Drug Abuse/Addiction /4 pages /Follow outline/Peer Reviewed Journals attached/ 8% Plagiarism

Alcohol, Drugs, and Violence: A Meta-Meta-Analysis

Aaron A. Duke Yale University School of Medicine

Kathryn M. Z. Smith Yale University School of Medicine and Wayne State University

Lindsay M. S. Oberleitner, Alexander Westphal, and Sherry A. McKee Yale University School of Medicine

Objective: The aim of this project was to provide a comprehensive overview and quantitative synthesis of the links between alcohol, drugs, and violence with established meta-meta-analysis methodology. Gender, psychotic illness (present vs. absent), role (perpetrator vs. victim), substance (alcohol, drugs, or both), operationalizations of violence (e.g., laboratory observed violence, community reported violence, homicide records, etc.), and study design (experimental, case-control, cross-sectional, and longitudinal) were evaluated as potential moderators. Method and Results: An extensive literature search resulted in 32 meta-analyses that met our inclusion criteria (i.e., quantitatively synthesized research assessing the link between alcohol or illicit drug use and violence perpetration/victimization) demonstrated a signif- icant relationship between substance use and violence (grand weighted mean effect size of d � 0.45, 95% CI [0.36, 0.54], p � .001). Male gender, psychotic illness, experimental study design, and combined alcohol and illicit drug use increased the associations between substance use and violence. More important, the degree of association was similar across the types of violence assessed and roles (victim/perpetrator). Conclusions: The current study shows that with respect to alcohol, illicit drugs, and violence, the overall relationship is a medium effect size that is robust across different populations, substances, types of violence, and both perpetration and victimization.

Keywords: violence, aggression, alcohol, drugs, meta-analysis

The link between substance use and violence is well estab- lished. According to the National Crime Victimization Survey, there were more than 750,000 violent crimes in the United States in 2007 during which the offender was perceived to be under the influence of alcohol or drugs at the time of the offense (Bureau of Justice Statistics, 2008). Substance-related violent crime represents an enormous burden on society with the eco- nomic toll of violent crimes involving alcohol and drugs esti-

mated to be in excess of 120 billion dollars per year in the United States alone (Miller, Levy, Cohen, & Cox, 2006). More- over, the phenomenon of alcohol and drug-related violence is not unique to the United States. In a recent meta-analysis of homicide offenders from nine different countries it was found that 48% of offenders had alcohol in their systems at the time of the incident, that this estimate was largely consistent across different geographic regions, and this finding demonstrates that alcohol has often been consumed by the perpetrator before one of the most severe classification of violence offenses— homi- cide (Kuhns, Exum, Clodfelter, & Bottia, 2014). The purpose of this project was to provide a comprehensive overview and quantitative synthesis of the links between alcohol, drugs, and violence with established meta-meta-analysis methodology, given the recent substantial growth of this research area.

As researchers continue to examine the role of alcohol and drugs with respect to violence, it has become increasing difficult to stay abreast of the large and expansive research literature on this important topic. Meta-analyses have been enormously helpful in this regard. In 1985, Steele and Southwick (1985) published the first meta-analysis assessing the impact of human alcohol con- sumption on social behaviors including aggression and conflict. Results of this early meta-analysis showed that alcohol was related to increased conflict and that this effect was dose-dependent. Since that time, the research on the link between alcohol, illicit drug use, and violence has expanded at a rapid pace. Figure 1 shows that the number of new academic publications on alcohol, illicit drugs, and violence has increased exponentially over the past 50 years, with over 1,000 publications per year since 2010.

This article was published Online First April 10, 2017. Aaron A. Duke, Department of Psychiatry, Yale University School of

Medicine; Kathryn M. Z. Smith, Department of Psychiatry, Yale Univer- sity School of Medicine, and Department of Psychology, Wayne State University; Lindsay M. S. Oberleitner, Department of Psychiatry, Yale University School of Medicine; Alexander Westphal, Department of Psy- chiatry and Child Study Center, Yale University School of Medicine; Sherry A. McKee, Department of Psychiatry and Women’s Health Research at Yale, Yale University School of Medicine.

Aaron A. Duke is now at Optimus, Alexandria, Virginia. Kathryn M. Z. Smith is now at the Department of Psychiatry, Columbia University Medical Center/New York State Psychiatric Institute.

This work was supported by the State of Connecticut, Department of Mental Health and Addiction Services, and the Substance Abuse and Mental Health Services Administration (Grant TI026330; PI: SM). The authors have no conflicts of interest related to the current work to disclose.

Correspondence concerning this article should be addressed to Lindsay M. S. Oberleitner, Department of Psychiatry, Yale University School of Medicine, 1 Long Wharf, Suite 7, New Haven, CT 06511. E-mail: [email protected]

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

Psychology of Violence © 2017 American Psychological Association 2018, Vol. 8, No. 2, 238–249 2152-0828/18/$12.00 http://dx.doi.org/10.1037/vio0000106

238

Despite the growth in literature, our understanding of the overall relationship of substance use and violence remains limited. The early meta-analysis by Steele and Southwick (1985) was narrow in focus, including only laboratory based studies of alcohol use and violence. The focus for meta-analyses in the area of substance use and violence has become narrower and the findings more difficult to generalize in recent decades. The narrower focus of meta- analyses has been because of both: (a) the scientific merit of gaining a greater depth of understanding of the relationship of violence and substance use in specific populations (i.e., for the purpose of the development of specific treatment or policy initia- tives for specific groups such as intimate partner violence only); and (b) the exceedingly high growth of the number of publications which prohibits the ability to analyze all individual studies in a single meta-analytic structure because there have been over 5,000 publications on this topic in general between 2010 and 2016 (see Figure 1). While the variety of research on substance use and violence makes the advantages of meta-analytic review apparent, the sheer volume and scope of the existing research literature makes a comprehensive meta-analysis a particularly arduous and cumbersome task. Because of these multiple reasons, some of the most recent meta-analyses have been conducted on focused pop- ulations (e.g., violence in inpatient settings; Dack, Ross, Papado- poulous, Stewart, & Bowers, 2013), or specific forms of violence (e.g., intimate partner violence victimization in women, DeVries et al., 2014; homicide offenses, Kuhns et al., 2014) in regards to the relationship of substances and violence. As one can see in review- ing these examples of recent meta-analyses, despite the vast body of literature that is capable of supporting more closely defined aspects of the relationship of substance use and violence published in recent years, this has also led us to have little sense of the consistency in patterns of interrelationships of alcohol, illicit

drugs, violence, and victimization when looking from a broader scope.

Although these more targeted meta-analyses are incredibly valu- able in addressing narrowly defined questions, the utility of having a comprehensive overview has necessarily been sacrificed as a result of the field’s increased specificity. The growth of narrowly focused research studies has provided valuable insight into very specified behaviors, however, they leave us with many unan- swered questions such as how lab-based violence compares with homicide in the community, or the similarity in relationship of substances in violence perpetration and violence victimization. The purpose of this study is to determine the overall magnitude of the relationship between alcohol, illicit drug use, and violence across these various studies, and whether there are variations in these effects when comparing key differences in study populations (e.g., men vs. women, perpetrator vs. victim, age, etc.) and meth- odology (e.g., alcohol vs. drug; community violence vs. laboratory violence, etc.).

