Clearing The Haze - The Complexities And... Es of Research On State Marijuana Laws

Ann. N.Y. Acad. Sci.
ISSN 0077-8923
ANNALS OF THE NEW YORK ACADEMY OF SCIENCES

Issue: Addiction Reviews
REVIEW
Clearing the haze: the complexities and challenges of

research on state marijuana laws
Esther K. Choo1 and Sherry L. Emery2
1
Department of Emergency Medicine, Center for Policy and Research in Emergency Medicine, Oregon Health & Science
University, Portland, Oregon. 2 Health Media Collaboratory, Chicago, Illinois
Address for correspondence: Esther K. Choo, Department of Emergency Medicine, Center for Policy and Research in
Emergency Medicine, Oregon Health & Science University, 3181 SW Sam Jackson Park Drive, Portland, OR 97239.
[email protected]
As states increasingly liberalize marijuana laws, high-quality research is needed that will inform the public and policy-
makers about the health and societal impact of these laws. However, there are many challenges to studying marijuana
policy, including the heterogeneity of the drug and its use, the variability in the laws and their implementation
from state to state, the need to capture a wide variety of relevant outcomes, and the poorly understood influence of
marijuana commercialization. Furthermore, current instruments generally fail to distinguish between types of users
and lack accurate and detailed measures of use. This review provides a background on marijuana laws in the United
States and an overview of existing policy research, discusses methodological considerations when planning analysis
of state marijuana laws, and highlights specific topics needing further development and investigation.
Keywords: cannabis; marijuana; policy; law; legislation
Introduction has emerged as a medical treatment for refractory

