HomeMy WebLinkAbout8 - Article: Relationship of alcohol outlet desity to heavy and frequent drinking-related problems a~~a,~~~-r, ~,,, ~l~~~~r~~{~
PERCAMOIY Health & Place 9(2003) 1-6
~'~ Y ~ ~~ ~ ; (~ I"-.~. ~':G `.,:::~"
H~..Tx
& Pr,~cE
www.elseviv rnmlloca Whealthplace
The relationship of alcohol autlet density to heavy and
frequent drinking and drinking-related problerns among
college students at eight universities
Elissa R. Weitzman*, Alison Folkman, Kerry Lemieux Folkman,
Henry Wechsler
Deparlment of Nealrh ~ Soc1a( Behaofar, Haraurd Schoaf ojPoblk Heal~h, Lwubnark Center, IO! Park Drme, P O Box 13678,
Bostar, MA 07215, USA
Absfract
To detcimme whether alcohol outlet den~ty was correlatcd with heavy and frequent drinldng and dnnlong-relatcd
problems, we compared ecologcal measures of outlet density with survey measures of drinl~ng using a geogxaphic
information system and the Harvard School of PubLc Health College Alcohol Study (n = 3,421, site n= 8). We
idenhfied 966 ouUets wiUrin 8 2-mile study azeas. Dencities/site ranged 1'rom 32 w 185, Density was correlated with
heavy drinhng (r = 0 82, p= 0 01~ 1'requent cltinktng (r = 0.73, p= 0 04) and drinking-mlated problems (r = 0.79,
p= 0 02). Women, undcrage students and studems who picked up binge drinking in college werc affected. Imphcarions
For prevention and reseazch are discussed. ~ 2002 Elsevier Scie~e Ltd. All rights resecved.
Keywords• Alwhol use, Dnnkmg problems, Aloohol outlet deostty, Preventlon; Eav¢oument; Co(kge
Tnhoductlon
Featurea of local alco6ol econoaues, in addihon to
chazacteristics of drinkers, may wfluence d:inkmg
Ixhavior among college studenta Diacount pricang of
akoholic drinks aad pmmohon of alcohoGc beveragea
have been ]inked to consumpdon amoug colkge
students (Chaloupka et al , 1998, WecLslet et al ,
2000a) Lower rates of binge drinldng e~dst among
students at schools whose administratocs report an
absencc of alcaho( outleta within a m~le of campus
(Wechsler et al, 1994). OuUet density may impact
dnnldng by maldng low cos; or volume discounted
akohol availabk to persons pmd~sposed to driuk heavily
(Gruenewald et al , 199~, for example young adults
High ouUet dens~ty may re8ect heavy dnnlong norms
and preferences (Scribner et al., 2000), or underl}wg
'Co~respondmg auWoe TeP + 1-6173848933, Caz. + 1-6fl-
3848730.
6-muU ad~ess ewmtzmaQhsph.hsrvard edu
(E R Wett~en)
community features, such as aocial disorganizatioa or
soclal capital linked W frequmt harvy dnnldng m
college (Wettzman and KawacLi,1A00).
The purpose of tltis sNdy was to. (a) pilot the
collection of secondaty data about local aloohol licenses
and assess thea availability and quality; aad,
(b) determme whether levels of heavy and frequent
drinldng and dnnking-mlatal probkms varied system-
aflcally with alcohol outlet density among students at
colleges participating in the "A Matter of Degree"
(AMOD) program to ieduce binge drinldng and related
harms.
Met6ods
Data collecaon jor geograpluc mjormat~on systems (GISJ
Outlet information was collected for venues wiUun a
2-m~le radws of a central location pomt (CLP) on or
near eight of ten AMOD campuses CLPs were
1353-8292/03/S-see &ont matter Q 2002 F,lsov~cr Siacnce LW Ali nghts reserved.