Current Study

The aim of the current project was to provide a comprehensive overview and quantitative synthesis of the links between substance use and violence. To accomplish this, we conducted a meta-meta- analysis (i.e., the use of meta-analytic techniques to review exist- ing meta-analyses) focused on the relationship between alcohol/ illicit drug and violence. This approach has many advantages including the ability to: (a) to assess variability between meta- analyses; (b) summarize very large research literatures in a parsi- monious fashion; and (c) compare across meta-analyses with dif- ferent populations (e.g., men vs. women, psychotic diagnosis vs. no psychotic diagnosis), operationalizations of violence (e.g., lab-

Figure 1. Number of articles published on substance use and violence per year. Data from Thomson Reuter’s Web of Science’s Citation Report feature, March, 2014.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

239ALCOHOL, DRUGS, & VIOLENCE: META-META-ANALYSIS

oratory observed violence, community reported violence, homi- cide records, etc.), inclusion/exclusion criteria, and methodologies (e.g., alcohol vs. drug vs. both alcohol and drug).

The use of meta-meta-analytic techniques are established and have been helpful in integrating other large bodies of research such as psychotherapy outcomes (Grissom, 1996), the role of choles- terol reduction in coronary heart disease (Katerndahl & Lawler, 1999), behavioral treatment approaches (Hofmann, Asnaani, Vonk, Sawyer, & Fang, 2012; Lipsey & Wilson, 2001), and the utility of correctional rehabilitation (Lipsey & Cullen, 2007). There is a clear need to find a way to integrate and analyze the many recent, but narrowly focused research studies in this area, and the meta-meta-analytic approach is one established pathway to understanding the consistency of these narrowly focused, but closely related topical areas.

In addition to our primary research question of the overall relationship of substance use and violence we were interested in whether the relationship is moderated by sample and study-level variables. Based on availability of data across multiple meta- analyses, we were able to evaluate the factors of age (adolescent/ young adult vs. adult), gender (as men are often accepted to be more violent though individual research studies have had conflict- ing findings), the presence of psychotic illness (as psychiatric diagnoses are often implicated by the news, the community, etc.), role (perpetrator vs. victim; as the relationship with violent vic- timization is not as widely understood), substance (alcohol, drugs, both; as the role of substances other than alcohol, and the com- bined effect of alcohol and substance use is less widely understood than alcohol alone), measurement or definition of violence used (criminal offending, intimate partner violence, and aggression in laboratory tasks), and study design (experimental, case-control, cross-sectional, and longitudinal). The moderator variables chosen were limited by the variables available in the meta-analyses in- cluded in this study. We were also interested in whether charac- teristics specific to the meta-analyses themselves account for dif- ferences in the magnitude of the relationship between substance use and aggression. These characteristics included year of publi- cation, the size of the meta-analysis, the use of random versus fixed effects, weighting scheme, the inclusion of unpublished research, the use of multiple coders, handling of outliers, and whether substance use and violence were the primary focus of the meta-analysis.

Hypotheses for the present study were as follows:

Hypothesis 1: Substance use and violence have a consistent, positive relationship across the meta-analyses examined which include alcohol and violence perpetration (Hypothesis 1a), illicit drug use and violence perpetration (Hypothesis 1b), and alcohol use and violence victimization (Hypothesis 1c).

Hypothesis 2: The relationship of substance use and violence will remain significant after accounting for characteristics specific to the meta-analyses examined for this study (i.e., number of studies included, year published, random vs. fixed effects, weighting, inclusion of unpublished studies, removal of outliers, use of multiple coders, and whether substance use and violence was the primary focus).

The relationship of substance use and violence exists across a variety of sample specific (i.e., age group, presence of psychotic

illness, gender, alcohol vs. drug, etc.) and study specific (i.e., operationalized measure of violence, perpetration vs. victimiza- tion, and included study designs) variations.

Method

Search Strategy

A comprehensive search of meta-analyses involving the relation between substance use and violence was conducted using the following electronic databases: PsycINFO, PubMed, EMBASE, CINAHL, National Criminal Justice Reference Service, ProQuest Criminal Justice, Criminal Justice Abstracts, Google Scholar, OpenGrey, and Cochrane Database of Systematic Reviews. Titles, keywords, and abstracts were searched in each database from 1970 to February 15, 2014. The following search terms were used across each database with some modification as necessary (e.g., � instead of � in Google Scholar) to reflect differences in search operators and optional limiters: (aggression OR homicid� OR murder� OR hostil� OR offend� OR anger OR violent OR violence OR assault�) AND (ethanol OR alcohol OR intoxic� OR substance OR drug�) AND metaanaly�. Initially, these searches revealed 1,572 results; however, there were a substantial number of duplicate references (n � 813). The screening process for the included meta-analyses is presented as a flowchart in Figure 2.

Inclusion/Exclusion Criteria

To be included, systematic reviews needed to involve a meta- analytic component (i.e., quantitative aggregation of effect sizes) and needed to present at least one aggregated effect involving substance use (alcohol or illicit drugs) and violence (aggression, intimate partner violence, violent crime, etc.). For inclusion, the review could present substances singularly (e.g., cocaine) or as a combined factor (e.g., any substance use). Additionally, the review could use a violence variable of perpetration or victimization.

Meta-analyses were excluded if they involved the effects of legally prescribed drugs on aggressive behavior such as the effi- cacy of mood stabilizers on aggression (Jones et al., 2011; Mercer, Douglass, & Links, 2009) or the effects of medications on overt

Figure 2. Flow diagram describing selection process for included meta- analyses.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

240 DUKE, SMITH, OBERLEITNER, WESTPHAL, AND MCKEE

aggression in children with attention-deficit-hyperactivity disorder (ADHD) or emotional disturbances (Connor, Boone, Steingard, Lopez, & Melloni, 2003; Connor, Glatt, Lopez, Jackson, & Mel- loni, 2002). Other noteworthy systematic reviews that were not included because they did not contain aggregated effect sizes are Marshal’s (2003) review of the effects of alcohol on marital violence, Exum’s (2006) review of experimental findings linking alcohol and aggression, and Sønderlund et al.’s (2014) systematic review of the association between sports participation, alcohol use, and violence.

Coding Procedures

All meta-analyses were initially coded by the second or third authors (K.S. and L.O.) and were recoded by the first author (A.D.). Initial intercoder agreement was high (87.5%, � � 0.871) and reached 100% after initial discrepancies were resolved. Dis- putes were resolved by discussion and consultation with an addi- tional author (S.M.) if necessary. When available, multiple aggre- gate effect sizes were coded for each meta-analysis for: different substances (e.g., alcohol, marijuana, illicit drugs, etc.), gender (i.e., male, female, mixed), methodology (e.g., longitudinal, cross- sectional, experimental, etc.), role (perpetration vs. victimization), population (e.g., clinical vs. community), age range (i.e., adoles- cent, adult, all ages), and operationalizations of violence (e.g., laboratory observed violence, community reported violence, ho- micide records, etc.). Most meta-analyses contributed multiple aggregate effect sizes (median � 3).

