Aggression involving alcohol: relationship to drinking patterns and social context

RESEARCH REPORT

© 2003 Society for the Study of Addiction to Alcohol and Other Drugs

Addiction,

98,

33–42

ABSTRACT

Aims

The present study examines the relationships between: (1) alcoholinvolvement/perceived intoxication level of participants and aggression sever-ity; (2) respondent drinking patterns and involvement in alcohol-related aggres-sion; and (3) social context and alcohol-related aggression.

Design

Random digit dialing (RDD) with computer assisted telephone inter-viewing (CATI) was used to obtain a random sample of Ontario adults aged 18–60 (response rate of 67%).

Participants

Respondents who reported that they had been involved personallyin physical aggression in the past 12 months were the focus of the present study.

Measurements

Questions were asked regarding the most recent incident ofphysical aggression, including whether the respondent and opponent drankalcohol prior to aggression, perceived intoxication levels at the time, number ofparticipants, relationship to opponent, social context of aggression, time of dayand day of week. Three items were used to assess aggression severity: injury torespondent, use of threats by respondent or opponent and police involvement.

Findings

(1) Injury to respondent and threats by respondent were not associ-ated with alcohol involvement

per se

, but were significantly related to perceivedlevel of alcohol intoxication; (2) drinking pattern of respondent was signifi-cantly associated with alcohol-related aggression but unrelated to aggressionthat did not involve alcohol; and (3) a number of contextual factors (e.g. gender,number of participants, time of day) were found to be associated with alcoholinvolvement in aggression.

Conclusions

The results suggest that both drinking pattern and contextual fac-tors are important in distinguishing between alcohol-related aggression andnon-alcohol-related aggression. As well, alcohol intoxication may be an impor-tant predictor of aggression severity.

KEYWORDS

Aggression severity, alcohol-related aggression, drinking

pattern, perceived intoxication level, social context.

Blackwell Science, Ltd

Oxford, UK

ADDAddiction

0965-2140© 2002 Society for the Study of Addiction to Alcohol and Other Drugs

97Original Article

Samantha Wells & Kathryn Graham Aggression, drinking pat-

terns and social context

Correspondence to:

Samantha WellsSocial, Prevention and Health Policy Research DepartmentCentre for Addiction and Mental Health100 Collip CircleSuite 200LondonOntario N6G 4X8CanadaTel:

+

1 519 858 5000Fax:

+

1 519 858 5199E-mail: [email protected]

Submitted 21 September 2001; initial review completed 7 January 2002;

final version accepted 29 April 2002

Aggression involving alcohol: relationship to drinking patterns and social context

Samantha Wells & Kathryn Graham

Social, Prevention and Health Policy Research Department, Centre for Addiction and Mental Health, London, Ontario, Canada

INTRODUCTION

Experimental research suggests that alcohol consump-tion combined with expectations about the effects of alco-hol plays a causal contributing role in aggression (seemeta-analysis by Bushman 1997). A substantial propor-tion of naturally occurring incidents of aggression andviolent crime involves one or more participants who havebeen drinking (Murdoch, Pihl & Ross 1990; Pernanen1991; Collins 1993; Wells, Graham & West 2000). Alco-

hol consumption has also been linked to the severity ofaggression, with aggression involving injury being morelikely to involve alcohol (Gerson & Preston 1979; Martin& Bachman 1997; Sharps

et al

. 2001). This association ofalcohol with aggression and greater aggression severitymay be due, at least in part, to pharmacological effects ofalcohol, such as increased power concerns (McClelland

et al

. 1972), increased emotional lability and focus on thepresent (Graham, West & Wells 2000), decreased aware-ness of internal cues or less self-awareness (Hull 1981),

34



Addiction,

98,

33–42

decreased ability to consider consequences (Hull & Bond1986; Pihl, Peterson & Lau 1993; Ito, Miller & Pollock1996) or reduced ability to problem solve (Sayette,Wilson & Elias 1993).

Given the possible role of the pharmacological effectsof alcohol, Pernanen (1991) has suggested that dose-related effects such as alcohol intoxication level might bemore important than the mere absence or presence ofalcohol in terms of increasing aggression severity. Severalstudies have found support for this hypothesis. In partic-ular, emergency room studies have found that injuryseverity is associated with level of intoxication (e.g.Shepherd

et al

. 1988; Honkanen & Smith 1990); how-ever, such studies focus on the alcohol consumption ofvictims and not perpetrators, and therefore provide littleinformation regarding the role of alcohol in the escala-tion process (Cherpitel 1993). More direct support for thishypothesis was obtained in a recent descriptive study ofnaturally occurring aggression among young adults inbars (Graham & Wells 2001a), which found that partici-pants in incidents of severe aggression were more intoxi-cated than participants in incidents in which aggressionwas less severe.

