REVIEW ARTICLE

Cannabis use among U.S. adolescents in the era of marijuana legalization:
a review of changing use patterns, comorbidity, and health correlates

Christopher J. Hammonda,b,c , Aldorian Chaneya,c, Brian Hendricksona,c and Pravesh Sharmad

aDivision of Child & Adolescent Psychiatry, Johns Hopkins University School of Medicine, Baltimore, MD, USA; bBehavioral
Pharmacology Research Unit, Johns Hopkins University School of Medicine, Baltimore, MD, USA; cDepartment of Psychiatry and
Behavioral Sciences, Johns Hopkins University School of Medicine, Baltimore, MD, USA; dDepartment of Psychiatry, Mayo Clinic
Health System and University of Wisconsin-Eau Claire, Eau Claire, WI, USA

ABSTRACT
Decriminalization, medicalization, and legalization of cannabis use by a majority of U.S. states
over the past 25 years have dramatically shifted societal perceptions and use patterns among
Americans. How marijuana policy changes have affected population-wide health of U.S. youth
and what the downstream public health implications of marijuana legalization are topics of sig-
nificant debate. Cannabis remains the most commonly used federally illicit psychoactive drug by
U.S. adolescents and is the main drug for which U.S. youth present for substance use treatment.
Converging evidence indicates that adolescent-onset cannabis exposure is associated with short-
and possibly long-term impairments in cognition, worse academic/vocational outcomes, and
increased prevalence of psychotic, mood, and addictive disorders. Odds of negative develop-
mental outcomes are increased in youth with early-onset, persistent, high frequency, and high-
potency D-9-THC cannabis use, suggesting dose-dependent relationships. Cannabis use disor-
ders are treatable conditions with clear childhood antecedents that respond to targeted preven-
tion and early intervention strategies. This review indicates that marijuana policy changes have
had mixed effects on U.S. adolescent health including potential benefits from decriminalization
and negative health outcomes evidenced by increases in cannabis-related motor vehicle acci-
dents, emergency department visits, and hospitalizations. Federal and state legislatures should
apply a public health framework and consider the possible downstream effects of marijuana pol-
icy change on paediatric health.

ARTICLE HISTORY
Received 15 September 2019
Accepted 6 January 2020

KEYWORDS
Adolescents; cannabis;
marijuana; legalization;
psychiatric comorbidity;
health correlates

Introduction

Cannabis is a psychoactive drug derived from the
plant species cannabis sativa and cannabis indica. It
contains greater than 500 bioactive chemicals and
more than 80 unique phytocannabinoids that have
distinct and dose-dependent effects in humans,
including D-9-tetra-hydrocannabidol (D-9-THC), the
primary psychoactive constituent, and cannabidiol
(CBD) another major constituent of the plant believed
to have potential medical properties (NCCIH, 2018).
Cannabinoids such as D-9-THC and CBD act cen-
trally and peripherally at receptors that are part of an
endogenous brain system involved in development
and homeostasis called the endocannabinoid system.
Cannabis is used by an estimated 183 million individ-
uals worldwide (WHO, 2019). Use of cannabis is
increasing among individuals living in North
America, in part related to legalization, decriminaliza-
tion, and expansion of availability in United States

(U.S.) and Canadian markets (Findlaw, 2019; WHO,
2019). Within the U.S. specifically, dramatic shifts in
public policy have occurred over the past two deca-
des. In 1996 California became the first U.S. state to
approve legislation allowing for the medical use of
marijuana. Since that time 32 other states and the
District of Columbia have passed laws allowing for
legal use of marijuana for medical purposes and 11
states have passed laws allowing for recreational use
of marijuana (Findlaw, 2019). Accompanying these
policy shifts in the U.S. have been changes in public
opinion, perception regarding the harms of cannabis,
cannabis use patterns, and the prevalence of cannabis
use disorders (CUD) among U.S. adults (Cerda, Wall,
Keyes, Galea, & Hasin, 2012; Hasin et al., 2017).

Over this period, there has also been a significant
expansion in the number and type of cannabis prod-
ucts available for consumption (e.g. plant, oil, edibles,
concentrates) (Hopfer, 2014) and three-fold increase

CONTACT Christopher J. Hammond [email protected] Johns Hopkins Bayview, 5500 Lombard Street, Baltimore, MD 21224, USA
� 2020 Institute of Psychiatry and Johns Hopkins University

INTERNATIONAL REVIEW OF PSYCHIATRY
2020, VOL. 32, NO. 3, 221–234
https://doi.org/10.1080/09540261.2020.1713056

in the average D-9-THC potency in cannabis products
(ElSohly et al., 2016). While cannabis use is perceived
by the average American to have few negative conse-
quences, there is increasing evidence for poorer men-
tal and physical health outcomes, including elevated
risk for serious psychiatric illness with chronic recre-
ational cannabis use, especially when high-potency
cannabis is used (Davis et al., 2016; Dutra, Parish,
Gourdet, Wylie, & Wiley, 2018). The risk for negative
health outcomes related to cannabis use are amplified
when use and progression to chronic use begin dur-
ing adolescence.

Given the dramatic shifts in marijuana legislation
that have occurred in the U.S. over the past 25 years,
it is more important than ever for clinicians and pol-
icy makers to be aware of the effects of adolescent
cannabis use on mental and physical health and to
understand potential down-stream consequences
related to these policy shifts on perceptions, use pat-
terns, and health outcomes of U.S. adolescents.
Knowledge about the impact of marijuana policy on
relevant health indicators in youth may inform cur-
rent and future public health efforts. In this article,
we provide a state of the science review on the effects
of adolescent-onset cannabis use on health and behav-
iour with a focus on cannabis use and CUDs among
U.S. adolescents in the current era of state-based
marijuana legalization. We discuss the epidemiology,
recent trends in use, and examine relationships
between adolescent cannabis use and developmental
outcomes. Using a public health framework, we
review marijuana-related legislation and public policy
changes in the U.S. and examine the implications,
positive and negative, for adolescent health outcomes.