Analyses

A variety of meta-analytic effect sizes were reported (e.g., mean standardized differences, correlation coefficients, and odds ratios). We used mean standardized differences as our primary outcome statistic given its ease of interpretation and because it was the most widely used outcome statistic among the meta-analyses included in our review (k � 13). Thus, all effect sizes were transformed into bias-corrected mean standardized differences (i.e., Hedge’s g—not Cohen’s d) following the formulas outlined by Borenstein (2009) and in accordance with recommendations published by the Co- chrane Collaboration (Deeks, Higgins, & Altman, 2011). Follow- ing the practice of Hedges and Olkin (1985), we refer to the bias-corrected mean standardized difference as d and consistent with Cohen’s (1992) descriptions of effect size, we refer to values less than 0.35 as “small,” values between 0.35 and 0.65 as “me- dium,” and values above 0.65 as “large.”

All analyses were conducted in R version 3.0.2 using the meta- for package (Viechtbauer, 2010). Aggregate effect sizes were weighted by the inverse of their variance (Shadish & Haddock, 2009). Mean weighted effect sizes are presented for random- effects models along with estimates of heterogeneity (Q and I2

statistics). Moderator analyses were conducted using a mixed- effects model when there was significant heterogeneity in the overall estimate (Lipsey & Wilson, 2001). Random-effects and mixed-effects models were estimated using restricted maximum- likelihood estimation (REML) with Knapp and Hartung (2003) adjustment to account for uncertainty in the amount of residual heterogeneity.

Moderator analyses were conducted in two phases. First, planned comparisons were carried out to address our primary

research questions. In the second phase, potential moderator vari- ables were tested by way of univariate and multivariate metare- gression analyses. Two multivariate metaregression analyses were conducted to look at the potential influence of meta-analysis level moderators (i.e., number of studies included, year published, ran- dom vs. fixed effects, weighting, inclusion of unpublished studies, removal of outliers, use of multiple coders, and whether substance use and violence was the primary focus) and sample/design-level moderators (i.e., age group, gender, presence of psychotic illness, operationalized measure of violence, perpetration vs. victimiza- tion, and included study designs). The specific variables analyzed are reported in the results section.

Results

Overall, 32 meta-analyses that assessed the relationship between alcohol/drugs and violence met our inclusion criteria and were included in our study, which had a random effects mean weighted effect size of d � 0.45, 95% confidence interval (CI) [0.36, 0.54], p � .001, which is considered a medium effect size. Studies are listed in Table 1. As predicted, the effect was highly heteroge- neous, I2 � 97.9%, p � .001, necessitating further exploration of potential moderators. The breakdown of meta-analyses by inclu- sion/exclusion criteria are presented in Table 2 along with infor- mation on literature search strategies (e.g., which databases were used), and methodological information about the meta-analyses themselves (e.g., weighting, coding, etc.).

Alcohol and Violence Perpetration—Hypothesis 1a

We identified 18 meta-analyses that assessed the relationship between alcohol use and violence perpetration (see Figure 3). The overall random-effects mean standardized difference for alcohol’s effect on violence perpetration was significant, d � 0.39, 95% CI [0.32, 0.47], p � .001 (medium effect), and heterogeneous, I2 � 92.6%, p � .001. Three meta-analyses included effect sizes for female perpetrators and provided an overall estimate of d � 0.23, 95% CI [0.13, 0.33], p � .001 (small effect), which was not significant for heterogeneity, I2 � 0%, p � .61. Eight meta-analyses included effect sizes for male perpetrators with a combined effect size of d � 0.48, 95% CI [0.45, 0.50], p � .001 (medium effect), which was significantly heterogeneous, I2 � 53%, p � .04. Finally, 11 meta-analyses provided mixed-gender effect sizes with an overall effect of d � 0.38, 95% CI [0.26, 0.50], p � .001 (medium effect), which was also heterogeneous, I2 � 90%, p � .001. Note that Lipsey, Wilson, Cohen, and Derzon (1997) actually includes two sep- arate meta-analyses, one focused on experimental laboratory paradigms and the other focused on cross-sectional studies, both of which provided mixed-gender estimates of alcohol’s impact on violence perpetration. Planned comparisons revealed that the overall effect of alcohol on violence perpetration was greater for men than for women, d � 0.24, 95% CI [0.05, 0.42], p � .05 (small effect) but did not differ between mixed-gender samples and men, d � 0.08, 95% CI [�0.04, 0.19], p � .20, or between mixed-gender samples and women, d � 0.16, 95% CI [�0.02, 0.34], p � .08.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

241ALCOHOL, DRUGS, & VIOLENCE: META-META-ANALYSIS

Illicit Drugs and Violence Perpetration—Hypothesis 1b

In total, 6 meta-analyses were identified that assessed the rela- tionship between illicit drug use and violence perpetration (see Figure 4). An additional 8 meta-analyses assessed the relationship between combined alcohol and drug use and violence perpetration. When aggregated together, the overall effect size for these two categories of meta-analytic findings was d � 0.49, 95% CI [0.34, 0.64], p � .001 (medium effect), which was significant for heter- ogeneity, I2 � 96.58%, p � .001. The pooled mean standardized difference for combined alcohol/illicit drug use was d � 0.55, 95% CI [0.28, 0.83], p � .001 (medium effect), and the pooled estimate for meta-analyses reporting the impact of illicit drug use alone on violence was d � 0.42, 95% CI [0.32, 0.53], p � .001 (medium effect). Both of these pooled estimates were also significant for heterogeneity, I2 � 96.07%, p � .001 and I2 � 96.58%, p � .001, respectively. With respect to violence perpetration, the mean weighted effect size for alcohol-only was significantly smaller than for combined alcohol and drugs, d � 0.16, 95% CI [0.06, 0.39], p � .05 (small effect). The mean weighted effect of illicit drugs did not differ significant from alcohol only, d � 0.03, 95% CI [�0.13, 0.19], p � .93, nor from combined alcohol/drugs, d � 0.13, 95% CI [�0.03, 0.35], p � .16.

Alcohol Use and Violence Victimization—Hypothesis 1c

With respect to victimization, we identified 5 meta-analyses assessing the link between alcohol use and violence victimization (see Figure 5). The weighted mean effect size was d � 0.42, 95% CI [0.19, 0.64], p � .001 (medium effect), which was significant for heterogeneity, I2 � 99.12%, p � .001. The difference between alcohol-related violence perpetration and alcohol-related violence victimization was not statistically significant, d � 0.03, 95% CI [�0.06, 0.09], p � .05.

Moderator Analyses—Hypotheses 2 and 3

Results from the metaregression analyses are presented in Table 3. Hypothesis 2 was examined in the first set of metaregression analyses, and meta-analysis–level variables were assessed includ- ing: year of publication, number of studies, type of pooling (ran- dom vs. fixed), type of weighting (inverse vs. other), whether unpublished studies were included (yes vs. no), whether outliers were removed, number of coders, and whether alcohol/drugs and violence was the primary focus of the meta-analysis. In support of the second hypothesis, none of these variables were significantly related to mean effect size.