Recent research across a number of countries has sug-gested that involvement in alcohol-related aggression ismore likely among people who are heavier drinkers, espe-cially those who have a usual drinking pattern character-ized by drinking large amounts per occasion or drinkingto intoxication (Room, Bondy & Ferris 1995; Rossow1996; Dawson 1997; Giesbrecht & West 1997). This rela-tionship could be due to heavier drinkers being generallymore aggressive than lighter drinkers [e.g. more aggres-sive people are more likely to drink heavily, as indicated byWhite, Brick & Hansell (1993)] or to increased risks asso-ciated with higher levels of intoxication (Lipsey

et al

.1997; Roizen 1997). In a multivariate analysis of self-reported aggression, Scott, Schafer & Greenfield (1999)found that, while drinking history was associated signif-icantly with both perpetration and victimization control-ling for other variables, this variable was not associatedsignificantly with drinking in the event (i.e. whether ornot the victim or perpetrator had been drinking at thetime of aggression). However, the measure of drinkinghistory employed by these authors was based on a dichot-omous variable of respondents drinking at least once inthe previous year and at least more than five drinks onone occasion in their lifetime versus not drinking in theprevious year and less than five drinks on a single occa-sion. A better understanding of the relationship betweenthese variables might be achieved if more sensitive mea-sures of drinking pattern are employed, along with thecontrol of potential confounders such as gender and age.

Social context factors (e.g. physical and social envi-ronment) have also been found to influence whether a

drinking occasion is likely to involve aggression (Graham

et al

. 1980; Homel & Clark 1994); however, there hasbeen relatively little research on contextual factors asso-ciated with aggression involving alcohol versus incidentsnot involving alcohol. The relationship among partici-pants in aggressive incidents is one factor that appears topredict alcohol involvement, although the results havenot always been consistent. In a survey study of violentcrime in one Canadian community, Pernanen (1991)found that aggressive incidents between strangers were

more likely

to involve alcohol (78%) than incidents inwhich the respondent knew the assailant a little (50%)and incidents in which the respondent knew the assailantvery or fairly well (46%). However, in his analysis of policedata, Pernanen (1991) found that marital violence hadthe highest alcohol involvement. Results from the USNational Crime Victimization Survey (Greenfeld 1998)indicated that the offender was less likely to have beendrinking at the time of the crime in incidents involvingstrangers (24% of incidents involved an offender who hadbeen drinking) compared to incidents of violence betweenintimates (40% involved a drinking offender). In terms ofother aspects of the social context, studies have suggestedthat alcohol involvement was highest for incidents thatoccurred at weekends and late at night or very early inthe morning (Gerson & Preston 1979; Greenfeld 1998).Pernanen (1991) found that alcohol was most likely to beinvolved in incidents between young people, especiallymales, and aggression involving more than two partici-pants. Other than these findings, little is known aboutsocial context and environmental factors associated withalcohol involvement in aggression compared to aggres-sion not involving alcohol (Pernanen 1993).

As part of a larger study on aggression, the presentresearch examines three main areas related to achievinga better understanding of alcohol-related and intoxicatedaggression. First, we explore whether aggression severityand injury are associated with any alcohol involvementand alcohol intoxication of participants. Secondly, weexamine whether heavier drinkers are more likely tobecome involved in any aggression (alcohol and non-alcohol-related) than non-heavy drinkers or whetherthey are only more likely to be involved in alcohol-relatedaggression (controlling for potential confounders suchas gender and age).Thirdly, we identify social contextvariables that discriminate between alcohol-involvedincidents versus incidents not involving alcohol.

METHOD

Random digit dialing (RDD) and Computer assisted tele-phone interviewing (CATI) were used to obtain a randomsample of Ontario adults age 18–60 years. A two-stage

Aggression, drinking patterns and social context

35


Addiction,

98,

33–42

probability procedure was used involving the randomselection of telephone numbers from listings of activearea codes and exchanges, and the random selection ofrespondents within households according to the mostrecent birthday of household members. The overall com-pletion rate for the survey was 67% based on 1753 inter-view completions from 2623 eligible households.

Measures

Most recent incident of aggression

The telephone interview consisted of an initial screeningsection where respondents were asked whether theyhad been involved personally in an incident of physicalaggression ever in their lives, followed by a second ques-tion as to how many times they had experienced physicalaggression involving only adults (i.e. age 18 and over) inthe previous 12 months. For those who reported at leastone incident of physical aggression in the previous12 months, specific questions were asked regarding themost recent incident. Notably, respondents were askedabout personal involvement in aggression rather thanvictimization experiences or violent crime perpetration.This manner of questioning allowed respondents toreport incidents in which they were mutual participants(i.e. not necessarily victims or perpetrators) or third par-ties who became involved in aggression. Among thosewho did not experience aggression in the previous12 months, a random sample of 107 respondents wasinterviewed for a subset of questions (i.e. demographicquestions and drinking pattern) and this sample wascompared with respondents who reported aggression.Those who reported physical aggression were signifi-cantly more likely to be male (

χ

2

=

8.8, df

=

1,

P

=

0.003)and in the younger age groups (

χ

2

=

23.0 df

=

3,

P

<

0.001).

Alcohol consumption and perceived intoxication level prior to incident of aggression

In terms of alcohol consumption, respondents were askedwhether they drank any alcohol in the 6 hours before theincident. Those who reported drinking were asked howmany drinks they had in the 6-hour period before theincident and to rate the effects of alcohol at the time from1: totally sober to 10: falling down drunk. They were alsoasked about the alcohol consumption and intoxicationlevel of the person who was most directly involved in theincident. Respondents were not asked about the numberof drinks consumed by their opponents because this infor-mation would be unknown to many respondents. Anunintended consequence of the interview protocol is thatrespondents sometimes reported on the consumption of

people on the same side as respondents (i.e. not theiropponents). Only consumption and perceived intoxica-tion levels of respondents’ opponents were used in thepresent paper. A dichotomous variable was computedreflecting whether the respondent and/or opponent hadbeen drinking (i.e. any drinking) or neither had beendrinking. For respondents who reported that they had notbeen drinking and they did not know whether their oppo-nent had been drinking, this variable was coded as ‘nei-ther drinking’.

