlynnFlores, Tavis Glassman, MeghanGreenfield, Devon Grimme, Sara Gullet,Will Hild, Tommy Huang, Sara Jameson,Amber Janson, Stephanie Knapp, BrockMiller, Megen Miller, Steven Omli, Jes-sica Osborne, Justin Page, Valerie Prince,Sonya Rane, Coley Sheriff, BryanSteinman, and Alvin Wong. •

REFERENCES1.Wechsler H, Lee JE, Kuo M, et al. Trends in

college binge drinking during a period ofincreased prevention efforts. Findings from 4Harvard School of Public Health College Alco-hol Study Surveys: 1993-2001. JAm Coll Health.2002:50:203-222.

2.US Department of Health and Human Ser-vices. Healthy People 2010: Understandingand Improving health and Objectives for Im-proving Health. (Vol. 2. 2nd ed). WashingtonDC: U.S. Government Printing Office, 2000.Available at: http://www.healthypeople.gov.Accessed: March 6, 2007.

3.Centers for Disease Control and Prevention.Youth risk behavior surveillance: national col-lege health risk behavior survey—United States.MMWR CDC Surueill Summ. 1997;46:l-54.

4.Hingson RW, Heeren T, Zakocs RC,et al.Magnitude of alcohol-related mortality andmorbidity among U.S. college students ages18-24. J Stud Alcohol. 2002;63:136-144.

5.Perkins HW. Surveying the damage: a reviewof research on consequences of alcohol mis-use in college populations. J Stud Alcohol.2002;14(Suppl 14):91-100.

O.Christie J, Fisher D, Kozup J, et al. The effectsof bar-sponsored alcohol beverage promotionsacross binge and nonbinge drinkers. Journalof Public Policy and Marketing. 2001;20(2):240-253.

7.Kuo M, Wechsler H, Greenberg P, et al. Themarketing of alcohol to college students: therole of low prices and special promotions. AmJ Prev Med. 2003:25(3):204-210.

8.Erenberg DF, Hacker GA. Last Call for High-risk Bar Promotions that Target College Stu-dents: A Community Action Guide. Washing-ton, D.C.: Center for Science in the PublicInterest: 1997.

9.Aaron H, Akerlof G, Baumöl WJ, et al. Econo-mists' declaration on federal alcohol excisetaxes (online). Coalition for the Prevention ofAlcohol Problems. May 2005. Available at:http://www.cspinet.org/new/pdf/petition-alcohol.pdf. Accessed February 24, 2007.

lO.Chaloupka FJ, Grossman M, Saffer H. Theeffects of price on alcohol consumption andalcohol-related problems. Alcohol Res Health.2002;26(l):22-34.

ll.Chaloupka FJ, Wechsler H. Binge drinkingin college: the impact of price, availability andalcohol control policies. Contemp Econ Policy.1996;14:112-124.

12.O sterberg E. Do alcohol prices affect con-sumption and related problems? In HolderHD, Edwards G, eds. Alcohol and PublicPolicy: Evidence and Issues. New York: Ox-ford University Press, 1995:145-163.

13.Babor TF, Mendelson JH, Greenberg I, et al.Experimental analysis of the "happy hour:"effects of purchase price on alcohol consump-tion. Psychopharmocology. 1978;58:35-41.

14.Pedersen, PJ. The influence of alcohol adver-tising on students' drinking behaviors. NASPAJournal 2002;40(l):25-38.

15.Clapp JD, McDonnell AL. The relationship of

418

Thombs et al

perceptions of alcohol promotion and peerdrinking norms to alcohol problems reportedby college students. Journal of College StudentDevelopment. 2000;41(l):19-26.

16.Harrell AV. The validity of self-reported druguse data: the accuracy of responses on confi-dential self-administered answer sheets. InHarrison L, Hughes A, eds. The Validity ofSelf-Reported Drug Use: Improving the Accu-racy of Survey Estimates, NIDA ResearchMonograph No. 167. NIH Publication No. 97-4174. Rockville, MD: Department of Healthand Human Services; 1997:37-58.

17.Thombs DL, Olds RS, Snyder B. Field assess-ment of BAC data to study late-night collegedrinking. J Stud Alcohol. 2003;64:322-330.

18.Insurance Institute for Highway Safety. DUI/DWI Laws as of March 2007. Highway LossData Institute (online). Arlington, VA. Avail-able at: http://www.iihs.org. Accessed May10, 2007.

19.Wright NR, Cameron D. The influence ofhabitual alcohol intake on breath-alcohol con-centrations following prolonged drinking.Alcohol Alcohol. 1998;33:495-501.

20.RUSS NW, Geller ES, Leland LS. Blood-alco-hol level feedback: a fedlure to deter impaireddriving. Psychol Addict Behav. 1988;2:124-130.

21.Dubowsky KM. Studies in breath-alcohol analy-sis: biological factors. Int J Legal Med.1975;76:93-117.

22.Slemeyer A. Detection of Residual MouthAlcohol Uusing Eelectrochemical Sensors.Paper presented at the 15* meeting of theInternational Council on Alcohol, Drugs, andTraffic Safety. Stockholm, Sweden, May 21-26, 2000. Available at: http://wvirw.icadts.org/proceedings. Accessed May 16, 2007.

23.Peduzzi P, Concato J, Kemper E, et al. Asimulation study of the number of events pervariable in logistic regression analysis. J ClinEpidemiol. 1996;49:1373-1379.

24.0'Malley PM, Johnston LD. Epidemiology ofEilcohol and other drug use among Americancollege students. J Stud Alcohol. 2002;(Suppl14):23-39.

25.Lange JE, Voas RB. Defining binge drinkingquantities through resulting blood alcoholconcentrations. Psychol Addict Behav.2001;15:310-316.

26.Olds RS, Thombs DL. Blood alcohol concen-trations of college women and men returninghome at night. Alcohol Research. 2003;8(6):271-275.

27.Illinois Higher Education Center for Alcohol,Other Drug and Violence Prevention. Mono-graph: 21 Minimum Bar Entry Age, 2005(online). Available at: h t tp : / /w WW. illinoishec.org/frame s/page s/docu-m e n t s / M o n o g r a p h 2 l M i n i m u m B a rEntryAge.doc. Accessed March 6, 2007.

28.Carey KB, Hustad JT. Are retrospectivelyreconstructed blood alcohol concentrationsaccurate? Preliminary results from a fieldstudy. J Stud Alcohol. 2002;63:762-766.

29.Werch CE, Bakema D, Ball M, et al. Catalogingblood alcohol level and alcohol consumptiondata in field settings: feasibility and findings.J Stud Alcohol. 1988;49:561-566.

30.Dowdall GW, Wechsler H. Studying collegealcohol use: widening the lens, sharpeningthe focus. J Stud Alcohol. 2002;(Suppl 14):14-22.

31.Osborn CJ, Thombs DL, Olds RS.Reconceptualizing research on undergradu-ate alcohol use: the need for student engage-ment. Eval Health Prof. 2007;30:118-137.

Am J Health Behav.™ 2008;32(4):411-419 419