Epidemiology and developmental trajectories

Prevalence and incidence of cannabis use among
U.S. youth

Cannabis is the most commonly used federally illicit
drug among U.S. adolescents and the most common
drug problem reported by U.S. teens presenting for
substance abuse treatment (SAMHSA, 2018, 2019).
Over 1.6 million individuals between the ages of 12
and 17 equating to 6.5% of the U.S. adolescent popu-
lation, and 7.6 million individuals between the ages of
18 and 25 equating to 22.5% of the U.S. young adult
population reported current use of cannabis
(SAMHSA, 2018). Among U.S. high school students,
nearly half of all 12th graders (44%), one third of
10th graders (33%) and one in seven 8th graders
(14%) reported lifetime cannabis use in 2018, with

22%, 17%, and 5.6% of U.S. 12th, 10th, and 8th graders
reporting past month use respectively (Johnston et al.,
2018). Daily cannabis use was reported by 5.8% of
12th graders, 3.4% of 10th graders and 0.7% of 8th

graders. Rates of current cannabis use amongst 12-to-
17 year-olds decreased between 2002 and 2009 and
have remained stable since 2015 (SAMHSA, 2018). In
contrast, rates of current cannabis use amongst young
adults have risen significantly since 2002 (SAMHSA,
2018). At the time of this writing, prevalence rates of
cannabis use in 19–22 year old ‘college age’ young
adults have gradually increased since 2009 and are
approaching their highest levels in the past three dec-
ades (Michigan, 2018). Some subgroups have shown
larger changes in cannabis use compared to others.
For example, while daily cannabis use among college-
enrolled U.S. youth peaked at 5.9% in 2014 and has
plateaued since then, daily use has continued to rise
for non-college enrolled U.S. youth and, at 13.2%, is
currently at an all-time high in the past three decades
(Schulenberg et al., 2018). This gap continues to
widen and is worrisome given the negative health and
functioning impact of recreational cannabis use.

Regarding incidence of cannabis initiation: In 2017,
1.2 million adolescents ages 12-to-17 and 1.3 million
young adults ages 18-to-25 reported first time use of
cannabis in the past year (SAMHSA, 2018). Since
2002, the incidence of cannabis use among U.S. ado-
lescents has remained stable, but has increased among
young adults (SAMHSA, 2018). Among youth, the
prevalence of past year CUD reached a peak in the
early 2000s, declined between 2002–2013, and has
remained stable since 2014. In 2017, 2.2% of adoles-
cents ages 12-to-17 equating to over 0.5 million U.S.
adolescents and 5.2% of young adults ages 18-to-25
equating to 1.8 million U.S. young adults met DSM-5
criteria for CUD in the past year (SAMHSA, 2018).

The types of cannabis products and methods of
administration among U.S. youth have also changed
in the past 25 years. While combustible cannabis
products (e.g. joints, blunts, bongs, water pipes)
remain the most common method of use, youth have
expanded to other types of cannabis products and
ways of using (Hopfer, 2014; Johnston et al., 2018). In
fact, data suggests that most current adolescent can-
nabis users use multiple administration methods to
consume cannabis (Peters, Bae, Barrington-Trimis,
Jarvis, & Leventhal, 2018). A 2015 cross-sectional sur-
vey study of 3177 adolescents found that 21.3% and
10.5% of 10th graders in California have tried edible
and vaporized cannabis products respectively (Peters
et al., 2018). The use of ‘concentrates’ (i.e. dabs, wax,

222 C. J. HAMMOND ET AL.

budder, shatter), cannabinoid products with very high
D-9-THC concentrations, are also increasing in popu-
larity among U.S. youth (Sagar & Gruber, 2018;
Zhang, Zheng, Zeng, & Leischow, 2016).

With the increasing availability of portable elec-
tronic ‘vaporized’ nicotine and cannabis delivery sys-
tems in the U.S., vaping cannabis is becoming more
common among youth. National survey data suggest
that 10% of high school students have vaped cannabis
with 12% and 14% of U.S. 10th and 12th graders
reporting vaping cannabis in the past year (Johnston
et al., 2018; Kowitt et al., 2019). While vaping has
been promoted as a safer alternative to combustible
smoking of tobacco and cannabis products, there
have been recent reports of vaping-related acute lung
disease, seizures, and deaths in young people, includ-
ing those who vape cannabis (Schier, Meiman, &
Layden, 2019). In response to this, the CDC has
recently issued health warnings related to vaping
(CDC, 2019).

In summary, across national surveys, prevalence
and incidence of cannabis use among U.S. teens have
remained stable and among U.S. young adults have
increased over the past decade. During this period,
there have been changes in the type and method of
cannabis product used by U.S. youth. Many questions
remain. Why cannabis use and CUD have not
increased in adolescents as they have in adults is
unclear. Additionally, over the same time period that
cannabis use rates have remained unchanged, alcohol,
combustible tobacco products, and non-marijuana
illicit drug use among U.S. adolescents ages
12–17 years have all decreased and are at or near all-
time low use patterns since data were first collected
in the 1970s (Johnston et al., 2018; SAMHSA, 2018).
Whether this comparative stability is a result of
changing U.S. perceptions and policies related to can-
nabis is unclear.

Individual differences in vulnerability and
different developmental trajectories

Significant individual differences exist in the response
to cannabis during youth (Coffey, Carlin, Lynskey, Li,
& Patton, 2003). Not all youth who try cannabis
develop problems related to its use. Some young peo-
ple experiment with cannabis, use it sporadically for a
limited duration, and experience few negative conse-
quences. Others become chronic cannabis users whose
use persists throughout adolescence and young adult-
hood and contributes to long-term health problems
and impaired functioning. Several studies have

identified different trajectories of cannabis use from
adolescence into young adulthood including a persist-
ent/chronic high use trajectory (�5–10%), a chronic
occasional use trajectory (�5–10%), a sporadic/transi-
ent use trajectory (�35%), and a no/low use trajectory
(�45%) (Lee, Brook, Finch, & Brook, 2018; Swift,
Coffey, Carlin, Degenhardt, & Patton, 2008; Windle &
Wiesner, 2004). Relative to no/low use and sporadic/
transient use trajectory groups, individuals in the per-
sistent/chronic high cannabis use trajectory group
have increased psychiatric problems, delinquency, and
other drug use by young adulthood (Lee et al., 2018;
Swift et al., 2008; Windle & Wiesner, 2004).
Understanding the risk and protective factors related
to cannabis engagement trajectories and developing
risk calculators to identify youth who are at elevated
risk for developing persistent/chronic cannabis use
may improve the precision and cost-effectiveness of
targeted prevention efforts.

A widening treatment gap for adolescent
cannabis use

Most individuals who experience drug-related prob-
lems or who meet criteria for a SUD, including can-
nabis, never receive treatment. In 2017, while
1.1 million U.S. adolescents met criteria for a SUD,
fewer than one in 10 received substance abuse treat-
ment (SAMHSA, 2018). Emerging evidence suggests
that for cannabis, this treatment gap may be widening
during the era of marijuana legalization. National
data on treatment episodes for substance-related dis-
orders indicate that fewer individuals are seeking
treatment for cannabis use/CUD in recent years
(Sahlem, Tomko, Sherman, Gray, & McRae-Clark,
2018; SAMHSA, 2019). For example, data on national
admissions to substance abuse treatment from the
treatment episode dataset (TEDS) showed a decrease
in admissions to adolescent substance abuse treatment
among 12–17-year-olds in general, and a 48%
decrease in the total number of cannabis-related
admissions between 2005 and 2015 (SAMHSA, 2017).
The reduction in treatment seeking rates may be a
result of increased social acceptability of cannabis use,
a reduction of the perceived risk of cannabis use, or a
result of the limited effectiveness of current treat-
ments (Sahlem et al., 2018). If expansion of marijuana
legalization results in increased cannabis engagement
among U.S. youth, this widening treatment gap could
portend a future public health crisis. Educational
campaigns aimed at increasing public awareness about
the harms of cannabis use to adolescents, and

INTERNATIONAL REVIEW OF PSYCHIATRY 223

continued research funding aimed at enhancing can-
nabis cessation outcomes and implementing and dis-
seminating effective cannabis treatments nationally
are needed. As more states change their views on can-
nabis use and revise their laws related to marijuana,
public health educational initiatives will be increas-
ingly important.