Characteristics of respondent

Standard questions regarding the age, marital status andemployment status of respondents were asked. Generaldrinking patterns of the respondent were assessed usingstandard alcohol use items (see Rehm 1998). In particu-lar, respondents were asked whether they had had a drinkof any alcoholic beverage in the past 12 months. Forthose who reported drinking in the previous year, theywere asked how often they drank alcoholic beverages,how often they had five or more drinks at the same sittingor occasion, usual number of drinks consumed on dayswhen they drank and the largest number of drinks theycould recall having on one occasion during the past12 months. A summary variable was computed from thestandardized scores of the latter four drinking patternvariables.

Characteristics of aggressive incident

For those who reported an incident of physical aggressionin the previous 12 months, contextual information wasobtained about the most recent incident of aggression.Time of incident was coded using the following catego-ries: midnight to 6 a.m., 6 a.m. to 6 p.m. and 6 p.m. tomidnight. Day of week was coded as weekday (Monday–Thursday) versus weekend (Friday–Sunday). Social con-text was coded from closed and open-ended responses tocapture both the location and the social setting using thefollowing categories: (1) in or around licensed premises(bar, licensed restaurant, pool hall); (2) a social event orsocial situation not at a licensed premise (e.g. a party orsocial occasion at the respondent’s home, another per-son’s home, a park, the beach); (3) an interaction thatoccurred at home between residents of the same home(i.e. a domestic situation between family members,spouses or roommates); (4) a public place other than asocial gathering (e.g. on the street, in an elevator); (5) allother contexts (e.g. work-place, ex-husband/boyfriend athome of former wife/girlfriend, between neighbours).

Variables were also developed to reflect the partici-pants in the incident. The number of people involved inthe incident was assessed by asking the respondent how

36



Addiction,

98,

33–42

many people, other than the respondent, aged 18 yearsor older, were involved directly in the incident (i.e. peoplewho actually participated in the incident as opposed tothose who were simply standing and watching). The gen-der composition of the incident (i.e. male only, femaleonly or males and females) was determined based on thenumber of males and females who were involved directlyin the incident of aggression. Respondents were asked toprovide the gender and age of another person, other thanthemselves, who was involved most directly in the inci-dent of aggression. In some instances, the respondentidentified someone in the incident who was actually onthe same side as the respondent rather than an opponent.These responses were treated as missing unless additionalinformation was available regarding the main opponent,in which case the appropriate code was given. Finally,respondents were asked how they were related to theperson who was involved most directly. This variable wascategorized as follows: (1) intimate couples and formerpartners; (2) friends, acquaintances or colleagues atwork; (3) family; (4) strangers; and (5) other (e.g. cus-tomer, patient).

Aggression severity

Respondents were asked whether they were injured orhurt, including minor injuries such as bruises or cuts,and whether they received medical attention from a doc-tor, nurse, paramedic or other professional at the time ofthe incident or in the next day or so. In addition, theywere asked whether police were involved in the incidentat any time. A final indicator of incident severity waswhether the respondent threatened the opponent orthe opponent threatened the respondent. Threat wasincluded as a measure of severity because it is likely tocontribute to the escalation of aggression. A summaryscore of severity was created based on the sum of thefour severity items (i.e. police involvement, respondentinjured, threats by opponent and threats by respondent).

Analyses

Bivariate analyses were conducted to assess the relation-ships of alcohol consumption and perceived intoxicationlevel prior to aggression with severity of aggression using

χ

2

tests and

t

-tests. Pearson’s correlation was computedto assess associations between continuous variables, suchas the summary of aggression severity score by respon-dent and opponent intoxication and general drinking pat-terns by intoxication levels at the time of the incident.One-way analysis of variance (ANOVA) tests were con-ducted to examine the relationship between drinking pat-tern and alcohol involvement in the aggressive incident.For analysis of the relationship between context of

aggression and alcohol involvement, odds ratios and con-fidence intervals (CIs) were computed along with

χ

2

testsof association. Finally, multiple logistic regression analy-sis was performed on the dichotomous outcome variable,respondent and/or opponent drinking versus neitherdrinking, for the social context variables, controlling forage and gender. Severity of aggression and drinking pat-tern of respondent were also entered into the regressionmodel as covariates.

RESULTS

A total of 178 respondents reported being involved inphysical aggression in the previous 12 months. Twenty-one respondents were excluded from the analysesbecause the respondent was only a witness to aggressionor no physical aggression was reported, or because theresponded described an aggressive incident that occurredas part of playing a sport or an incident in which a par-ticipant was less than 18 years of age.