PII 51353-8292(02)00014-X
~R Wertzman e! oi / Health & Place 9(2003J !-b
identi&ed by local evaluators and corresponded to a
student union, administrative location or major inter-
sectron 1Le 2-mile diatance was chosen becausc it
encompasced major busmesses and student residences
(on- and off'~campus), One site was excladed because the
responsc rdte on the stude¢t behavioral survey was too
low (<50%), anotlter because data deacnbing onHet
density were unreliable.
Enumeration of Lcensad outlets wiUun study areas
was accomplished by matchmg lists of ]ocal Lcenses to
study areas addresses using ArcView 3.1 GIS software
(Environmental Systems Research Inshtutc, Inc,C
2000), as follows. We compiled a mastcr list of licensed
venues by s~ta &om government licensing hoazds,
secondary data sources and physical obaervahon for
the 1999-Z000 academic year. Lists mcluded: (1) name;
(2) address; (3) licensc type(s); (4) venue category (i.e.,
reafautant, bar, nightclub, package store/liquor
store/Ixer distributor, other); and, (~ license category
(i.e., whetLer Lcense supporta alcohol cons~ption
on- andJor off-premise). Outleu wem excladal that did
not typicaily serve college atudents; venues with a
combined on-site and catenng licensc were cbded as
`bn-site" basad on how tLey functioned for college
students Exclusione and recategori~ations were made
case by case.
Nex; ate CLP addtesses were entered into AtcView,
and a 2-mile cadial boundary was drawn, Atter
idenLfying Uu 2-mite study areas we geocoded addmss
and zip code Selds of each ouUet nsing ArcView's
automaud geocoding function, whrneby the aoftware
attempted to mntch each address element with its spawlt
street database. Addmsses for w6ich a 100% match was
found wen mapped without furtha insPection. Ad-
dresses for w}uch eithv no match or a pa~ai match was
found wero verifial using multiple rosources. Soumes of
error included• (1) misspelled street names, (2) incorrcct
street types, (3) incomct or misa~ng street dtrections,
(4) incorrect street numbers, and (~ i~ornxt ap codes.
Incomct elements were repaiied, and a modified subset
af addresses was aubvutted for a second round of
8~~~8•
The Srst two rounds of geocoding produced spatial
coordinates for 93-100% of the licensed ouUeta by s~te.
Remaming addresses were Lkely creaud subsequent to
the ArcView street database. In such cases, a proxy
geocode was Qenuated using local data. Su venues were
located manually using infocmxtion from paper maps
scnt by site evaluators. O~e plotted, wc visually
inspected maps and idenufied outkts wi'thm the study
azeaa 1'hese were counted and included in the analyses
(CAS) survey (mshtuhonal n= S for this study, student
n= 3421). Information about the CAS meihods and
measures ~s pubhshed elsewhere (R'echsler et al , 1994,
Wechsler et al , 1998; Wechsler et al., 7A006).
Student dnnking behaviors at the AMOD sites
included: Heavy drinking (~ercentage of drinkers who
reported coaeum~ng five or morc drinks at aa ofl:
campua party m the past 30 days), Frequenr drmking
(peroentage of drinkers who reported drinking on at
leas[ 10 occasions m the past 30 days); and, Drudcirtg-
related prohlems (percentage of drinkecs reporting five or
more problems associated with one's own alcohol
coasumption since the beginning of the school year).
Measures are cons~stent with othet large nationa]
surveya of youth drinking (Presley et al., 1996; Dougtas
et al., 1997; Johnaton et al., 1999).
Initinl analyses testad ranlc order correlationa between
outlet den~ty and drinlung among all s[udmt drinkers
Nact, we teatai rank order cornelations between onUet
dansity and d:inkmg measures among subgroups of
studmt drinkers. Because the elastiaty of demand for
akohol dilfers far college women and meu as do their
aocess patterns (Chaloupka and Wechsler, 199~, we
acamined gender ddferences in dfect among all student
drinkera. On all analyses, ties wene taken iato accomt by
Statistical Analysis SoRware (SAS) (SAS Inshtute, Tna,
Cc,1999-2000) Fmdinga arc roported for probabJtty
thPesltolds of p<0.05 with a two-taihxl test of sigaifi-
cance. We note all significant cormlauons and annotate
those with mulhple tiea
Eeaults
School settfng and sNdent chm~acteruhcs
Study aites were located in differeat ~eograpltic
regions of the United Staus and set in different types
of communities (i.e., small town, urban, suburban)
(Table 1). All of the universities wae public and all but
one had full-time undugraduate student enrollments
> 10,000.