Health outcomes related to adolescent
cannabis use

Association with cognitive outcomes

The cognition impairing effects of cannabinoids in
humans are well documented in the scientific literature
in both adults and adolescents (Gorey, Kuhns,
Smaragdi, Kroon, & Cousijn, 2019; Lisdahl, Wright,
Kirchner-Medina, Maple, & Shollenbarger, 2014).
Cannabis-related cognitive deficits may be both global
and domain specific and can be temporally framed as
acute/intoxication-, short-, and long-term effects. Acute
cannabis intoxication is associated with transient mood
alterations that may include euphoria, anxiety, or para-
noia, along with impairments in cognitive function
and sensory processing (Bloomfield et al., 2019;
Hunault et al., 2009). When not intoxicated, adolescent
cannabis users exhibit signs of cognitive impairments
when compared to matched controls in the domains of
attention, memory, executive function (including work-
ing memory and inhibition), visual processing, and
processing speed, along with decreased full-scale IQ
and verbal IQ (Dougherty et al., 2013; Gruber, Sagar,
Dahlgren, Racine, & Lukas, 2012; Jacobus, Bava,
Cohen-Zion, Mahmood, & Tapert, 2009). Many of the
domain-specific cognitive deficits improve with abstin-
ence, but some may persist beyond early abstinence
and represent long-term deficits (Hanson et al., 2010;
Medina et al., 2007; Meier et al., 2012; Volkow et al.,
2016). For example, a study by Hanson and colleagues
found that following 3 weeks of abstinence from canna-
bis that adolescent regular cannabis users showed
improvements in verbal memory and inhibition but
continued to exhibit impaired attention compared to
controls (Hanson et al., 2010). Studies of adolescents
with earlier, heavier, and more persistent cannabis use
patterns generally show larger effect sizes for cognitive
outcomes (Gorey et al., 2019; Gruber et al., 2012;
Meier et al., 2012; Pope et al., 2003). Some studies
indicate that long-term cognitive impairments occur
primarily in early adolescent-onset of cannabis users
(� age of 15 or 16), and that adult- or late adolescent-
onset of cannabis users do not exhibit the same risk
for long-term cognitive impairments (Fontes et al.,

2011; Gruber et al., 2012; Pope et al., 2003). Studies
have also largely shown that adolescent-onset cannabis
exposure produces similar cognitive deficits for both
sexes (Levine, Clemenza, Rynn, & Lieberman, 2017;
Meier et al., 2012; Pope et al., 2003).

Whether the cognitive impairments related to ado-
lescent-onset cannabis exposure are permanent or
resolve with abstinence is unclear. Mixed findings
have been reported in the literature. A longitudinal
study by Meier et al. (2012) found that full-scale IQ
decrements related to adolescent cannabis exposure
persisted into adulthood for individuals who contin-
ued to use cannabis and did not fully remit in early
adolescent-onset users who abstained from cannabis
during adulthood (Meier et al., 2012). Analyses from
other longitudinal studies suggest that some cognitive
deficits including global FSIQ decrements remit with
cessation and that length of abstinence is associated
with greater cognitive recovery (Mokrysz et al., 2016;
Tait, Mackinnon, & Christensen, 2011). Twins studies
discordant for cannabis engagement during adoles-
cence have also reported mixed findings related to
whether cannabis-associated cognitive deficits remit
or persist with abstinence in adulthood after control-
ling for genetic and early-childhood vulnerability
(Jackson et al., 2016; Lyon, Bar, & Panizzon, 2004).

Association with psychiatric symptoms and
mental health outcomes

The associations between adolescent-onset cannabis
use, psychiatric symptoms, and mental health out-
comes are complex (Levine et al., 2017). Cannabis use
and CUD commonly co-occur with psychiatric disor-
ders in youth. Concurrent cannabis use and psychi-
atric symptoms may track together in adolescents
with studies suggesting parallel improvement in both
during successful treatment (Hser et al., 2017; Jacobus
et al., 2017; Moitra, Anderson, & Stein, 2016). Aside
from concurrent relationships, long-term associations
have been described in the literature. Longitudinal
studies generally show that adolescent-onset cannabis
exposure is associated in a dose-dependent way with
an increased prevalence of psychotic, mood and
addictive disorders and worse courses of these disor-
ders into adulthood (Levine et al., 2017).

Psychotic disorders

Among all cannabis-related mental health outcomes,
the evidence is strongest for a relationship between
adolescent-onset cannabis use and psychotic disorders

224 C. J. HAMMOND ET AL.

(D’souza et al., 2016; Moore et al., 2007). Adolescent-
onset cannabis use increases the likelihood of devel-
oping attenuated psychotic symptoms, psychotic like
experiences, and full psychotic disorders by young
adulthood (Fridberg, Vollmer, O’Donnell, & Skosnik,
2011; Hall & Degenhardt, 2008; Moore et al., 2007).
Furthermore, among youth with psychotic disorders,
cannabis use is associated with worse course and
prognosis including greater risk for relapse to psych-
osis, increased hospitalizations, and poorer medication
adherence (Miller et al., 2009; Schoeler et al., 2016).
In parallel with other developmental outcomes, dose-
response relationships exist between cannabis use and
psychosis with early-onset, high frequency (�weekly),
and high-potency (�10% D-9-THC) conveying
greater risk (Di Forti et al., 2014; Hall & Degenhardt,
2015). Additively, early-onset daily use of high-
potency cannabis starting before age 15 poses a five-
to six-fold greater risk for the development of a
psychotic disorder compared to non-use (Di Forti
et al., 2014). Different cannabis-psychosis relation-
ships exist along different time-scales (Wilkinson,
Radhakrishnan, & D’Souza, 2014). Acute cannabis
intoxication can produce transient psychotic symp-
toms that resolve spontaneously without intervention
in some youth. Cannabis use can also lead to a pro-
tracted psychotic symptom course including a transi-
ent persistent psychosis or cannabis-induced
psychosis, which lasts beyond the period of acute
intoxication (Wilkinson, Radhakrishnan, & D’Souza,
2014). Cannabis-induced psychotic disorders may be
a harbinger for long-term risk for chronic psychosis.
A recent longitudinal study from Scotland found that
50–75% of individuals who experienced a cannabis-
induced psychotic disorder went on to develop
schizophrenia or bipolar disorder within 15 years
(Alderson et al., 2017). The link between cannabis
and psychosis is greater in individuals with child mal-
treatment or genetic vulnerabilities (Wilkinson et al.,
2014). Some but not all of the risk for developing
psychosis in adolescent cannabis users is related to
pre-existing vulnerability to psychosis. For example,
while adolescent cannabis users with a family history
of psychosis in a first-degree relative have a higher
risk of developing psychosis, cannabis use still con-
veys increased risk in adolescent cannabis users with
no familial risk (D’souza et al., 2016).