Of the remaining 157 respondents who reported inci-dents of aggression, 57.3% were male, 49.7% were nevermarried, 32.5% were married or living with a partner,17.2% were divorced or separated and 0.6% were wid-owed. The mean age was 33.7 (age groups: 18–24,28.2%; 25–34, 26.9%; 35–44, 23.7%; 45–60, 21.2%).Drinking in the 6 hours before the incident was reportedby 59 respondents (37.6% of those who reported aggres-sion). Of respondents who were able to assess drinking onthe part of the opponent (

n

=

132), 59.0% reported thattheir opponent had been drinking. Overall, respondentsreported that they and/or their opponent had been drink-ing in 50.3% (

n

=

79) of incidents. In terms of perceivedintoxication levels, the mean rating of intoxication forrespondents was 4.2 (

n

=

58) and the mean rating foropponents was 6.5 (

n

=

57). Number of drinks consumedprior to aggression was significantly correlated withperceived intoxication level of respondent (

r

=

0.63,

P

<

0.001) and opponent (

r

=

0.36,

P

<

0.05). These cor-relations remained significant, controlling for the demo-graphic variables (i.e. partial

r

=

0.58,

P

<

0.001 andpartial

r

=

0.38,

P

<

0.05).

Alcohol involvement, number of drinks, and perceived intoxication levels by incident severity

As shown in Table 1, alcohol involvement

per se

was notrelated significantly to indicators of aggression severity.However, a significantly larger number of drinks was con-sumed by respondents in the 6 hours prior to aggressionfor those who reported an injury. Similarly, higher levelsof perceived intoxication were reported by respondentswho had an injury or threatened someone else.

Aggression, drinking patterns and social context

37


Addiction,

98,

33–42

Tabl

e 1

Alc

ohol

invo

lvem

ent,

num

ber

of d

rinks

con

sum

ed b

y re

spon

dent

, per

ceiv

ed in

toxi

catio

n le

vel o

f th

e re

spon

dent

, and

per

ceiv

ed in

toxi

catio

n le

vel o

f th

e m

ain

oppo

nent

by

seve

rity

of a

ggre

ssio

n.

n

Resp

onde

nt a

nd/o

r opp

onen

t drin

king

vers

us

no o

ne d

rinki

ng (

n

=

157

)

Num

ber

of d

rinks

con

sum

ed b

y re

spon

dent

in 6

hou

rs b

efor

e in

ciden

t (

n

=

58)

Perc

eive

d in

toxi

catio

n le

vel o

f re

spon

dent

(

n

=

58)

Perc

eive

d in

toxi

catio

n le

vel o

f m

ain

oppo

nent

(

n

=

57)

% c

onsu

min

g al

coho

lO

R95

% C

IM

ean

ratin

g

t

-val

ueM

ean

ratin

g

t

-val

ueM

ean

ratin

g

t

-val

ue

Resp

onde

nt w

as in

jure

d or

hur

t

a

Yes

4447

.7%

0.9

0.4–

1.7

7.8

t

(56

df)

=

2.1

,

P

=

0.0

445.

4

t

(56

df)

=

2.9

,

P

=

0.0

057.

1

t

(49

df)

=

1.8

,

P

=

0.0

84N

o11

351

.3%

5.0

3.7

6.3

χ

2

=

0.2

,

P

=

0.6

85Re

spon

dent

thr

eate

ned

som

eone

Yes

1553

.3%

1.1

0.4–

3.2

9.0

t

(56

df)

=

2.0

,

P

=

0.0

506.

3

t

(56

df)

=

2.9

,

P

=

0.0

058.

2

t

(52

df)

=

2.3

,

P

=

0.0

28N

o13

951

.1%

5.3

3.9

6.3

χ

2

=

0.0

,

P

=

0.8

68O

ppon

ent

thre

aten

ed r

espo

nden

tYe

s80

48.8

%0.

90.

5–1.

76.

0

t

(55

df)

=

0.5

,

P

=

0.6

444.

4

t

(56

df)

=

1.0

,

P

=

0.3

217.

0

t

(51

df)

=

1.7

,

P

=

0.0

95N

o76

51.3

%5.

43.

96.

1

χ

2

=

0.1

,

P =

0.74

9Po

lice

wer

e in

volv

edYe

s30

50.0

%1.

00.

4–2.

15.

4t(

56 d

f) =

0.3

, P =

0.7

594.

3t(

56 d

f) =

0.2

, P =

0.8

686.

6t(

52 d

f) =

0.1

, P =

0.9

13N

o12

750

.4%

5.8

4.1

6.5

χ2 =

0.0,

P =

0.9

69

a Incl

udin

g m

inor

inju

ries

such

as

brui

ses

or c

uts.

38 Samantha Wells & Kathryn Graham

© 2003 Society for the Study of Addiction to Alcohol and Other Drugs Addiction, 98, 33–42

Opponents who threatened the respondent were also per-ceived to be more intoxicated than opponents who did notthreaten. In addition, a summary score of aggressionseverity (i.e. a sum of police involvement, injury, threatsby opponent, threats by respondent) was significantlycorrelated with respondent’s intoxication level (r = 0.36,P < 0.01) and the perceived intoxication level of the oppo-nent (r = 28, P < 0.05).