Table 1 Basic Informational About Meta-Analyses Included in Meta-Meta-Analysis

Study Cit. Focus Years

Devries et al., 2014 1 IPV victimization and women’s alcohol use �2013 Gupta et al., 2013 2 Psychosis and substance use 1967–2012 Gutierrez et al., 2013 6 Recidivism in Aboriginal offenders 1988–2010 Witt et al., 2013 11 Risk factors for violence in psychosis 1960–2011 Dack et al., 2013 1 Risk factors of psychiatric in-patient aggression 1960–2009 Kuhns et al., 2014a 2 Alcohol-involved homicide offending 1954–2010 Rothman et al., 2012 16 Alcohol and dating violence perpetration 1985–2010 Large & Nielssen, 2011 40 Violence in first-episode psychosis 1960–2010 Kuhns et al., 2011a 10 Toxicology findings among homicide victims 1950–2009 Collins, 2010 15 Effect of gender on violent recidivism 1970–2008 Douglas et al., 2009 137 Psychosis and violence to others �2006 Fazel et al., 2009 238 Schizophrenia and violence 1970–2009 Kuhns et al., 2009 19 Toxicology findings among homicide victims 1950–2007 Foran & O’Leary, 2008 153 Alcohol and intimate partner violence 1980–2006 Moore et al., 2008 101 Drug abuse and intimate partner aggression 1967–2005 Bunde & Suls, 2006 49 Hostility and coronary artery disease risk factors 1966–2004 Gil-Gonzalez et al., 2006 52 Alcohol and male-perpetrated intimate partner violence 1966–2004 Stith et al., 2004 354 Risk factors of male-perpetrated physical IPV 1980–2000 Rosenfeld, 2004 87 Violence risk factors in stalking and harassment 1990–2002a

Corrao et al., 2004 565 Alcohol and risk of 15 diseases 1966–1998 Derzon, 2001 35 Antisocial behavior and the prediction of violence �2001a

McKay & Schare, 1999 35 Alcohol’s effects on subjective reports 1972–1998 Golding, 1999 1,117 Intimate partner violence and mental disorders �1996a

Bonta et al., 1998 805 Violent recidivism of mentally disordered offenders 1959–1995 Bushman, 1997 234 Effects of alcohol on human aggression 1992–1995a

Lipsey et al., 1997 203 Alcohol use and violence �1995a

Lipsey et al., 1997 203 Alcohol use and violence �1995a

Ito et al., 1996 286 Effects of alcohol on human aggression 1966–1994 Bushman, 1993 103 Effects of alcohol/drugs on human aggression 1974–1993 Bushman, 1997 626 Effects of alcohol on male aggression 1972–1989 Hull & Bond, 1986 447 Social effects of alcohol consumption/expectancy 1972–1985 Steele & Southwick, 1985 291 Alcohol and social behavior 1967–1985

Note. Cit. � Number of times cited according to Google Scholar as of March 10, 2014. a The two studies by Kuhn et al. reported proportions of homicide offenders/victims intoxicated at the time of the incident and did not provide a comparison group from which to calculate an effect size, therefore, these studies did not contribute to the overall effect size estimates.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

242 DUKE, SMITH, OBERLEITNER, WESTPHAL, AND MCKEE

Hypothesis 3 was examined in a second set of regression anal- yses, in which sample-level variables were tested including: age group, the presence of psychotic illness, gender, role (perpetrator vs. victim), substance (alcohol vs. drugs, and combined vs. drugs), measure of violence (criminal offending/recidivism, intimate part- ner violence, and laboratory aggression paradigm contrasted with mixed/other), and study design (randomized experiment, case- control, and cross-sectional contrasted with mixed/other study design). In the univariate analyses, the only significant predictor of mean effect size was the presence of psychotic illness, which was linked to a stronger relationship between alcohol/drugs and vio- lence, b � 0.23, 95% CI [0.04, 0.43], p � .05. The multivariate analysis revealed a different pattern of findings with two signifi- cant predictors: all-male sample (b � 0.32, 95% CI [0.01, 0.63], p � .05) and the use of randomized experimental design (b � 0.52, 95% CI [0.02, 1.03], p � .05), both of which were linked to larger mean effects.

Discussion

We identified 32 meta-analyses spanning nearly 30 years that included information on the aggregate effect of alcohol and illicit drug use on violence published between 1985 and 2014. Our first hypothesis was supported, with every meta-analysis demonstrated a positive relationship between substance use and violence, with the vast majority indicating statistically significant mean effects. Overall, the relationship between substance use and violence ap- pears to be remarkably robust and can be classified as a “medium”

effect with the average impact of substances leading to an increase of about half a SD in overall violence, d � .45. The findings of this study supported the second hypothesis, that the relationship of substance use and violence would remain significant after account- ing for characteristics of the meta-analyses included in the study. The overall effect of the relationship of substance use and violence was not significantly effected by the inclusion of any characteris- tics of the meta-analyses included in this meta-meta (e.g., year of publication, use of multiple coders, etc.).

The findings of this study also supported the third hypothesis that the relationship of substance use and violence exists across a variety of sample and study specific variations. In character- istics specific to the meta-analyses included in the present metameta, it was found that the relationship of substance use and violence was consistent across operalizations of violence being examined including community reported violence, violent criminal recidivism, or laboratory-based measures of aggres- sion. Moreover, the relationship appears to be bidirectional, at least with alcohol, in that we found the relationship between alcohol and violence perpetration to be almost identical to that of alcohol and violence victimization, although the latter rela- tionship has been studied in much less detail. Our findings also highlight that the combination of alcohol and drugs may be more detrimental than either alcohol alone given that meta- analyses which looked at combined alcohol and drug use showed a significantly stronger relationship with violence than meta-analytic results just looking at alcohol.

Table 2 Breakdown of Meta-Analysis Characteristics and Reporting Rates

Sample n % Literature search n % Analyses n %

Age Databases included Aggregation of within-study effects Unrestricted 12 37.5 PsycINFO 25 78.1 Simple average 11 34.3 Adults 7 21.9 MEDLINE 19 59.4 Randomly select one 4 12.5 11–21 1 3.1 EMBASE 7 21.9 Largest effect 1 3.1 Not reported 12 37.5 NCJRS 6 18.7 Multivariate technique 2 6.2

Gender Criminal justice abstracts 5 15.6 Not reported 14 43.8 Men 2 6.2 Google scholar 4 12.5 Weighting of effects Women 2 6.2 ERIC 4 12.5 Inverse of their variance 17 53.1 Men and women 7 21.9 Current contents 4 12.5 By sample size 3 9.4 Mixed gender 21 65.6 Dissertation abstracts 4 12.5 Unweighted 2 6.2

Substance CINAHL 3 9.4 Not reported 10 31.2 Alcohol 17 53.1 Not reported 5 15.6 Pooling Illicit drugs 3 9.3 Unpublished studies Fixed effects 3 9.4 Alcohol and illicit drugs 5 15.6 Yes 15 46.9 Random effects 16 50.0 Combined 7 21.9 No 13 40.6 Not reported 13 40.6

Role Not reported 4 12.5 Bias Perpetrator 27 84.4 Search strategy Outlier removal 5 15.6 Victim 5 15.6 Explicit search terms 27 84.4 Vote counting 3 9.4

Violence Initial search results 17 53.1 Failsafe N 5 15.6 Criminal offending 11 34.4 Final number of studies 30 93.8 Funnel plot 6 18.7 Laboratory aggression 6 18.8 Contacted authors for data 8 25.0 Coding Intimate partner violence 8 25.0 Used reference sections 22 68.8 Multiple coders 19 59.4 Mixed/other 7 21.9 Searched specific journals 8 25.0 Procedure for discrepancies 13 40.6