Mood and anxiety disorders

Complex relationships exist between cannabis use and
mood and anxiety disorders. Many cannabis using

youth self-report using cannabis to ‘self-medicate’
depression and anxiety, and to reduce stress (Hyman
& Sinha, 2009). When adults are administered canna-
binoids in the laboratory, D-9-THC exerts a biphasic
dose-dependent acute effect on anxiety, with low-dose
D-9-THC preparations being anxiolytic and high-dose
D-9-THC preparations being anxiogenic (Bloomfield
et al., 2019). In direct contradiction to these short-
term effects, the majority of longitudinal studies show
the opposite association, whereby adolescent-onset
cannabis use is associated with increased likelihood
and poorer course of mood and anxiety disorders
during adolescence and into adulthood (Degenhardt,
Hall, & Lynskey, 2003; Levine et al., 2017; Moore
et al., 2007).

Adolescent-onset cannabis use has been linked to
adverse mood outcomes including increased depres-
sion and suicidality in both population-wide cross-
sectional studies and a number of large longitudinal
cohort studies (Degenhardt et al., 2003; Gobbi et al.,
2019; Horwood et al., 2012; Moore et al., 2007; Silins
et al., 2014). It is important to note that while numer-
ous studies have reported a positive association
between adolescent cannabis use and depression,
there are also studies that failed to find this associ-
ation (Scholes-Balog, Hemphill, Evans-Whipp,
Toumbourou, & Patton, 2016). Results from a recent
meta-analysis found that adolescent-onset cannabis
use is associated with a modest increase in the odds
of depression, suicidal ideations, and suicide attempts
in young adulthood (Gobbi et al., 2019). Earlier meta-
analyses have found evidence for a dose-dependent
increase in risk for depression from adolescent canna-
bis use (Lev-Ran et al., 2014) and an age of onset
effect whereby initiation of cannabis before age
17 years conveys greater risk (Moore et al., 2007).
Results from a large retrospective study involving
13,986 twins found that the monozygotic twin who
reported more frequent cannabis use had a two-fold
greater odds of depression and suicidal ideations
compared with their twin counterpart who used less
frequently (Agrawal et al., 2017). This study indicates
that the association between cannabis and depression
cannot be explained by common vulnerabilities alone.
Interestingly, in parallel with cognitive sequelae, most
studies have shown that adolescent-onset cannabis
exposure results in similar mood outcomes for both
males and females (Coffey & Patton, 2016;
Degenhardt et al., 2013; Horwood et al., 2012).

Cross-sectional and longitudinal studies show that
compared to controls adolescent and young adult
cannabis users have increased anxiety, anxiety

INTERNATIONAL REVIEW OF PSYCHIATRY 225

sensitivity, and panic attacks, and increased rates of
co-occurring anxiety disorders including social anx-
iety disorder, panic disorder, and post-traumatic stress
disorders (PTSD) (Buckner et al., 2017; Cornelius
et al., 2010; Zvolensky et al., 2006). A recent meta-
analysis in adults suggests a modest relationship
between cannabis use and anxiety (Kedzior & Laeber,
2014). Findings regarding the directionality of these
associations are mixed. Adverse childhood experiences
and chronic stress generally precede the onset of can-
nabis use and increase the risk for early-onset canna-
bis use and progression to CUD by young adulthood
(Buckner et al., 2012; Hyman & Sinha, 2009). Some
longitudinal studies have shown a long-term relation-
ship between adolescent-cannabis use and increased
risk for anxiety disorders in adulthood (Degenhardt
et al., 2013; Fergusson & Horwood, 1997), while
others have shown no association (Buckner et al.,
2012) or the inverse association (Buckner et al.,
2008). Adolescent cannabis use and CUDs have been
prospectively associated with increased odds for panic
attacks and disorders, but some of the variances in
this relationship may be related to co-occurring
tobacco use (Zvolensky et al., 2008). More recently, a
2019 meta-analysis of longitudinal studies failed to
find an association between adolescent cannabis use
and later anxiety disorders (Gobbi et al., 2019).
Future research is needed in this area, especially given
that many states have added anxiety disorders and
PTSD as ‘indicated conditions’ for medical marijuana
(Mitchell, 2019).

Addictive disorders

As with cognitive, mood, and psychotic outcomes,
adolescent cannabis use is also associated with
increased risk for developing addictive disorders in
adulthood to cannabis and other drugs. Both cross-
sectional and longitudinal studies have shown a sig-
nificant relationship between early-onset cannabis
exposure and a greater likelihood of developing an
addictive disorder (Coffey & Patton, 2016; Fergusson,
Boden, & Horwood, 2015; Nocon, Wittchen, Pfister,
Zimmermann, & Lieb, 2006; E. Silins et al., 2017;
Swift et al., 2008; Swift et al., 2012). Drug engagement
typically follow a general sequence from ‘lite’ drugs
such as alcohol, tobacco, and cannabis to ‘heavier’
illicit drugs including opioids, stimulants. Much
debate has been made about the ‘gateway’ hypothesis,
and reverse gateways going from cannabis to tobacco
use have also been described (Patton, Coffey, Carlin,
Sawyer, & Lynskey, 2005). While part of the variance

in association between early-onset cannabis use and
escalation to other drugs can be explained by the
common liability hypothesis, there is also evidence
suggesting that cannabis’s effects on the developing
brain may also play a role (Hall & Degenhardt, 2015).
In animal models, adolescent-onset cannabinoid
exposure sensitizes striatal dopamine systems and
increases self-administration of opioids and stimulants
in adulthood (Pistis et al., 2004; Renard et al., 2016).
Some studies also suggest unique cannabis-opioid
associations whereby cannabis, more so than other
drugs, increases the risk for progression to opioid
misuse (Kaminer, 2017; Olfson, Wall, Liu, &
Blanco, 2018).