Alcohol-related aggression by drinking patterns of respondents

Table 2 displays the mean values of the drinking patternvariables for three groups: those reporting no aggression(i.e. randomly selected comparison group), those report-ing aggression not involving alcohol (i.e. neither respon-dent nor opponent were drinking) and those reportingaggression involving alcohol (i.e. drinking by respondentand/or the main opponent). Those who reported notdrinking in the previous 12 months were excluded fromthese analyses, reducing the sample to 140 for those whoexperienced aggression and 94 for the comparison group.Respondents who reported that their most recent inci-dent of aggression involved alcohol were significantlymore likely to drink frequently and in greater amountsthan respondents who reported either no aggression inthe past 12 months or that their most recent incident didnot involve alcohol. This relationship weakened, butremained significant controlling for other factors relatedto drinking and/or aggression (i.e. sex, age, educationlevel, whether married and whether employed, a studentor other employment status).

To investigate whether drinking patterns were associ-ated with intoxication at the time of the incident amongrespondents who were drinking at the time of the incident(or reported that the opponent had been drinking), cor-relations were computed for the four drinking patternvariables with number of drinks consumed prior toaggression and perceived intoxication levels of the respon-dent and opponent (see Table 3). Number of drinks con-sumed in the 6 hours before the incident was significantlyassociated with all drinking variables except frequency ofdrinking. Most of these relationships remained significantwhen age, gender, marital status, education level andemployment status were controlled for. Perceived intoxi-cation level was significantly associated with usual num-ber of drinks per occasion and largest number of drinksconsumed on a single occasion. None of the relationshipsbetween respondent’s drinking pattern and perceivedintoxication level of opponent reached the significance cri-terion, although the relationship for usual number ofdrinks and frequency of consuming five or more drinks peroccasion were in the same general direction.

Alcohol-related aggression and social context

As shown in Table 4, alcohol-related incidents were morelikely to involve only males, respondents who were aged30 or less, three or more people and strangers, and weremore likely to occur after midnight, at weekends and atbars or social events.

Table 5 provides the results of the multiple logisticregression analysis of the relationship between socialcontext characteristics and whether the respondent and/

Table 2 No aggression, alcohol and non-alcohol-related aggression by drinking patterns.

Drinking variable

Mean score

No aggressionin past 12 months(n = 94)

Most recent incidentdid not involvealcohol (n = 66)

Most recentincident involvedalcohol (n = 74)

F for one-way ANOVAa (df )

F controllingfor respondentcharacteristicsb (df )

Frequency of drinking during past12 months (range 1–8)c

3.6 3.1 4.5 10.7 (2, 230)*** 7.4 (2, 212)**

Usual number of drinks per drinkingoccasion (range 1–24)

2.6 2.4 5.7 25.1 (2, 227)*** 10.1 (2, 209)***

Frequency of consuming 5 + drinksper occasion (range 1–9)d

2.5 2.4 4.5 28.9 (2, 231)*** 12.4 (2, 213)***

Largest number of drinks consumedon a single occasion during thepast 12 months (range 1–36)

5.8 5.9 12.4 29.5 (2, 219)*** 7.6 (2, 203)**

aFor significant F-values, pairwise comparisons using Scheffé post hoc test indicated that mean score for most recent incident involved alcohol was significantly dif-ferent from the other two means. No other pairwise comparisons were significant. bGender, age, education level, whether married, whether employed full-time, student or other employment status. c1 = less than once a month, 2 = once a month, 3 = two to three times a month, 4 = once a week, 5 = two to three times a week, 6 = four to five times a week, 7 = about every day, 8 = more than once a day. d1 = never, 2 = one to five times a year, 3 = six to 11 times a year, 4 = about once a month, 5 = two or three times a month, 6 = once or twice a week, 7 = three or four times a week, 8 = five or six times a week, 9 = every day. **P < 0.01; ***P < 0.001.

Aggression, drinking patterns and social context 39


or opponent had been drinking at the time of the incident,controlling for gender and age. Due to the small samplesize, not all variables examined in the bivariate analyses(i.e. Table 4) were included in the multiple regressionmodel. Variables excluded were those with a low numberof valid cases and those correlated with other importantvariables (e.g. time of day, day of week and location werehighly correlated). As well, 10 cases in the ‘other’ cate-gory of relationship of respondent to opponent wereexcluded to reduce the likelihood of producing unstableodds ratios. Selected variables were entered as threeblocks. First, age and gender of respondent were enteredinto the model. Gender was associated significantly withalcohol versus non-alcohol-related aggression, withmales being significantly more likely to report alcohol-related aggression. Secondly, the contextual variablesweekend (versus weekday), number of people involved(three or more versus 2), and relationship of respondentto main opponent, were entered into the regressionmodel. With the inclusion of these contextual variables,gender remained significant, and aggression occurringon weekends was significantly more likely to be alcohol-related. Finally, two variables were entered as a block,drinking pattern of respondent (as measured by a stan-dardized summary score of the four drinking patternvariables) and severity of aggression (a sum of policeinvolvement, threats and injury). Drinking pattern ofrespondent was associated significantly with alcohol-related aggression, with heavier drinkers being morelikely to report alcohol-related aggression, controlling forage, gender and contextual characteristics.

DISCUSSION

Aggression severity, as measured by injury and threat bythe respondent, was not associated with alcohol involve-

ment per se, but was associated positively with perceivedintoxication level of the respondent and to some extentintoxication of the opponent. Some support, therefore,was found for the theory posited by Pernanen (1991) thatalcohol intoxication, rather than the mere involvement ofalcohol, is important in terms of the escalation of aggres-sion into more severe violence.