Research designs Language Inter-coder agreement 11 34.3 Experiments 8 25.0 English only 12 37.5 Effect size statistic Cross-sectional 7 21.9 Unrestricted 5 15.6 Cohen’s d or Hedge’s g 13 40.6 Case-control 4 12.5 Not reported 15 46.9 Odds ratio 8 40.6 Longitudinal 1 3.1 Mental illness Correlation coefficient 6 18.7 Multiple/other 12 37.5 Psychotic disorder 7 21.9 Multiple/other 6 18.7

Note. n � number of meta-analyses reporting variable. % � percentage of the studies. NCJRS � National Criminal Justice Reference Service.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

243ALCOHOL, DRUGS, & VIOLENCE: META-META-ANALYSIS

Despite the robust relationship of substance use and violence, there were sample and study specific factors that had significant influence on that relationship. More important, we identified that gender and psychotic illness moderated the relationship between alcohol, drugs, and violence. First, the impact of alcohol on vio- lence perpetration was found to be larger for men than for women. The moderating role of gender was also significant in a multivar- iate metaregression analysis with all substances included. There is a growing literature on the gender differences in the relationship of alcohol/drugs and violence. The pattern of gender differences have varied greatly depending upon the population (e.g., psychiatric disorders vs. none), operationalizations of violence (e.g., labora- tory observed violence, community reported violence, homicide records, etc.), and the setting (e.g., incarcerated vs. not; laboratory vs. community). Therefore, despite our overall findings, gender differences in the relationship of alcohol/drugs and violence re- mains one topic area in which a more detailed analysis of moder- ating factors is warranted.

The second sample specific moderating variable that we iden- tified was the presence of psychotic illness. It is worth highlighting

that while the relationship between alcohol/drugs and violence appears to be stronger for individuals with psychotic illnesses (examples of psychotic disorders include schizophrenia, schizoaf- fective disorder, mood disorders with psychosis, etc.), the vast majority of individuals with psychotic illnesses do not engage in seriously violent behavior (Large & Nielssen, 2011). Additionally, the link between alcohol/drugs and violence in individuals diag- nosed with a psychotic illness may be partially driven by increased rates of alcohol and illicit drug use documented within this pop- ulation (Elbogen & Johnson, 2009; see also Van Dorn, Volavka, & Johnson, 2012). Unfortunately, the same attention paid toward those with psychotic illnesses with respect to substances and violence has not been directed toward those with other mental illnesses; therefore, we cannot speak to whether these results are reflective of mental illness more generally.

An interesting study specific variation finding was that experi- mental studies of substance use and aggression were found to be linked to slightly larger outcomes than nonexperimental studies when analyzed in multivariate metaregression controlling for sam- ple characteristics. One advantage of experimental studies is that

Figure 3. Forest plot, estimates of the effect of alcohol on violence perpetration.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

244 DUKE, SMITH, OBERLEITNER, WESTPHAL, AND MCKEE

we know individuals are actually under the influence of the sub- stance in question, while in community samples, there is much more variability in the link between substance use and violence both temporally and from a dose-response standpoint. That is to say that in community based samples, the recorded alcohol or other substance use may have been used acutely as related to the re- corded offense, or it may reflect a more chronic use (e.g., if urine drug screens were the operationalized definition of substance use prior meta-analyses it does not guarantee that substance use oc- curred at the time of violence and alternatively may have been in the days prior). Thus, experimental designs afford greater sensi-

tivity (i.e., observation of the substance consumption and control over what is contained in that substance) and specificity (i.e., exact timing in reference to the recorded violence) in identifying effects of substances on violent and aggressive outcomes.

Limitations and Future Directions

Meta-meta-analytic approaches provide a parsimonious strategy for reviewing a large amount of research; however, they are in turn, less focused on narrow aspects of a relationship. As noted by Kazrin, Durac, and Agteros (1979, p. 398), “the onerous effort of

Figure 4. Forest plot, estimates of the effect of substance use (mixed alcohol and illicit drugs) and illicit drug use on violence perpetration.

Figure 5. Forest plot, estimates of the effect of alcohol use on violence victimization.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

245ALCOHOL, DRUGS, & VIOLENCE: META-META-ANALYSIS

reading individual studies and the taxing task of evaluating their methodology are obviated by meta-meta-analysis.” In other words, this approach provides a 10,000 foot view of the relationship between alcohol, illicit drugs, and violence, a much needed per- spective as the science in many areas becomes increasingly nar- row. This wide-lens approach also provides key information needed for policy decisions. However, it is important to note that the subtle details that remain critical to understanding the incred- ibly complex relationship between substance use and violence are equally necessary.

As noted by others (Munder, Brutsch, Leonhart, Gerger, & Barth, 2013) another issue to raise is that meta-analytic studies included in the current investigation may include overlap in the individual studies reviewed. In this regard the meta-meta-analytic approach is akin to creating a simple sampling distribution of means. In a sampling distribution, a sample is taken from the entire population, a mean is calculated from this sample and plotted on the sampling distribution. When enough sample means are plotted, there is a normal distribution of sampling means with the distri- bution approximating that of the population. With the meta-meta- analysis, the entire population of studies examining substance use and violence are sampled, sample means are plotted (i.e., prior individual meta-analytic studies), and a sampling distribution (i.e., results of meta-meta-analysis) is created in which the distribution

approximates the population. In both examples, overlap in indi- vidual population level data points which create a sampling mean is possible and expected. While some prior meta-meta-analysis (e.g., Munder et al., 2013) have attempted to correct for this overlap, it is unclear how this may have changed potential find- ings.

Additionally, as with any meta-analytic review, the analysis is limited by the existent literature. A well-known limitation of existent literature is that most nonsignificant findings are not published. We reviewed the gray literature to counteract the pub- lication bias, however, this bias is inherent to the meta-analytic approach. We found that all meta-analyses included in these anal- yses had a positive relationship between alcohol/other substances and violence, and there is always the possibility that other unpub- lished works found no relationship between these variables.

Another limiting factor of meta-analytic procedures are that the potential moderatos examined are limited by the variables existing in the existent literature. Future research studies and meta-analyses should examine factors such as the relationship between perpetra- tor and victim, type of violence (e.g., differentiating between physical, verbal, and psychological forms of violence), individual factors (e.g., baseline anger, psychiatric diagnoses other than psy- chosis, etc.) situational characteristics (e.g., stressors/triggers for aggression, community based risks, etc.). Additionally, further research in the field should aim to address gaps in the literature that were identified through this comprehensive analysis, includ- ing: substance use and homicides, the role of individual drug categories in violence perpetration and victimization, and the role of substance use and violence in women. Future research should examine the aspects of study design that may contribute to the stronger finding for laboratory based studies versus community samples (e.g., acute vs. chronic use; inclusion of individuals with higher aggression or violence histories; more clearly defined con- trol groups). Further research should also aim to better understand the association between psychotic symptoms, substance use and violence, as well as substance use and victimization, as these links have not been fully elucidated.