Summary of health outcome findings

In summary, current evidence indicates that adoles-
cent-cannabis exposure is associated with a number of
negative life outcomes including impairments in cogni-
tion and increased prevalence and worse course of
psychotic, mood, and addictive disorders. These associ-
ations are stronger in adolescents with earlier age of
onset, frequent and heavy use, and high-potency can-
nabis use, which is worrisome given the rising potency
and availability of cannabis throughout the U.S. As
findings are associative, causality cannot be deter-
mined. Next-level studies should aim to characterize
associations between adolescent-onset cannabis expos-
ure and health outcomes across different time scales
and at different levels of analysis (i.e. genomics, brain
circuits, behaviour, self-report). Future research …

Characteristics Associated with Attitudes toward Marijuana Legalization in Michigan
Jennifer D. Ellis, M.A.a, Stella M. Resko, Ph.D. b,c, Kathryn Szechy, M.S.W.b, Richard Smith, Ph.D.b,
and Theresa J. Early, Ph.D.d

aDepartment of Psychology, Wayne State University, Detroit, MI, USA; bSchool of Social Work, Wayne State University, Detroit, MI, USA; cMerrill
Palmer Skillman Institute, Wayne State University, Detroit, MI, USA; dCollege of Social Work, The Ohio State University, Columbus, OH, USA

ABSTRACT
Support for legalization of marijuana has increased over the past several years. While studies have
examined correlates of favoring marijuana legalization, less attention has been placed on those
who are unsure about legalization, despite the potentially important role of this group as states
vote on legalization in upcoming elections. Using data from a statewide sample of adults in
Michigan, this study examined whether those who support, oppose, or are unsure about legaliza-
tion differed based on demographics, marijuana use, and perceived risk associated with marijuana
use. Those who were older and perceived marijuana use to pose a greater risk had lower odds of
being unsure about legalization (relative to opposing legalization); those who were politically left
of center or centrist and those who reported lifetime marijuana use had higher odds of being
unsure about legalization. Older respondents, women, and those who perceived marijuana use to
be risky had lower odds of supporting legalization; those who were politically left of center or
centrist, and those who reported recent or lifetime marijuana use had higher odds of supporting
legalization. Better understanding correlates of being unsure about marijuana legalization may
help inform political and prevention efforts as states continue to vote on these issues.

ARTICLE HISTORY
Received 12 June 2018
Accepted 11 March 2019

KEYWORDS
Marijuana; cannabis;
legalization; perspectives

Introduction

Although marijuana is illegal under federal United States
(US) drug policy, state and local marijuana policies have
shifted dramatically over the past two decades (Millhorn
et al. 2009; Nielsen 2010). As of November 2018, 33 states
have legalized medical marijuana use and 13 states, as well
as several local jurisdictions, have decriminalized mari-
juana possession. Additionally, 10 states and the District
of Columbia have legalized recreational marijuana use for
adults over 21 years of age. Michigan became the first
Midwestern state to legalize recreational marijuana when
56% of its voters approved a November 2018 ballot mea-
sure. Other states are expected to vote on medical mar-
ijuana, marijuana decriminalization and recreational
legalization initiatives in future elections (Caulkins,
Kilmer, and Kleiman 2016). As of early 2018, medical
marijuana is supported by an overwhelming majority of
Americans (>90%) (Quinnipiac University/Poll 2018), and
as of October 2018, approximately two in three supported
legalization of recreational marijuana (McCarthy 2017).

Public opinion researchers have examined correlates
of support for recreational marijuana legalization. In
recent studies of adults, male gender (Galston and
Dionne 2013; Looby, Earleywine, and Gieringer 2007;
Musgrave and Wilcox, 2013; Nielsen 2010), minority

racial/ethnic status (Looby, Earleywine, and Gieringer
2007; Nielsen 2010), having children (Caulkins et al.
2012; Cruz, Queirolo, and Boidi 2016; Nielsen 2010),
and being politically liberal (Looby, Earleywine, and
Gieringer 2007; Musgrave and Wilcox 2013; Nielsen
2010), were associated with supporting recreational
marijuana legalization. In the past two decades, spend-
ing a greater amount of time engaged with media (e.g.,
watching television or reading the newspaper) also was
associated with favoring legalization, although this rela-
tionship was not observed prior to 1990, and may be
due to more positive depictions of marijuana in recent
years (Stringer and Maggard 2016). Other researchers
have found that residents of states and countries that
have legalized medical or recreational marijuana sup-
port legalization at higher rates than residents of states
or countries where marijuana is not legal (Sznitman
and Bretteville-Jensen 2015; McGinty et al. 2017);
Schuermeyer et al. 2014). Exceptions to this trend
have been seen in young adult samples where rates of
support for legalization of marijuana for recreational
use are high regardless of the state’s policies (Cohn
et al. 2017; Moreno et al. 2016).

Additionally, an individual’s own use of substances
contributes to their broader opinions on substance use

CONTACT Jennifer D. Ellis [email protected] Department of Psychology, Wayne State University, 5447 Woodward Ave., Detroit, MI 48202

JOURNAL OF PSYCHOACTIVE DRUGS
2019, VOL. 51, NO. 4, 335–342
https://doi.org/10.1080/02791072.2019.1610199

© 2019 Taylor & Francis Group, LLC

issues (Hilton and Kaskutas 1991; Latimer et al. 2001;
Resko 2014; Wagenaar et al. 2000). Cruz, Quierolo, and
Boidi’s (2016) analysis of samples from the United
States, El Salvador, and Uruguay found that past mar-
ijuana use was among the strongest predictors for sup-
porting recreational marijuana legalization. Williams,
Van Ours, and Grossman (2011) analysis of the
Australian Drug Strategy’s National Household Survey
similarly found current and past marijuana use were
significantly associated with supporting recreational
marijuana legalization. A sample of adults living in
Houston found that frequency of past 30-day use of
marijuana, as well as use of other drugs, was associated
with supporting legalization of marijuana (Trevino and
Richard 2002). In a US-based sample of young adults,
Cohn et al. (2017) found past month use of alcohol,
tobacco, marijuana, and other drugs were associated
with greater likelihood of supporting recreational mar-
ijuana legalization.

Much of the research on attitudes toward recrea-
tional marijuana legalization has used binary measures
that force respondents to indicate support for or against
marijuana legalization; however, some research sug-
gests that a number of individuals are still undecided
about whether they believe marijuana should be lega-
lized (Geiger 2016). Individuals who are undecided
about legalization have the potential to play an impor-
tant role as they decide how to vote in future elections.
Therefore, characterizing those who are unsure about
legalization may be useful. Additionally, examining
perceived risk towards marijuana use as a potential
correlate of attitudes towards legalization is important,
as perceived risk of marijuana has decreased over time
in the United States (Okaneku, Vearrier, McKeever,
LaSala & Greenberg, 2015), as support for legalization
has increased.