Drinking pattern of the respondent was significantlyassociated with alcohol in the event; that is, respondentswho reported alcohol-related aggression were morefrequent and heavier drinkers than those who reportedaggression that did not involve alcohol or no aggression.This association was found even controlling for age,gender, education, marital status and employment sta-tus. This finding contradicts results from a study by Scottet al. (1999), which measured drinking pattern with asingle dichotomous measure (i.e. drinking in the pastyear plus drinking at least five drinks on a single occasionversus not drinking or never drinking at least five drinks)and suggests that greater attention needs to be paid toindividuals who become aggressive only when they drink.In particular, a significant correlation was found betweenheavy drinking, as measured by usual number of drinks,and alcohol intoxication in the aggressive event and thisrelationship remained significant controlling for demo-graphic variables. These findings suggest that heavierdrinkers might be at greater risk of alcohol-related andintoxicated aggression rather than non-alcohol-relatedaggression; that is, it may not be that heavier drinkers aremore aggressive generally, but that they drink more peroccasion and are therefore at greater risk of alcohol-related aggression, perhaps because of the effects ofalcohol or possibly because of environmental factors indrinking contexts that increase the likelihood of aggres-sion (e.g. competitive games). Jaffe, Babor & Fishbein(1988) found that alcoholics reported more anger andaggression when drinking than when sober, suggesting

Table 3 Correlations between perceived intoxication levels, number of drinks and drinking patterns.

Drinking variable

Number of drinks con-sumed 6 hours before aggression (n = 57)

Perceived intoxication level of respondent (n = 57)

Perceived intoxication level of opponent (n = 51)

r Partial ra r Partial ra r Partial ra

Frequency of drinking during past 12 months (range 1–8)b 0.09 0.04 0.16 0.14 −0.07 −0.03Usual number of drinks per drinking occasion (range 1–24) 0.47*** 0.42*** 0.38** 0.36** 0.23 0.24Frequency of consuming 5 + drinks per occasion (range 1–9)c 0.32* 0.24 0.25 0.18 0.14 0.14Largest number of drinks consumed on a single occasion during

the past 12 months (range 1–36)0.57*** 0.51*** 0.33* 0.22 0.03 0.04

aControlling for gender, age, education level, whether married, whether employed full-time, student or other employment status. b1 = less than once a month, 2 = once a month, 3 = two to three times a month, 4 = once a week, 5 = two to three times a week, 6 = four to five times a week, 7 = about every day, 8 = more than once a day. c1 = never, 2 = one to five times a year, 3 = six to 11 times a year, 4 = about once a month, 5 = two or three times a month, 6 = once or twice a week, 7 = three or four times a week, 8 = five or six times a week, 9 = every day. *P < 0.05; **P < 0.01; ***P < 0.001.



an enhanced effect of alcohol on aggression amongheavy drinkers. Analyses of amount of alcohol consump-tion and intoxication level among those who had beendrinking at the time of the incident suggested that heavierdrinkers consumed more alcohol and were more intoxi-cated at the time of the incident compared with lighterdrinkers.

The bivariate analyses of social context variables andalcohol- versus non-alcohol-related aggression revealedthat respondents who reported incidents that occurred inbars, between males, with a large number of people,between strangers, late at night and at weekends were

more likely to report that they and/or their opponent hadbeen drinking. These results are consistent with previousfindings that alcohol-related aggression is more likely tooccur at night and on weekends (Greenfeld 1998) andamong males rather than females (Graham & Wells2001a). In terms of relationship to opponent, the resultssupport findings from Pernanen (1991) rather thanthose of Greenfeld (1998). For example, Greenfeld (1998)reported that drinking by the offender was greatest forviolence between spouses and ex-spouses and consider-ably lower between strangers. In the present study, simi-lar to the survey results of Pernanen (1991), alcohol

Table 4 Social context variables by alcohol versus non-alcohol-related aggression.

n

Respondent and/or opponent drinking versus no one drinking (n = 157)

% consuming alcohol OR 95% CI

All incidents 157 50.3 42.2–58.4Gender composition of participants

Male only 72 62.5 2.4* 1.2–4.8Female only 16 37.5 0.9 0.3–2.7Male and female (reference) 69 40.6

χ2 = 7.9, P = 0.019Respondent age 30 or less

Yes 75 61.3% 2.3* 1.2–4.4No (reference) 81 39.5%

χ2 = 7.4, P = 0.006Time of incident

Midnight to 6 a.m. 26 92.3% 60.0** 12.0–300.36 p.m. to midnight 67 62.7% 8.4*** 3.5–20.16 a.m. to 6 p.m. (reference) 54 16.7%

χ2 = 46.9, P < 0.0001Day of week

Weekend 78 71.8 5.8*** 2.8–12.4Weekday (reference) 56 30.4

χ2 = 22.6, P < 0.0001Location of aggression

Bar 47 95.7 48.8*** 8.8–270.9Public place 33 24.2 0.7 0.2–2.4Social event 25 52.0 2.4 0.7–8.2Work/other 33 21.2 0.6 0.2–2.1Home (reference) 19 31.6

χ2 = 61.7, P < 0.0001Number of people involved

3 or more people 61 60.7 2.0* 1.0–3.82 people (reference) 96 43.7

χ2 = 4.3, P = 0.038Relationship of respondent to main opponent

Friend/acquaintance 52 46.2 1.2 0.5–3.0Family 10 40.0 0.9 0.2–4.1Stranger 52 65.4 2.7* 1.1–6.8Other 10 10.0 0.2 0.0–1.4Intimate couple (reference) 29 41.4

χ2 = 12.8, P = 0.012

*P < 0.05; **P < 0.01; ***P < 0.001.