Research Implications

The meta-meta approach provides an invaluable step back from the recent literature and allows analysis of the patterns and con- ditions under which a given relationship exists. The tendency for meta-analyses in a particular field to become increasingly nar- rower in focus over time (as a result of both necessity because of the size of existent literature and focused research and clinical needs) directly hinders the ability to speak to the big picture, which remains of interest to researchers and policymakers alike. The meta-meta approach speaks to the “big picture” in literature areas where comprehensive meta-analyses are increasingly intractable to conduct. The current study shows that with respect to alcohol, illicit drugs, and violence, the big picture is a medium effect size that is robust across different populations, substances, method of operationalizing violence (e.g., lab-based studies, community vi- olence, and intimate partner violence), and both perpetration and victimization. Although the relationship between alcohol and per- petration of violence has been established in the literature, the current finding that a similar pattern exists for drugs and violence, and for the relationship of alcohol and drugs to victimization is

Table 3 Test of Weighted Regression Models for Potential Moderators

Moderator

Univariate effects

b [95% CI]

Simultaneous effects

b [95% CI]

Meta-analytical moderators Mean effect .42 [.12, .71]��

Number of studies �.05 [�.15, .04] .00 [�.01, .01] Year published .00 [�.09, .09] �.02 [�.04, .01] Random effects .10 [�.09, .28] .22 [�.15, .59] Inverse weighting �.14 [�.32, .04] �.02 [�.27, .22] Unpublished studies included �.09 [�.27, .09] �.05 [�.31, .21] Outliers removed �.17 [�.39, .05] �.09 [�.41, .23] Multiple coders .06 [�.12, .25] .05 [�.18, .27] Primary focus �.12 [�.31, .06] �.17 [�.50, .16]

Sample/design moderators Mean effect .20 [�.50, .90] Psychotic illness .23 [.04, .43]� .28 [�.12, .68] Female (vs. mixed gender) .11 [�.14, .36] .37 [�.02, .76] Male (vs. mixed gender) .13 [�.08, .34] .32 [.01, .63]�

Victimization (vs. perpetration) .12 [�.33, .57] .10 [�.46, .65] Alcohol (vs. drugs) .09 [�.46, .63] �.33 [�1.03, .37] Combined substances (vs. drugs) .28 [�.26, .83] .03 [�.63, .68] Criminal offending/recidivisma .20 [�.05, .45] .21 [�.14, .56] Intimate partner violencea .03 [�.22, .28] .07 [�.37, .51] Laboratory aggressiona �.02 [�.30, .25] �.24 [�.64, .17] Experimental (vs. mixed design) �.09 [�.30, .13] .52 [.02, 1.03]�

Case-control (vs. mixed design) .04 [�.25, .34] �.11 [�.49, .26] Cross-sectional (vs. mixed design) �.19 [�.48, .10] .21 [�.18, .60]

Note. Summary statistics for simultaneous test of meta-analytical mod- erators: QM(8) � 5.95, p � .67, QE(32) � 546.58, p � .001. Summary statistics for simultaneous test of participant and design moderators: QM(12) � 19.38, p � .08, QE(27) � 302.38, p � .001. QM � test of whether moderators account for a significant amount of the variability in effect sizes; QE � test of residual heterogeneity. a Comparison: Mixed/other measure of violence. � p � .05. �� p � .01.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

246 DUKE, SMITH, OBERLEITNER, WESTPHAL, AND MCKEE

novel for meta-analyses conducted in this area. The difference in effect size found between different methods of operationalizing violence should be examined further in future research. Specifi- cally, laboratory based methods had a stronger effect size than community based evaluations, and it is unclear as to what aspect of study design has led to this difference. Potential questions that should be further explored are: Is it the acute exposure to sub- stances in the laboratory which cannot be controlled in community based samples that leads to a stronger effect? Is it the increased control and increased specificity of the measurement of aggres- sion/violence in the laboratory that leads to the stronger effect? Does the recruitment of participants with histories of violence/ aggression in laboratory settings make a difference in the relation- ship?

Clinical and Policy Implications

The present findings have demonstrated that alcohol and other substances are closely intertwined with aggression and violence, and that this relationship is robust across different measures of violence and for both alcohol and other drugs. These findings suggest that there is a clear need to provide integrated care for violence prevention and substance use. There have been recent efforts to increase treatment for co-occurring alcohol and perpe- tration of violence (Easton et al., 2007), however, these findings also suggest that in addition to increasing those efforts, we also need to develop, evaluate, and disseminate treatments that are focused on the co-occurrence drugs and violence as well. The robust relationship of alcohol and other drugs to violence also suggests that to decrease violence community-wide we need to continue to increase policy efforts to improve the availability and access to evidence-based care for substance use. Those substance use treatment efforts should include assessment of aggression and violence, as well as focused treatment for violence prevention. Special focus on the development of effective prevention and treatment efforts for the co-occurrence of substance use and vio- lence should be prioritized for individuals who are using both alcohol and other substances, who are diagnosed with a psychotic disorder, and for men, as those groups were identified as having a stronger relationship—though further research should be con- ducted to examine whether those factors are influenced by other individual or situational factors. Further, the relationship of vio- lence to substance use was found to be bidirectional; thus, these findings also suggests that providers who are treating individuals with substance use disorders should also be assessing current risk of violence victimization as a standard of care. Substance use and violence both individually represent serious public health risks, and the finding that they are closely intertwined suggests that there is a clear need to address violence perpetration and risk of victim- ization in all substance use prevention and treatment efforts to improve our communities. Effective, accessible, and affordable substance use treatment is likely one of the key steps in reducing violence in our communities.

References �References Marked With an Asterisk Indicate meta-analyses Included

in the metameta-analysis.

�Bonta, J., Law, M., & Hanson, K. (1998). The prediction of criminal and violent recidivism among mentally disordered offenders: A meta-

analysis. Psychological Bulletin, 123, 123–142. http://dx.doi.org/10 .1037/0033-2909.123.2.1239522681

Borenstein, M. (2009). Effect sizes for continuous data. In H. Cooper, L. V. Hedges, & J. C. Valentine (Eds.), The handbook of research synthesis and meta-analysis (2nd ed., pp. 221–235). New York, NY: Russell Sage Foundation.

�Bunde, J., & Suls, J. (2006). A quantitative analysis of the relationship between the Cook-Medley Hostility Scale and traditional coronary artery disease risk factors. Health Psychology, 25, 493–500. http://dx.doi.org/ 10.1037/0278-6133.25.4.49316846324

Bureau of Justice Statistics. (2008). Criminal victimization in the United States, 2007, Statistical Tables, Table 32.

�Bushman, B. J. (1993). Human aggression while under the influence of alcohol and other drugs: An integrative research review. Current Direc- tions in Psychological Science, 2, 148–151. http://dx.doi.org/10.1111/ 1467-8721.ep10768961

Bushman, B. J. (1997). Effects of alcohol on human aggression: Validity of proposed explanations. In M. Galanter (Ed.), Alcohol and violence: Epidemiology, neurobiology, psychology, family issues (pp. 227–243). New York, NY: Plenum Press.