Thera were 3421 survey reapondents among the eight
AMOD ~tes (avezage response rate was 62%, ranBmg
from 51% to 73%) From onahalf to two-thirds of the
student respondents at the sites wera female. A ma~onty
of students reported tLey were Wlute and betwcen 48%
repor[ed they werc younger than 21-64 yeazs of a~, tbe `~
legal age for purchasing and consummg akohol. From
10% to 21% of the respondents reportqd they were
membeis of fratermues and soronhes.
S~udent surcey dqta
We used behavioral survey data from We 1999
Harvard School of Public Health College tllcohol Stndy
Oudet charactertshcs
We idenhfied 2304 alcohol ouUets uamg master lists at
We eight AMOD sites, of which we were able to geocode
E R We~tzmnn et aL / Hen/th h Place 9(1003J J-6 3
Tabk I
Srto chawctensdcs and survey respondent soaodemograplucs' LV
`~_
S~[c
A B C D E F ~- C3 j H
Se(nng
Regioa Northeast South South North NorW North West South
Central Cmtral Centcal
Location Smalt Small Sub- Urban Smeli Urbaa Sub U~bea
town town ufian town urbsn
Sne&n[ character'utic+
Total N 391 728 348 388 412 462 382 310
Respoase rale (%) ~63) ~5~ ~62) ~bZ) ~6~ ~~3) (63) ~51~
°/. Femela 58 67 56 55 60 56 51 62
% py~~~ 94 88 73 90 89 89 83 83
% Undcage 63 64 54 48 55 54 58 63
% Caak-at6hatod 10 17 20 21 l7 12 l9 l6
Out&t cdara~terestlcs # (%J
Total deasity, 2 m~es 156 32 I85 117 85 15b 152 83
On-a~te venup, 0-1 mles 4l (2~ 17 (53) 0(0) 26 (22) 0(0) 0(0) b0 (39) 12 (14)
Off-site vmues, 0-1 mila 13 (8) 4(I3) 18 (10) 6(~ l4 (1~ 12 (S) 15 p0) 13 p~
On- & ofT-s~te venues, 0-1 mtles l(1) 1(3) 63 (34) 46 (39) 50 (59) 54 (3~ 4(3) 0(0)
On-e~fn venues, 1-2 mdce 52 (33) 7(22) 0(0) 7(~ 0(0) 0(0) 49 (32) 34 (41)
O(f-aRe venuev. 1-2 mdes 47 (30) 3(9) 42 (23) 20 (L'n 7(8) 15 (10) 21 (14) 24 (29)
On- R oH site venues, 1-2 miles 2 p) 0(0) 62 (34) 12 (l0) 14 (1~ 75 (48) 3(2) 0(0)
•perantages may not add W IUO due W roundmg error
~
*
~'
Q bdhm6dtele
Q alfaMe
Q meps
F~g 1 Study commumty vnth the lowest alcohol outlet density
ER Weltzman et aL l Hea!!h k Place 9(2003J !-h
and map 2217 (96%), rangmg from 93% to 100% across
sites. Site H bad the largest proporGOn of unmapped
outlets but, 6ecausc it drew on soune data descnbmg a
much lazger geograpluc uwt thaa the others, was not
cousidered to 6ave a d~sproportionate amount of
m~ssmg data. Almost half (e = 966, 44%) of the mapped
outlets wero located within two miles of the CLPa Of
these 470 fell within the first mde (i.e., a 1-mile radius
from t6c CLP), and 496 fell betwan one a~ two miles.