The present study examines the extent to which recrea-
tional marijuana legalization was supported by adults in
Michigan, as well as factors associated with supporting,
opposing, and being unsure about marijuana legalization.
In line with previous work demonstrating that one’s own
substance use is associated with perspectives about lega-
lization (Cohn et al. 2017; Trevino and Richard 2002), we
hypothesized that past year marijuana users would be
more likely to favor legalization, whereas those with less
recent marijuana use would be more likely to be unsure.
Additionally, because political orientation has been
related to supporting or opposing marijuana legalization
(Looby, Earleywine, and Gieringer 2007; Musgrave and
Wilcox, 2013; Nielsen 2010), we expected that those who
identify as politically centrist would be more likely to be
unsure about legalization. Finally, because women have
historically been more likely to be opposed to marijuana

legalization, but otherwise tend to lean politically liberal
relative to men (Galston and Dionne 2013), we expected
that women would be more likely to oppose or be unsure
about legalization when given this option.

Methods

Sample and study procedures

The present study was exempted from the Wayne State
University institutional review board. Data were drawn
from an anonymous web-based survey conducted by the
Michigan Prevention Association (MPA). The MPA is
a statewide advocacy group focused on substance misuse
prevention. The purpose of the statewide survey was to
examine attitudes toward marijuana legalization in
Michigan. In 2008, Michigan voters passed a ballot mea-
sure to legalize medical marijuana. Ten years later,
Michigan voters passed a measure to legalize recreational
marijuana (November 2018). For the current survey, the
MPA recruited a sample of adults (ages 18+) in Michigan.
Recruitment emails were sent to community partners
(e.g., health-care providers, mental health providers,
city/county health departments, county substance abuse
coalitions, non-profit/government agencies serving
families or older adults). Ads also were placed online
(e.g., Facebook), and recruitment announcements were
made at community events (e.g., community meetings,
health fairs). Advertisements stated MPA was looking for
adults (ages 18+) in Michigan to respond to a brief survey
(5–10 min) regardless of their experiences with marijuana.
Data collection took place over a three-week period in
August and September of 2016, prior to Michigan voters
passing a ballot measure for marijuana legalization. As
a validity check for the online survey, the researchers
included “please leave this item blank” in the middle of
the survey and examined the length of time spent com-
pleting the survey. The recruitment efforts yielded
a sample of 2,608 participants, of which 2,190 passed
validity checks and had complete data for all relevant
study measures. This group of 2,190 was included in the
following analyses.

Measures

Survey questions were grounded in previous public opinion
and substance use survey research. The MPA steering
committee developed the survey with input from its mem-
bers who worked in non-profit and government agencies
that provide community-based substance abuse treatment
and prevention services. Because participants were not
provided incentives for survey participation, efforts were
made to ensure brevity.

336 J. D. ELLIS ET AL.

Marijuana legalization
Attitudes toward marijuana legalization were assessed
with a question from the Pew Research Center (Geiger,
2016) that asks: “Do you think marijuana should be
made legal for recreational use, or not?” Response
options adapted for this study were: “Yes, Legal”, “No,
illegal” and “Unsure”.

Demographic variables
Demographic questions were selected or adapted from
the 2015 National Survey on Drug Use and Health
(Substance Use and Mental Health Administration
2016). These self-report questions included gender
(male, female, or other), age, race or ethnic origin
(White/Caucasian, Black/African American, Latino/
Hispanic, American Indian, Asian/Pacific Islander,
Arab/Middle Eastern, Multiracial, Other), highest level
of education (less than high school diploma or general
education degree (GED), high school diploma/GED,
bachelor’s degree, or graduate degree), and whether
participants had children age 18 or under. Because
small portions of respondents endorsed Latino/
Hispanic, American Indian, Asian/Pacific Islander,
Arab/Middle Eastern, multiracial, or other race/ethni-
city, those participants were combined into an “Other
race” group. Political affiliation was assessed with
responses grouped into three categories: left of center,
right of center, and center/other, a framework that has
been used in other studies (e.g., Furnham and
Thomson 1996).

Marijuana use
Participants were asked to self-report the last time that
they used marijuana. Response options were never,
within the past year, and more than one year ago.

Perceived risk associated with Marijuana
Participants were asked to rate their agreement with four
items addressing perceived risks associated with mari-
juana use. The items said 1) “Marijuana use is harmless”
(reverse scored), 2) “Marijuana is addictive”, 3) “Using
marijuana poses serious mental health risks”, and 4)
“Using marijuana poses serious physical health risks.”
Responses were rated on a 4-point scale ranging from
1 = Strongly Disagree to 4 = Strongly Agree. Scores for
the items were summed to create a composite measure
assessing the perceived risks of marijuana use, which
demonstrated good reliability in the present sample
(α = .886). These items were developed for this study,
and were informed by existing research examining com-
mon perceived risks associated with marijuana use (e.g.,
Berg et al. 2015; Danseco, Kingery, and Coggeshall 1999;
Wilkinson et al. 2016).

Data analysis

Data were analyzed using SPSS version 25 (IMB Corp.
2017). Descriptive statistics were examined for all study
variables. We conducted bivariate analyses using Pearson’s
χ2 tests for categorical independent variables and Analysis
of Variance (ANOVA) for continuous independent vari-
ables to compare those who supported, opposed and were
unsure about marijuana legalization. Independent variables
included demographics (age, gender, race/ethnicity, having
children in household, education level), political orienta-
tion and substance use-related variables (current or past
marijuana use, perceived risk associated with marijuana
use). After examining the bivariate relationships,
a multinomial logistic regression model was conducted
where all correlates were examined simultaneously. Being
opposed to marijuana legalization was used as the reference
category. To investigate potential problems with multicol-
linearity, we conducted correlational (Spearman’s rho)
analyses of the predictor variables (Menard 2001). No
issues with multicollinearity were found.

Results

Descriptive statistics are presented in Table 1.
Participants ranged in age from 18 to 88 years, with
an average age of 45.37 years (SD = 13.39). The sample
was 71.7% female and 90.7% of the sample reported
their race as “white.” Just under half of the sample
(47.4%) supported legalization of recreational mari-
juana, 41.6% did not support legalization, and 11.1%
were unsure.

Bivariate relationships between variables of interest
and opinions towards marijuana legalization are
reported in Table 2. Given the uneven group sizes and
a violation of the homogeneity of variance assumption
for both age (p = .004) and perceived risk associated with
marijuana use (p = .001), post-hoc tests using the
Games-Howell statistic were performed for the
ANOVA models. The results of these tests indicated
that individuals who opposed legalization were more
likely to be older than those who were unsure about
legalization (Mean Difference = 3.43, p = .001) and
who supported legalization (Mean Difference = 4.36,
p < .001). Individuals who opposed legalization per-
ceived marijuana to be more risky than those who were
unsure about legalization (Mean Difference = 0.63, p <
.001) and those who supported legalization (Mean
Difference = 1.46, p < .001). Those who were unsure
about legalization reported significantly greater per-
ceived risk than those who supported legalization
(Mean Difference = 0.83, p < .001). In addition to these
findings, chi-square tests indicated that, at the bivariate

JOURNAL OF PSYCHOACTIVE DRUGS 337

level, gender, education level, political orientation, and
marijuana use were significantly associated with opi-
nions regarding legalization of recreational marijuana.
No differences were observed for race/ethnicity or being
a parent.