�Bushman, B. J., & Cooper, H. M. (1990). Effects of alcohol on human aggression: An integrative research review. Psychological Bulletin, 107, 341–354. http://dx.doi.org/10.1037/0033-2909.107.3.341

�Cherpitel, C. J., Ye, Y., & Bond, J. (2005). Attributable risk of injury associated with alcohol use: Cross-national data from the emergency room collaborative alcohol analysis project. American Journal of Public Health, 95, 266–272. http://dx.doi.org/10.2105/AJPH.2003.031179

Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159. http://dx.doi.org/10.1037/0033-2909.112.1.15519565683

�Collins, R. E. (2010). The effect of gender on violent and nonviolent recidivism: A meta-analysis. Journal of Criminal Justice, 38, 675–684. http://dx.doi.org/10.1016/j.jcrimjus.2010.04.041

Connor, D. F., Boone, R. T., Steingard, R. J., Lopez, I. D., & Melloni, R. H., Jr. (2003). Psychopharmacology and Aggression: II. A meta-analysis of nonstimulant medication effects on overt aggression-related behaviors in youth with SED. Journal of Emotional and Behavioral Disorders, 11, 157–168. http://dx.doi.org/10.1177/10634266030110030301

Connor, D. F., Glatt, S. J., Lopez, I. D., Jackson, D., & Melloni, R. H., Jr. (2002). Psychopharmacology and aggression. I: A meta-analysis of stimulant effects on overt/covert aggression-related behaviors in ADHD. Journal of the American Academy of Child & Adolescent Psychiatry, 41, 253–261. http://dx.doi.org/10.1097/00004583-200203000-00004

�Corrao, G., Bagnardi, V., Zambon, A., & La Vecchia, C. (2004). A meta-analysis of alcohol consumption and the risk of 15 diseases. Preventive Medicine, 38, 613–619. http://dx.doi.org/10.1016/j.ypmed .2003.11.027

�Dack, C., Ross, J., Papadopoulos, C., Stewart, D., & Bowers, L. (2013). A review and meta-analysis of the patient factors associated with psy- chiatric in-patient aggression. Acta Psychiatrica Scandinavica, 127, 255–268. http://dx.doi.org/10.1111/acps.12053

Deeks, J. J., Higgins, J. P., & Altman, D. (2011). Chapter 9: Analysing data and undertaking meta-analyses. In J. J. Deeks, J. P. Higgins, & D. Altman (Eds.), Cochrane handbook for systematic reviews of interven- tions (Vol. 5.1.0, updated March 2011). The Cochrane Collaboration. Retrieved from www.handbook.cochrane.org

�Derzon, J. H. (2001). Antisocial behavior and the prediction of violence: A meta-analysis. Psychology in the Schools, 38, 93–106. http://dx.doi .org/10.1002/pits.1002

�Devries, K. M., Child, J. C., Bacchus, L. J., Mak, J., Falder, G., Graham, K., . . . Heise, L. (2014). Intimate partner violence victimization and alcohol consumption in women: A systematic review and meta-analysis. Addiction, 109, 379–391. http://dx.doi.org/10.1111/add.1239324329907

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

247ALCOHOL, DRUGS, & VIOLENCE: META-META-ANALYSIS

�Douglas, K. S., Guy, L. S., & Hart, S. D. (2009). Psychosis as a risk factor for violence to others: A meta-analysis. Psychological Bulletin, 135, 679–706. http://dx.doi.org/10.1037/a0016311

Easton, C. J., Mandel, D. L., Hunkele, K. A., Nich, C., Rounsaville, B. J., & Carroll, K. M. (2007). A cognitive behavioral therapy for alcohol-dependent domestic violence offenders: An integrated substance abuse-domestic vio- lence treatment approach (SADV). The American Journal on Addictions, 16, 24–31. http://dx.doi.org/10.1080/10550490601077809

Elbogen, E. B., & Johnson, S. C. (2009). The intricate link between violence and mental disorder: Results from the National Epidemiologic Survey on Alcohol and Related Conditions. Archives of General Psy- chiatry, 66, 152–161. http://dx.doi.org/10.1001/archgenpsychiatry.2008 .537

Exum, M. L. (2006). Alcohol and aggression: An integration of findings from experimental studies. Journal of Criminal Justice, 34, 131–145. http://dx.doi.org/10.1016/j.jcrimjus.2006.01.008

�Fazel, S., Gulati, G., Linsell, L., Geddes, J. R., & Grann, M. (2009). Schizophrenia and violence: Systematic review and meta-analysis. PLoS Medicine, 6, e1000120. http://dx.doi.org/10.1371/journal.pmed.1000120

�Foran, H. M., & O’Leary, K. D. (2008). Alcohol and intimate partner violence: A meta-analytic review. Clinical Psychology Review, 28, 1222–1234. http://dx.doi.org/10.1016/j.cpr.2008.05.001

�Gil-González, D., Vives-Cases, C., Alvarez-Dardet, C., & Latour-Pérez, J. (2006). Alcohol and intimate partner violence: Do we have enough information to act? European Journal of Public Health, 16, 278–285. http://dx.doi.org/10.1093/eurpub/ckl016

�Golding, J. M. (1999). Intimate partner violence as a risk factor for mental disorders: A meta-analysis. Journal of Family Violence, 14, 99–132. http://dx.doi.org/10.1023/A:1022079418229

Grissom, R. J. (1996). The magical number: 7 �. 2: Meta-meta-analysis of the probability of superior outcome in comparisons involving therapy, placebo, and control. Journal of Consulting and Clinical Psychology, 64, 974–982. http://dx.doi.org/10.1037/0022-006X.64.5.9738916626

�Gupta, P., Mullin, K., Nielssen, O., Harris, A., & Large, M. (2013). Do former substance users with psychosis differ in their symptoms or function from non-substance users? A systematic meta-analysis. Austra- lian and New Zealand Journal of Psychiatry, 47, 524–537. http://dx.doi .org/10.1177/0004867412474071

�Gutierrez, L., Wilson, H. A., Rugge, T., & Bonta, J. (2013). The predic- tion of recidivism with Aboriginal offenders: A theoretically informed meta-analysis. Canadian Journal of Criminology and Criminal Justice, 55, 55–99. http://dx.doi.org/10.3138/cjccj.2011.E.51

Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. Orlando, FL: Academic Press.

Hofmann, S. G., Asnaani, A., Vonk, I. J. J., Sawyer, A. T., & Fang, A. (2012). The efficacy of cognitive behavioral therapy: A review of meta-analyses. Cognitive Therapy and Research, 36, 427–440. http://dx .doi.org/10.1007/s10608-012-9476-1

�Hull, J. G., & Bond, C. F. (1986). Social and behavioral consequences of alcohol consumption and expectancy: A meta-analysis. Psychological Bulletin, 99, 347–360. http://dx.doi.org/10.1037/0033-2909.99.3 .3473714923

�Ito, T. A., Miller, N., & Pollock, V. E. (1996). Alcohol and aggression: A meta-analysis on the moderating effects of inhibitory cues, triggering events, and self-focused attention. Psychological Bulletin, 120, 60–82. http://dx.doi.org/10.1037/0033-2909.120.1.608711017

Jones, R. M., Arlidge, J., Gillham, R., Reagu, S., van den Bree, M., & Taylor, P. J. (2011). Efficacy of mood stabilisers in the treatment of impulsive or repetitive aggression: Systematic review and meta-analysis. The British Journal of Psychiatry, 198, 93–98. http://dx.doi.org/10.1192/ bjp.bp.110.08303021282779