Aggression, drinking patterns and social context 41


consumption by the respondent and/or the opponent wasmost common in incidents of aggression between strang-ers. Alcohol-related aggression is known to vary greatlyby culture (MacAndrew & Edgerton 1969); therefore, it ispossible that this difference reflects differences in alcohol-related aggression between Canadians (the present study,Pernanen 1991) versus Americans (Greenfeld 1998).Future research is needed on cross-cultural variations inthe social context of alcohol-related aggression to betterunderstand the inconsistency in this finding.

In the multivariate analyses, the most important vari-ables associated with alcohol-related aggressive incidentswere gender of respondent, weekend (versus weekday),and drinking pattern of respondent. Incidents reported bymales and occurring on weekends were significantlymore likely to involve alcohol. These relationshipsremained significant controlling for age and drinking pat-tern of the respondent as well as other contextual char-acteristics of aggressive incidents. Drinking pattern ofrespondent was also significant, controlling for the con-textual variables, age and gender of respondent. Again,this finding suggests that heavy drinkers are more likelyto experience alcohol-related aggression than non-alcohol-related aggression.

A possible limitation of the present study was that,despite the large original sample size, the number ofrespondents who reported aggression was relativelysmall, which limited the number of variables that couldbe included in the multivariate analyses. A larger samplesize is needed in future research to assess the combinedeffect of various variables on alcohol-related aggression.Another limitation of the study is that perceived intoxica-tion levels may be somewhat biased. Respondents may

underestimate their own intoxication levels and overesti-mate that of others (see Graham & Wells 2001b based onthe same data). Therefore, results based on these mea-sures of intoxication should be interpreted with caution.In addition, whether the respondent was a victim, perpe-trator or mutual participant was not determined in thestudy and may be associated with alcohol involvement. Inparticular, it is possible that aggression involving mutualparticipants rather than clear victims or perpetrators ismore likely to involve alcohol. It would be useful in futureresearch to examine these relationships. Finally, theresponse rate of 67%, though comparable to mosttelephone surveys, may not be fully representative of thepopulation.

In summary, the present results suggest that bothdrinking pattern and contextual factors distinguishalcohol- and non-alcohol-related aggression. These fac-tors merit further research both in terms of personcharacteristics and environmental aspects that dif-ferentially predict alcohol versus non-alcohol-relatedincidents. Another notable finding is that intoxicationlevel may be a more important predictor of aggressionseverity and risk of injury than alcohol involvement.Research is needed that examines intoxication levelsand the escalation of aggression, especially in the studyof aggression severity and injury. A possible implicationof these findings is that prevention initiatives need tofocus on reducing excessive drinking among high-riskgroups, such as young males, and target high risk socialcontexts, such as weekend drinking. As well, heavierdrinkers may be an important group to target for inter-ventions focussed on reducing aggression resulting fromdrinking.

Independent variables OR 95% CI

Respondent male 4.4* 1.2–15.9Respondent age less than 30 2.1 0.6–6.8Weekend 10.4** 3.1–35.7Number of people involved

3 or more people2 people (reference)

1.9 0.5–7.3

Relationship of respondent to main opponentFriend/acquaintance 0.2 0.0–1.1Family 0.5 0.0–7.6Stranger 0.4 0.1–2.6Intimate couple (reference)

Drinking pattern of respondent 6.7*** 2.4–19.0Severity of aggression 0.7 0.4–1.2

Model χ2 = 56.3, df = 9, P < 0.0001

*P < 0.05; **P < 0.01; ***P < 0.001.

Table 5 Logistic regression of gender, age,social context variables, drinking patternof respondent and severity of aggressionregressed onto alcohol-versus non-alcohol-related aggression (n = 104).



REFERENCES

Bushman, B. J. (1997) Effects of alcohol on human aggression:validity of proposed explanations. In: Galanter, M., ed. RecentDevelopments in Alcoholism, pp. 227–243. New York: PlenumPress.

Buss, A. H. & Perry, M. (1992) The aggression questionnaire.Journal of Personality and Social Psychology, 63, 452–459.

Cherpitel, C. J. (1993) What emergency room studies revealabout alcohol involvement in violence-related injuries. AlcoholHealth and Research World, 17, 162–166.

Collins, J. J. (1993) Drinking and violence: an individualoffender focus. In: Martin, S. E., ed. Alcohol and InterpersonalViolence: Fostering Multidisciplinary Perspectives, pp. 221–235.Research Monograph no. 24. Rockville, MD: NIH.

Dawson, D. A. (1997) Alcohol, drugs, fighting and suicideattempt/ideation. Addiction Research, 5, 451–472.

Gerson, L. & Preston. D. (1979) Alcohol consumption and theincidence of violent crime. Journal of Studies on Alcohol, 40,307–312.

Giesbrecht, N. & West, P. (1997) Drinking patterns anddrinking-related benefits, harm and victimization experi-ences: reports from community-based general population sur-veys. Contemporary Drug Problems, 24, 557–579.