Results of the multinomial logistic regression analy-
sis are presented in Table 3. Politically left of center or
centrist views, past year marijuana use, and lifetime
marijuana use were associated with higher odds of
supporting legalization, relative to opposing

legalization. Older age, female gender, and greater per-
ceived risk of marijuana were associated with lower
odds of supporting legalization. Similarly, politically
left of center or centrist political views and lifetime
marijuana use were associated with higher odds of
being unsure about legalization. Older age and greater
perceived risk of marijuana were associated with lower
odds of being unsure about legalization, relative to
opposing legalization. After controlling for other vari-
ables, education level was not a significant predictor of
attitudes towards legalization.

Notably, prior marijuana use and perceived risk of
marijuana were the strongest predictors of opinions
regarding legalization in the multivariate model.
Among those who had used marijuana in the
past year, 94.1% favored legalization, compared to
46.0% of those who had used prior to the
previous year, and 24.1% of those who had never used
marijuana. On a measure of perceived risk of marijuana
ranging from 1 to 4, those who opposed legalization
had the highest perceived risk (M = 3.4), followed by
those who were unsure about legalization (M =2.7), and
those who favored legalization (M = 1.9).

Discussion

As marijuana policy continues to evolve, an understand-
ing of attitudes toward legalization for recreational pur-
poses provides insights useful for political and prevention
efforts. It is particularly important to characterize those
who are undecided about marijuana legalization before

Table 1. Descriptive statistics.
Variable N % M SD

Age – – 45.37 13.39
Gender
Male 619 28.3 – –
Female 1571 71.7 – –

Children under 18 in Household 903 41.2 – –
Race
White 1986 90.7 – –
African American 97 4.4 – –
Other Race 107 4.9 – –

Education Level
High School Diploma or Less 459 21.0 – –
Bachelor’s Degree 892 40.7 – –
Graduate/Professional Degree 839 38.3 – –

Political Orientation
Left of Center 888 40.5 – –
Right of Center 536 24.5 – –
Center/Other 766 35.0 – –

Past Marijuana Use
Never used 781 35.7 – –
Lifetime Use 994 45.4 – –

Past Year Use 415 18.9 – –
Perceived Risk of Marijuana (scores range
from 1 to 4)

– – 2.59 0.94

Support Marijuana Legalization
Yes 1037 47.4 – –
No 911 41.6 – –
Unsure 242 11.1 – –

Table 2. Bivariate relationship between the predictors and support, unsure or opposition to legalization of recreational marijuana
use in Michigan.

Marijuana Legalization Marijuana Legalization

Variable Legal % Illegal % Unsure % Legal M(SD) MIllegal (SD) Unsure M(SD) X2 F P value

Age – – – 43.5(13.6) 47.8(12.9) 44.4(13.2) – 27.06 < .001
Gender 38.14 – < .001
Male 57.7 34.6 7.8 – – –
Female 43.3 44.4 12.3 – – –

Children under 18 in Home 44.7 44.2 11.1 – – – 4.69 – .096
Race 3.21 – .524
White 47.4 41.6 11.0 – – –
African American 41.2 47.4 11.3 – – –
Other Race 52.3 35.5 12.1 – – –

Education Level 34.54 – < .001
HS Diploma or Less 54.9 33.3 11.8 – – –
Bachelor’s Degree 50.2 40.2 9.5 – – –
Graduate/Professional 40.2 47.6 12.3 – – –

Political Orientation 112.80 – < .001
Center 50.7 37.5 11.9 – – –
Right of Center 31.5 61.0 7.5 – – –
Left of Center 54.1 33.4 12.5 – – –

Marijuana Use 591.97 – < .001
Never 24.1 65.9 10.0 – – –
Lifetime Use 46.0 38.5 15.5 – – –
Past Year Use 94.5 3.1 2.4 – – –

Perceived Risk of Marijuana (range 1–4) – – – 1.9(0.69) 3.4(0.60) 2.7(0.58) – 1267.63 < .001

N = 2190

338 J. D. ELLIS ET AL.

they decide how they will vote, as prevention or advocacy
groups may benefit from giving particular attention to
these populations prior to elections. While previous
research on attitudes toward marijuana legalization (e.g.,
Cruz, Queirolo, and Boidi 2016; Galston and Dionne
2013) has yielded considerable insight on support for or
against marijuana legalization, less is known about those
who are undecided about marijuana legalization. Results
from the present study suggest that 47.4% of adults in

Michigan supported marijuana legalization and a sizable
minority were still undecided.

Findings from this study extend upon previous
research (Caulkins et al. 2012; Cruz, Queirolo, and Boidi
2016; Looby, Earleywine, and Gieringer 2007; Musgrave
and Wilcox 2013) that has shown conservative or liberal
political attitudes differentiate between those who support
and oppose marijuana legalization. Our findings suggest
that adults who are unsure about marijuana legalization
may be more similar, in terms of political orientation, to
those who support marijuana legalization, as being left of
center or center was associated with supporting or being
unsure about marijuana legalization. A notable minority
of politically conservative adults in our sample (31.5%)
supported legalization. This may reflect some ideological
and partisan crossover, where conservatives are increas-
ingly supporting legalization because they favor states’
rights and are suspicious of federal enforcement
(Galston and Dionne 2013).

Recent work found gender was significantly associated
with opinions regarding marijuana legalization, while
other demographic factors such as race, income, educa-
tion, and geography had only small relationships with
attitudes toward legalization (Galston and Dionne 2013).
The present study also suggested that women were more
likely than men to oppose legalization. The gender differ-
ences observed in the present study are notable in that
women, as a group, are more likely to vote left-of-center,
but are more aligned with political conservatives on this
issue (Galston and Dionne 2013). Interestingly, in the
present study, the bivariate results replicated previous
findings that higher education level is associated with
lower likelihood of favoring legalization, but this finding
disappeared when we controlled for other relevant factors.
Future research into demographic factors most strongly
associated with attitudes towards legalization, and poten-
tial reasons for these beliefs, may be beneficial.

The present study is only one of a few that has exam-
ined rates of supporting legalization among recent mar-
ijuana users. Those who were recent marijuana users
were more likely to support legalization compared to
those who had never used. It should also be noted that
recent marijuana users were not more likely to be unsure
than to be opposed to legalization, and rates of being
unsure about or opposing legalization were low among
past year users. This finding is in line with previous
studies (e.g., Cohn et al. 2017; Cruz, Queirolo, and
Boidi 2016; Williams, Van Ours, and Grossman 2011)
that have shown personal experiences with marijuana,
and one’s own use of marijuana, in particular (e.g., Cohn
et al. 2017; Cruz, Queirolo, and Boidi 2016; Trevino and
Richard 2002), is one of the strongest predictors of
supporting marijuana legalization.