Total ouUet densties mthtn tLe 2-mik tadii of the
study s~tes ranged from 32 (S~te B) to 185 vmues (Site G~
with an average of 121(Table 1). Figs. l and 2 depict the
lowest- and tughestdeosity communitiea, respectively.
'ILere were fewer off-site venues than there were on-
site or both on-JolGaite venues in the study areas. '[he
proportion of off-site venues increased at greater
dishences fmm the CLPe. Closcr ia, 10% (n = 9~ of
the 966 outleta wetc Hcenaod for off-site conaumption
compaeed to abont one-Sfth (n =179, 19°/a) in the one-
to-two mile tadial ring.
celated problems (Table 2) From 27% to 41% of
studenu reported heavy dnnldng, consuuung five or
more drinks at an offcampus party m the past 30 days.
From apprwumately one-Gtth to one-third of studeats at
the sdes reported frequent drinking (coaguming akohol
on ten or more occas~ons during the past 30 days), and
large percentages oF students (1 ~32%) reported e~cperi-
encwg five or more problems resulting 1'rom their
drinking.
Hetween 20% and 46% af the respondentv repor[ed
frequent drunkennesa (.e., they drank enough to get
drunk thta or more timcv dunag the past 30 days). A
minority of respo¢dents reported that they dtank but
drd not do so frequently and heavAy. W~th few
excepaons, the majority of etudents reported that they
usually binge when they drink. When asked why they
drink alcohol, 44fi5% of the students across stes
mdicated "to get drunk" as an important mason
Slvdent drinkmg behautors
All sites had student populations tLat exh~bited lugh
levels of heavy md fiequent drinking a¢d drinhng-
Assceiationr hetween outlet density artd heauy drmkmg.
frequent druikirtg and drudcutg-related pro6lems
pretlet densrty and /ieavy drinkurg. Overall there was a
~gnificant correlation between outlet density and heavy
drinldng (i.e., consumed 5+ dnnks at an offcampus
Legend
~ ~
A bdhmAdleka
~ dleRa
Fig 2. Study commumty mtlt the highest alcodol ouUet dens~ty.
ER Weitzmm et a1 / Healfh & Place 9(1IN13) 1-6
Tablc 2
Dnnkmg charactenshcs by s~m, n(%)a
Srte
A
B
C
D
E
F
i. ~
rG~
H
Drlnking behaoWr ~
Heavy dnnling 138 (4l) 200 (3~ 99 (39) I03 (3~ 124 (3'n 147 (3~ 117 (3~ 53 (2~
Frtquent dnnluog 106 (3O 147 (2~ 74 (29) 63 (21) 82 (24) 129 (32) 9l (29) 40 (19)
Dnoking-related problems 116 (32) 13fi (22) 87 (31) 88 (2~ 102 (2~ 127 (30) 109 (32) 44 (18)
Freqaent drunlcenoess 142 (43) 226 (4l) 93 (3'n T3 (?.~ l3l (39) 180 (4~ t l6 (3~ 40 (2l1)
Non "Mnge" drinldog 90 (2i) 198 (2'n 1I l(33) 120 (]2) 116 (28) l13 (2~ I09 (29) 125 (42)
Usunlly bmgea wLw drwks 184 (54) 323 (5'n 138 (54) 158 (54) 187 (5~ 217 (54) I45 (4~ 78 (38)
Dndcs to get drunk 240 (6~ 390 (61) 167 (58) l63 (48) 221 (56) 262 (62) 225 (64) 112 (44)
Absteins 40 (~ 90 (12) 55 (l~ 46 (12) 28 ('n 36 (8) 29 (8) 52 (1'n
•Percentages may not add to I00 dm m mundmg e[ror.
party) fot all drinkers (r = 0.82, p= 0.01), with several
s~tes tied in rank Tlus finding was found to hold for
multipk subgroups of stadents, specifically for men
(r = 0.73, p= 0 04) and students who picked up bmge
drinking in college (r = 0 75, p= 0.03)
Ou[!et den.srty and frequent dru~king. OuUet dmsity
was cornelated with frequent drinl~ng (i e., drank on
10+ occasiona m past 30 days) for atl drinkers (r = 0.73,
p= 0.04) where there wero muNiple ties ia iank, no¢-
Greek atT'iliated students (r = 0.75, p= 0.03), women
(r = 0.72, P= 0.04), underage sludents (r = 0.79,
p= 0.02) wttich had mnl6pk hes, aod students who
picked up binge drinking in college (r = 0.84, p= 0.01).