Graham, K., Larocque, L., Yetman, R., Ross, T. J. & Guistra, E.(1980) Aggression and barroom environments. Journal ofStudies on Alcohol, 41, 277–292.

Graham, K. & Wells, S. (2001a) Aggression among young adultsin the social context of the bar. Addiction Research and Theory,9, 193–219.

Graham, K. & Wells, S. (2001b) ‘I’m okay, you’re drunk!’ Self–other differences in the perceived effects of alcohol in real-lifeincidents of aggression. Contemporary Drug Problems, 28,441–462.

Graham, K., West, P. & Wells, S. (2000) Evaluating theories ofalcohol-related aggression using observations of young adultsin bars. Addiction, 95, 847–673.

Greenfeld, L. A. (1998) Alcohol and Crime: An Analysis of NationalData on the Prevalence of Alcohol Involvement in Crime.Washington, DC: Bureau of Justice Statistics.

Homel, R. & Clark, J. (1994) The prediction and prevention ofviolence in pubs and clubs. Crime Prevention Studies, 3, 1–46.

Honkanen, R. & Smith, G. S. (1990) Impact of acute alcoholintoxication and severity of injury: a cause-specific analysis ofnon-fatal trauma. Injury, 21, 353–357.

Hull, J. G. (1981) A self-awareness model of the causes andeffects of alcohol consumption. Journal of Abnormal Psychol-ogy, 90, 586–600.

Hull, J. G. & Bond, C. F. (1986) Social and behavioral conse-quences of alcohol consumption and expectancy: a meta-analysis. Psychological Bulletin, 99, 347–360.

Ito, T. A., Miller, N. & Pollock, V. E. (1996) Alcohol and aggres-sion: a meta-analysis on the moderating effects of inhibitorycues triggering events, and self-focused attention. Psychologi-cal Bulletin, 120, 60–82.

Jaffe, J. H., Babor, T. F. & Fishbein, D. H. (1988) Alcoholics,aggression and antisocial personality. Journal of Studies onAlcohol, 49, 211–218.

Lipsey, M. W., Wilson, D. B., Cohen, M. A. & Derzon, J. H. (1997)Is there a causal relationship between alcohol use and vio-lence? In: Galanter, M., ed. A Synthesis of Evidence. RecentDevelopments in Alcoholism, Vol. 13, pp. 245–282. New York:Plenum Press.

MacAndrew, C. & Edgerton, R. B. (1969) Drunken Comportment:A Social Explanation. Chicago: Aldine.

Martin, S. E. & Bachman, R. (1997) The relationship of alcoholto injury in assault cases. In: Galanter, M., ed. Recent Develop-ments in Alcoholism, vol. 13, pp. 42–56. New York: PlenumPress.

Mcclelland, D. C., Davis, W. N., Kalin, R. & Wanner, E. (1972)The Drinking Man. Alcohol and Human Motivation. Toronto,Canada: Collier-Macmillan Canada.

Murdoch, D., Pihl, R. O. & Ross, D. (1990) Alcohol and crimes ofviolence: present issues. International Journal of the Addictions,25, 1065–1081.

Pernanen, K. (1991) Alcohol in Human Violence. New York:Guilford Press.

Pernanen, K. (1993) Research approaches in the study ofalcohol-related violence. Alcohol Health and Research World,17, 101–107.

Pihl, R. O., Peterson, J. B. & Lau, M. A. (1993) A biosocial modelof the alcohol–aggression relationship. Journal of Studies onAlcohol, 11, 128–139.

Rehm, J. (1998) Measuring quantity, frequency and volume ofdrinking. Alcoholism: Clinical and Experimental Research Supple-ment, 22, 4S–14S.

Roizen, J. (1997) Epidemiological issues in alcohol-related vio-lence. In: Galanter, M., ed. Recent Developments in Alcoholism:Alcohol and Violence, vol. 13, pp. 7–40. New York: PlenumPress.

Room, R., Bondy, S. J. & Ferris, J. (1995) The risk of harm to one-self from drinking, Canada 1989. Addiction, 90, 499–513,.

Rossow, I. (1996) Alcohol-related violence: the impact of drink-ing pattern and drinking context. Addiction, 91, 1651–1661.

Sayette, M. A., Wilson, T. & Elias, M. J. (1993) Alcohol andaggression: a social information processing analysis. Journal ofStudies on Alcohol, 54, 399–407.

Scott, K. D., Schafer, J. & Greenfield, T. K. (1999) The role of alco-hol in physical assault perpetration and victimization. Journalof Studies on Alcohol, 60, 528–536.

Sharps, P. W., Campbell, J., Campbell, D., Gary, F. & Webster, D.(2001) The role of alcohol use in intimate partner femicide.American Journal on Addictions, 10, 122–135.

Shepherd, J., Irish, M., Scully, C. & Leslie, I. (1988) Alcoholintoxication and severity of injury in victims of assault. BritishMedical Journal, 296, 1299.

Wells, S., Graham, K. & West, P. (2000) Alcohol-related aggres-sion in the general population. Journal of Studies on Alcohol, 61,626–632.

White, H. R., Brick, J. & Hansell, S. (1993) A longitudinal inves-tigation of alcohol use and aggression in adolescence. Journalof Studies on Alcohol, 11, 62–77.

Documents

Aggression involving alcohol: relationship to drinking patterns and social context