Katerndahl, D. A., & Lawler, W. R. (1999). Variability in meta-analytic results concerning the value of cholesterol reduction in coronary heart

disease: A meta-meta-analysis. American Journal of Epidemiology, 149, 429–441. http://dx.doi.org/10.1093/oxfordjournals.aje.a009830

Kazrin, A., Durac, J., & Agteros, T. (1979). Meta-meta analysis: A new method for evaluating therapy outcome. Behaviour Research and Therapy, 17, 397–399. http://dx.doi.org/10.1016/0005-7967%2879%2990011- 1486043

Knapp, G., & Hartung, J. (2003). Improved tests for a random effects meta-regression with a single covariate. Statistics in Medicine, 22, 2693–2710. http://dx.doi.org/10.1002/sim.1482

�Kuhns, J. B., Exum, M. L., Clodfelter, T. A., & Bottia, M. C. (2014). The prevalence of alcohol-involved homicide offending: A meta-analytic review. Homicide Studies, 18, 251–270. http://dx.doi.org/10.1177/ 1088767913493629

�Kuhns, J. B., Wilson, D. B., Clodfelter, T. A., Maguire, E. R., & Ain- sworth, S. A. (2011). A meta-analysis of alcohol toxicology study findings among homicide victims. Addiction, 106, 62–72. http://dx.doi .org/10.1111/j.1360-0443.2010.03153.x20955489

�Kuhns, J. B., Wilson, D. B., Maguire, E. R., Ainsworth, S. A., & Clodfelter, T. A. (2009). A meta-analysis of marijuana, cocaine and opiate toxicology study findings among homicide victims. Addiction, 104, 1122–1131. http://dx.doi.org/10.1111/j.1360-0443.2009.02583 .x19438418

�Large, M. M., & Nielssen, O. (2011). Violence in first-episode psychosis: A systematic review and meta-analysis. Schizophrenia Research, 125, 209–220. http://dx.doi.org/10.1016/j.schres.2010.11.02621208783

Lipsey, M. W., & Cullen, F. T. (2007). The effectiveness of correctional rehabilitation: A review of systematic reviews. Annual Review of Law and Social Science, 3, 297–320. http://dx.doi.org/10.1146/annurev .lawsocsci.3.081806.112833

Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousand Oaks, CA: Sage.

Lipsey, M. W., Wilson, D. B., Cohen, M. A., & Derzon, J. H. (1997). Is there a causal relationship between alcohol use and violence? A synthe- sis of evidence. In M. Galanter (Ed.), Alcohol and violence: Epidemi- ology, neurobiology, psychology, family issues (pp. 245–282). New York, NY: Plenum Press.

Marshal, M. P. (2003). For better or for worse? The effects of alcohol use on marital functioning. Clinical Psychology Review, 23, 959–997. http:// dx.doi.org/10.1016/j.cpr.2003.09.00214624823

�McKay, D., & Schare, M. L. (1999). The effects of alcohol and alcohol expectancies on subjective reports and physiological reactivity: A meta- analysis. Addictive Behaviors, 24, 633–647. http://dx.doi.org/10.1016/ S0306-4603%2899%2900021-010574301

Mercer, D., Douglass, A. B., & Links, P. S. (2009). Meta-analyses of mood stabilizers, antidepressants and antipsychotics in the treatment of bor- derline personality disorder: Effectiveness for depression and anger symptoms. Journal of Personality Disorders, 23, 156–174. http://dx.doi .org/10.1521/pedi.2009.23.2.156

Miller, T. R., Levy, D. T., Cohen, M. A., & Cox, K. L. C. (2006). Costs of alcohol and drug-involved crime. Prevention Science, 7, 333–342. http:// dx.doi.org/10.1007/s11121-006-0041-616845591

�Moore, T. M., Stuart, G. L., Meehan, J. C., Rhatigan, D., Hellmuth, J. C., & Keen, S. M. (2008). Drug abuse and aggression between intimate partners: A meta-analytic review. Clinical Psychology Review, 28, 247– 274. http://dx.doi.org/10.1016/j.cpr.2007.05.00317604891

Munder, T., Brütsch, O., Leonhart, R., Gerger, H., & Barth, J. (2013). Researcher allegiance in psychotherapy outcome research: An overview of reviews. Clinical Psychology Review, 33, 501–511. http://dx.doi.org/ 10.1016/j.cpr.2013.02.002

�Rosenfeld, B. (2004). Violence risk factors in stalking and obsessional harassment: A review and preliminary meta-analysis. Criminal Justice and Behavior, 31, 9–36. http://dx.doi.org/10.1177/0093854803259241

�Rothman, E. F., McNaughton Reyes, L., Johnson, R. M., & LaValley, M. (2012). Does the alcohol make them do it? Dating violence perpetration

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

248 DUKE, SMITH, OBERLEITNER, WESTPHAL, AND MCKEE

and drinking among youth. Epidemiologic Reviews, 34, 103–119. http:// dx.doi.org/10.1093/epirev/mxr027

Shadish, W. R., & Haddock, C. K. (2009). Combining estimates of effect size. In H. Cooper, L. V. Hedges, & J. C. Valentine (Eds.), The handbook of research synthesis and meta-analysis (pp. 257–277). New York, NY: Russell Sage Foundation.

Sønderlund, A. L., O’Brien, K., Kremer, P., Rowland, B., De Groot, F., Staiger, P., . . . Miller, P. G. (2014). The association between sports participation, alcohol use and aggression and violence: A systematic review. Journal of Science and Medicine in Sport, 17, 2–7. http://dx.doi .org/10.1016/j.jsams.2013.03.011

�Steele, C. M., & Southwick, L. (1985). Alcohol and social behavior I: The psychology of drunken excess. Journal of Personality and Social Psy- chology, 48, 18–34. http://dx.doi.org/10.1037/0022-3514.48.1.18

�Stith, S. M., Smith, D. B., Penn, C. E., Ward, D. B., & Tritt, D. (2004). Intimate partner physical abuse perpetration and victimization risk fac- tors: A meta-analytic review. Aggression and Violent Behavior, 10, 65–98. http://dx.doi.org/10.1016/j.avb.2003.09.001

Van Dorn, R., Volavka, J., & Johnson, N. (2012). Mental disorder and violence: Is there a relationship beyond substance use? Social Psychiatry and Psychiatric Epidemiology, 47, 487–503. http://dx.doi.org/10.1007/ s00127-011-0356-x

Viechtbauer, W. (2010). Conducting meta-analyses in R with the metafor package. Journal of Statistical Software, 36, 1–48. http://dx.doi.org/10 .18637/jss.v036.i03

�Witt, K., van Dorn, R., & Fazel, S. (2013). Risk factors for violence in psychosis: Systematic review and meta-regression analysis of 110 stud- ies. PLoS ONE, 8, e55942. http://dx.doi.org/10.1371/journal.pone .0055942

Received June 1, 2015 Revision received February 1, 2017

Accepted February 2, 2017 �

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an Ps

yc ho

lo gi

ca l

A ss

oc ia

tio n

or on

e of

its al

lie d

pu bl

is he

rs .

T hi

s ar

tic le

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

249ALCOHOL, DRUGS, & VIOLENCE: META-META-ANALYSIS