Outlet densiry and dr6dcmg-relared prob7ems, Ftnally,
ouUet den~ty was correlated with problem drinking (.e ,
reporting 5+ probletns aince tho begtnn~ng of ffie school
year) among all driakecs (r = 0 79, p= 0.02), women
(r = 0.90, p= 0.002), underage students (r = 0.73,
p= 0.04), ovcrage studmts (r = 0.79, p= 0.02), and
studente who reporbed picking up binge drinking in
college (r = 0.76, p= 0 03).
Outlet denrtty and student deinagiap/uc charactensncs.
When raalced by provalence, student demographic
cheracteristics at the study sites were unrelated to the
rank ordering of outlet densuy.
Diacussbu
We found associahons betwcen ouUet density, heavy
and frequent drinkmg aud drinking-related problems
among all student drinkers and among several sub-
groups. These assoaahons are notable.lf outlet den~ty
wem a trivial Pactor we might not expect u to m8uence
less cowmitted and/or expenenced dnnke:s (i e, women
or students who report p~cking up bmge dnnkmg in
college). In fact, it appears tfiat tLe "wettest" commu-
nides may be particularly nsky for young people whose
drinking dces not reflect entrenched h~g6-nsk patterns.
Thorough invasflgation of these assoaaUons and
mechanisms nnderlying them arc naoded.
Severat kssons wen learned in Utis explo~uoa F~rst,
hcense categones vary considecably across state and
local bomdanes, challenging both researchers and
policyroakers. Development and adopuon of a atandaz-
dized licensing system may makc sense. We also fomd
considerable vanapon m the quality and currency of
Lcense infotmaflon fmm local Lcensing boarda. It was
helpful to supplement these data with data from
electromc and phyaical sources., including web site
yellow pages and buswess directories. A slulled local
evaluation slaff was inshvmental W both the creation of
a licensc typology that could be applied acroas s~tea, and
the collectioa of reGable local data.
Given the stnall sampk oC this sNdy it w~ll be
important to take a broadtr more compnheneive look
us~ng naUonal data. That larger look vrill address some
of tlus study's limitattons. We uaed an anaty4c tecLnique
appmpriate for nonparametne data and small sample
sizes. 'Ihis techniqu* did not allow us to control for
other variables. We nmited the chance that ow Sndings
we~e due to differmces in underlying student chazacter-
istics by usting whether ouUet densrty and student
sample characteristics were related and confounding the
observed relationsltips. They were not Future work
us~ng a national autvey sample will use mulhvanau
multilevel methods to accouut for iadiv~dual and
communuy cLaraccerisacs
Cross-sxtional data llke oura constrazn us from
making causal ini'erences about the relationslup between
outlet dens~ty and dnnhng. While we cannot determme
the chronological order of supply and demand patterns
at t6ese sites, it is unlikely that supply fully followed
demand. AMOD s~tea were selected based on their very
high levels of heavy episodic or bmge dnnking-levels
that had been in place for several years as Lave thetr
patterus of bar and atcohol oudet density. Finally, we
used as our outlet measure total density wuhin a
~R We~tzmatt e! aL l Heo!!h & Place 9(2003) !~f
bouuded geograpluc azea spec~fic to the AMOD coUege
rowns and students 77tis made sensc for our purposes
but &ndings cannot be generelized to other sethngs ar
populationa
Ackoowledgemeab
Tlus reseamh was supportal by grants from the
Robert Wood Jolwson Foundation We thank Toben
Nelsoa and Kathy McCsbe, Gaml Bormann, Ann
Dorwaldt, Brenda Egolf, Kathryn Farris, Julic Kearney,
Diane Leiva, R+ilLam Lugo, Steve Martin, Duane SLell
and Brenda Woods
Refereoces
Chaloupka, Fl, Wec6slv, H, 1996. Singe driddng in
collega tLe mpset of pnce, availaMhty, and atcohol control
pohwes. Contemporary Economtc Pohcy 14 (4), 112-12i
Chaloupka, FJ , Grossman, M, Saffer, H.,1998. T6e e6uas of
pna on tho consequwas of alcohol use and abuse. Recrnt
Developmrnts m Alcoholism 14, 33l 346.