Table 3. Multinomial logistic regression analysis predicting atti-
tudes toward legalization of marijuana for recreational use in
Michigana.

Variable B SE Wald p-value OR 95%CI

Yes, Legal
Age −0.03 0.01 25.35 < .001 0.97 0.96–0.98
Female
Gender

−0.79 0.17 20.72 < .001 0.46 0.32–0.64

Has children −0.16 0.16 1.05 .305 0.85 0.63–1.16
Raceb

African
American

−0.23 0.35 0.43 .515 0.80 0.40–1.58

Other −0.01 0.34 0.001 .974 0.99 0.51–1.92
Political Orientationc

Left of
Center

1.05 0.20 28.80 < .001 2.86 1.95–4.20

Center 0.63 0.20 10.14 .001 1.87 1.27–2.76
Education Leveld

Bachelor’s
Degree

0.10 0.21 0.24 .623 1.11 0.74–1.66

Graduate
Degree

0.03 0.21 0.02 .894 1.03 0.68–1.55

Marijuana usee

Past year 2.32 0.36 41.30 < .001 10.13 5.00–20.53
Lifetime 1.15 0.16 54.19 < .001 3.15 2.32–4.28
Perceived
risk

−2.85 0.13 486.56 < .001 0.06 0.05–0.08

Unsure
Age −0.03 0.01 14.43 < .001 0.98 0.96– 0.99
Female
Gender

−0.10 0.20 0.26 .610 0.90 0.61–1.34

Has children −0.14 0.17 0.72 .397 0.87 0.62–1.21
Raceb

African
American

−0.24 0.38 0.39 .531 0.79 0.37–1.66

Other −0.02 0.37 0.003 .954 0.98 0.48–2.01
Political Orientationc

Left of
Center

1.03 0.22 21.35 < .001 2.81 1.81–4.35

Center 0.86 0.22 14.99 < .001 2.37 1.53–3.67
Education Leveld

Bachelor’s
Degree

−0.38 0.22 2.92 .088 0.68 0.44–1.06

Graduate
Degree

−0.24 0.22 1.21 .271 0.79 0.51–1.21

Marijuana usee

Past year 0.46 0.48 0.92 .337 1.58 0.62–4.00
Lifetime 0.94 0.17 31.26 < .001 2.56 1.84–3.56
Perceived
risk

−1.51 0.13 138.37 < .001 0.22 0.17–0.28

Nagelkerke Pseudo Rb = 0.645
aNo, does not support legalization was the reference group for the multi-
nomial logistic regression.

b Caucasian/White was used as the reference group for Race.
c Right of Center was the reference group for the Political Orientation
variable.

d High School education was the reference group for the Education Level
variable.

e Never Used Marijuana was the reference group for the Marijuana Use
variable.

JOURNAL OF PSYCHOACTIVE DRUGS 339

Our findings regarding perceived risk associated
with marijuana extend earlier research that has focused
on adolescent samples. Campbell, Twenge and Carter’s
(2017) analysis of twelfth-graders in the Monitoring the
Future dataset, for example, found perceptions of risk
and support for legalization appear in tandem, with
higher perceived risk linked to low support for legaliza-
tion and lower perceived risk linked to strong support
for legalization. The current findings suggest a similar
pattern for adults. Those who supported or were unsure
about marijuana legalization had lower levels of per-
ceived risk compared to those who were opposed to
legalization. As perceived risk associated with mari-
juana use declines (Okaneku et al. 2015), findings on
personal experiences and risk perception may have
implications for the messaging of marijuana preven-
tion. Prevention messages that are designed to increase
risk perception may help to reduce marijuana use
among some, but also could potentially backfire for
those who perceive marijuana use to be benign.

The present study brings needed attention to people
who are undecided about marijuana legalization. Future
research may also benefit from examining what might
change decisions for individuals who have already formed
an opinion about marijuana legalization. Looby,
Earleywine, and Gieringer (2007), for example, found
that adults were more likely to support legalization if
a reliable roadside test could be administered.
Additional research is needed on how state regulations
for marijuana, including tax rates or limitations on who
may grow and distribute marijuana, may shape voting.
Regulations vary considerably from state to state and
evidence suggests they have already played a part in recent
elections. Legalization initiative that failed to pass, for
example, included a limit of 10 facilities given exclusive
commercial rights to grow marijuana in that state. This
regulation divided those who advocated for marijuana
legalization and is likely one of the reasons the initiative
failed (Graham 2015).

Additional research on the reasons people support,
oppose or are unsure about legalization is also
needed as attitudes toward marijuana continue to
shift (Kilmer and MacCoun 2017; Resko et al.
2019). Many people who support marijuana legaliza-
tion do so despite recognition of potential negative
consequences (Resko et al. 2019). As more states
move to legalize recreational marijuana, understand-
ing the influence of marijuana policy changes on
public attitudes is increasingly important. Some indi-
viduals may consider marijuana legalization as
a governmental endorsement that marijuana is safe
for recreational use (Moreno et al. 2016). Prevention
and education efforts may be needed to inform the

public that state-level marijuana legalization does not
imply governmental endorsement, nor does it suggest
that marijuana products have been deemed safe
(Moreno et al. 2016).

The results of this study should be considered in the
context of their limitations. First, the study was cross-
sectional. Given the ever-changing nature of public
opinion on marijuana (Kilmer and MacCoun 2017),
future studies may benefit from conducting longitudi-
nal analyses examining how attitudes toward legaliza-
tion change with time, and particularly whether these
attitudes change after legalization. The Michigan
Prevention Association opposes marijuana legalization,
and prevention providers distributed the survey, which
may have resulted in a sample that over-represents
adults with lower levels of support for marijuana lega-
lization. Additionally, the sample was not representa-
tive of the state of Michigan as a whole, and a yielded a
disproportionately high rate of White and female
respondents. Recruitment techniques that would yield
more representative and diverse samples, or use of
existing national datasets, may be ideal in the future.
For example, future studies could consider analyzing
data from a nationally representative sample, such as
the General Social Survey. Additionally, future studies
may want to consider oversampling among historically
underrepresented groups to ensure adequate represen-
tation of perspectives of those groups who have not
been well represented (e.g., racial-ethnic minorities and
those who identify as transgender or non-binary).
Recruiting a diverse sample may be particularly impor-
tant, as criminalization of marijuana and of substance
use more broadly has disproportionately impacted low-
income people of color (Koch, Lee, and Lee, 2016).
While evidence of racial or ethnic differences in atti-
tudes toward legalization has been mixed (e.g., …