~P~4 KA . CoUins, 7.L , Waaen, C., Kanq L, Gold, R,
Claytoq S, Ross, I.G , Kolbe, L 7, 1997 Rawlts fram We
1995 Nationai College Hr,alth Rislc Behav~or Survey
Jownat of Amencan College Health 46 (2~ 55-66.
Env~romnentat Systema Resaa[ch Inshtuty I~, p?A00
Arc~ew 31 [COmputa SoRwara] Redlnnds. CA.
Gruenewatd, P J, Millar, A B, Roepes, P., 1996. AccGVS to
alcohol. geograP6y and pmvendon for local co~umhes
Ncohol Health dc Research World 20 (4k 244-251
7ohnston, L D, O'Malley, P.M , Hachmaq I.G , 1999.
Natlonal Survey R~auita on Dmg Uae &om the Monitoring
lhe Future Study, 1975-1998. Vol. II: College Stndmts and
Yomg Adults NIIi Pubhcatioa pumber 99--0661 Nanonal
Insbtute on Drug Abuse. Betheda, MA
Presley, CA, Matmaq P W, Cashm, 7 R, Lyula, R, 1996
Alcohol and Dcugs on Armerican Collega Campuses: Uu,
Consequences, md Pemzpnons of tLe Campvs Envuon-
ment. Vol IV• 1992-94 Southem Llinois UmvecsLLy,
Carbondsiq II.
SAS Insatute, Ine . 0 1999-2000 The SAS Systan Vers~on 81
~Computer SoRwarel Cary, NC.
Scnbner, R.q . Coheq D.A., F~aha. W, 21100 Ev~dence of a
snuctueal dt'xt Cor akoho! outlU dmnty. a muldleeel
analyss. Alcoho6am: Clinical and Eapenmcntal Resean:h
21 (2), I88-195.
Wmhsla, H, DavenpoR, A., Dowdali. G, Moeyktos, B,
Casdllo, S, 1994 Health and bd~evioral coosequences
oY bmge dnnlong i¢ colkpe. A mtioaal aurvcy of studentt
at 140 campuses loumal of the Amencan Medical
Assoaation 272 (21~ 1672-1679.
Wxhster, H. Dowdall, G W, Maennet, 6., Gkdhdl-lioyt, 7,
Lee, H., 1998 Chaogd in bwge dnalong sud rolattd
problemv among Awa~can collega studmts betwan
1993 aad l997. Resdta of the Harvatd School of
Pub6c Health Co11eBe Alcohot Study. louroal oC Ama~mn
Cotlege Health 47 (2), 57-68.
Wahskr, H., Kuo, M, Ia, I-I, Dowdail, GW., ZODOa
Env~ronmentai cortelates of mdetage alcohol use md
relatal problems of college studen4a Amaicaa Soumal of
Prevm4ve Mediciee l9 (1), 2429
Wxhsler, H,1.ee. J.E , Kuo, M, I.ee, H, 20006 Colkge bin~o
dnntong in the 1990s a continuing problem Resulta of the
Iiarvacd School of Pubhc Health 1999 Colkge Alcohd
Study loumal of Ame~iian Co~ege Iiealth 48 (~,
199-210.
Wdtzman, E R, ICawactu, I., 21q0 Givmg means rearvmg• the
prokcuve etl'ea of soc~al caprtal on bmge dnddng on
coliege campuses. Amencan 7ournal oC Pubhc Health 90
(1~, L93l~1939