conditions.
Historical perspective
The turn of the century ushered in a remarkable
Marijuana, a drug derived from the cannabis plant,
shift in the landscape of marijuana laws in the United
has had a volatile history in the United States. Ini-
States. While the federal government retains central
tially a medicinal by-product of hemp production,
control over restrictions on marijuana,3 many indi-
it was introduced as a recreational drug in the early
vidual states––beginning with California in 1996
1900s with an influx of immigrants from Mexico
with Proposition 215, the Compassionate Use Act4 –
and rapidly became associated with crime and vio-
–have loosened regulations, passing laws accept-
lence. In sharp contrast to recent times, individual
ing some degree of use without penalty, ranging
states, responding to public campaigns against the
from medical use only to some amount of recre-
drug, restricted use of marijuana long before the
ational use. At the same time, individual cities have
enactment of federal laws. More lenient political and
also enacted laws in opposition to their own state
cultural attitudes emerged in the 1960s. Neverthe-
laws. The result has been a heterogeneous patch-
less, marijuana was classified as a schedule I drug––
work of marijuana policies, with states and cities
defined by the Drug Enforcement Administration
introducing and implementing laws allowing the
as having “no currently accepted medical use and
acquisition and use of marijuana to varying degrees
a high potential for abuse”1 ––under the Controlled
and by various means. Reflecting or perhaps driv-
Substances Act of 1970 and has retained this des-
ing legal reform, social norms about marijuana use
ignation ever since.2 The schedule I classification
have evolved quickly and dramatically. As recently
has restricted research on the safety, medical effi-
as 2006, less than one third of U.S. adults supported
cacy, and potential for abuse, even as marijuana
doi: 10.1111/nyas.13093
C 2016 New York Academy of Sciences.
Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 ! 55
Research on state marijuana laws Choo & Emery
marijuana legalization; in 2015, a majority (53%) quality evidence suggesting cannabinoids could
supported legalization.5 improve nausea and vomiting due to chemother-
There are many factors that have driven the sus- apy, weight gain in HIV, sleep disorders, and tics
tained effort to legalize marijuana: its ongoing use from Tourette syndrome; and very low-quality evi-
across social strata; its relatively low and benign dence that marijuana could lead to an improvement
morbidity; perceptions that marijuana has little in anxiety. The study also found that cannabinoids
addictiveness;6 its ready availability; and its favor- were associated with an increased risk of adverse
able profile in terms of association with violence events, such as dizziness, nausea, vomiting, diarrhea,
or injury, especially compared with other schedule I fatigue, asthenia, confusion, balance problems, and
drugs, such as cocaine or heroin, and even other legal hallucinations.
drugs, such as alcohol and prescribed opioids.7-9 Indeed, there are persistent concerns in the med-
Its epidemiology reflects its status as a socially sanc- ical community about the safety, health effects––
tioned substance.10 Among U.S. adolescents, exper- particularly on youth––and overall societal cost of
imentation with marijuana is routine, second in legalizing marijuana. Heavy cannabis use is associ-
prevalence only to alcohol; in some states, almost ated with a number of negative health and social
half of school-age adolescents have used it at least sequelae, including early onset of schizophrenia
once.11 Worldwide, marijuana is the most com- and other psychotic disorders,15,16 neurocognitive
monly used illicit drug.12 deficits,17,18 lower educational attainment,19,20 and
lower employment and income.21 The American
Current status of health research Association of Pediatrics continues to oppose any
on marijuana marijuana use in children up to age 21 and legaliza-
Despite its entrenched cultural and societal role, tion outside of medical marijuana laws (MMLs).22
research on cannabis remains in its infancy. Even Concerns around various degrees of legalization of
the clinical effects of cannabis are incompletely marijuana include the extent to which increased
understood. While the medical literature has gen- availability will lead to increases in the potency
erally supported cannabis use for a variety of con- of available marijuana and increased crime associ-
ditions that are difficult to treat with conventional ated with dispensaries.23,24 Alcohol and other drugs
therapies,13 the evidence base for its positive health tend not to be used in isolation, but to influence
effects is surprisingly thin. A 2015 systematic review the use of each other, whether as complements or
and meta-analysis published in JAMA searched 28 substitutes.25,26 Therefore, collateral effects on other
databases to identify randomized controlled trials types of substance use are also important consider-
(RCTs) of cannabis for the treatment of nausea and ations.
vomiting due to chemotherapy, appetite stimula- Most states have adopted some form of mari-
tion in HIV/AIDS, chronic pain, spasticity due to juana laws (Fig. 1). Accordingly, there is an emerg-
multiple sclerosis (MS) or paraplegia, depression, ing scientific literature on the various effects of these
anxiety disorder, sleep disorder, psychosis, intraoc- laws.23,27-33 However, this area of investigation is
ular pressure in glaucoma, or Tourette syndrome. also relatively new, and published studies to date
The authors identified 79 published RCTs; only have been limited by numerous factors, such as lack
four of these were judged as having low risk of of detailed national and state measures of drug use,
bias.14 One third of the studies were at least 25 years the complexity and heterogeneity of cannabis laws,
old. The majority of the studies examined cannabis the complexity of the drug itself, and insufficient
use for nausea and vomiting due to chemother- passage of time to gauge the effects of laws. As we
apy, chronic pain, or spasticity due to MS or para- discuss below, these types of challenges may affect
plegia. There were four or fewer studies examin- the results of studies on the effects of marijuana laws
ing its use for the six other studied indications: on drug use28,29,34,35 and related outcomes.24,36 The
no RCTs were identified that studied cannabis use purpose of this review is to highlight some of the
for depression. The review found moderate-quality methodological challenges that can be anticipated
evidence to suggest that cannabinoids may be ben- going forward, as investigators and policymakers
eficial for the treatment of chronic neuropathic seek a meaningful understanding of the effects of
or cancer pain and spasticity due to MS; low- changes in cannabis-related laws.
56 C 2016 New York Academy of Sciences.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 !
Choo & Emery Research on state marijuana laws
Figure 1. Types of marijuana laws in the United States. MML, comprehensive medical marijuana laws only; states with limited
MMLs are not counted. Only state laws are counted; individual cities, including Portland, ME, and Jackson, MI, have passed laws
legalizing marijuana that are discordant with state laws.37,39
Differences in individual marijuana laws: sive laws are in the intervention group, and all other
medical marijuana, decriminalization, and states are controls. Thus, control groups include
legalization for recreational use some states with limited laws, which––even given
the restricted nature of these laws––may differ from
Legalization of medical marijuana
states that do not allow any legal use of cannabis
As of March 2016, 23 states and the District
whatsoever. An even greater heterogeneity is repre-
of Columbia have passed comprehensive MMLs,
sented within comprehensive MMLs.38 Individual
according to the National Conference of State Legis-
states allow medical marijuana for specific medi-
latures (NCSL).37 Comprehensive laws are defined
cal conditions (e.g., Crohn’s disease) or treatments
by the NCSL as providing protection from criminal
of medical conditions (e.g., chemotherapy). These
penalties for using marijuana for a medical pur-
range from a handful of conditions to extensive
pose; providing access to marijuana through home
lists; Illinois has 45 approved conditions. Many also
cultivation, dispensaries or some other system that
include open-ended language allowing for a wide
is likely to be implemented; allowing a variety of
variety of medical conditions to be treated with mar-
strains, including those with more than low levels of
ijuana if they cause debilitating symptoms. Some
trans-!9 -tetrahydrocannabinol (THC); and allow-
states allow individual conditions to be added to
ing either smoking or vaporization of some kind
the established list at the discretion of the health
of marijuana products, plant material, or extract.
department. There are generally requirements for
Another 17 states have passed limited laws, which
written documentation of the medical condition by
allow products with low levels of THC and high lev-
a physician; sometimes two physicians must cer-
els of cannabidiol (CBD) for medical reasons in lim-
tify the necessity of marijuana for treatment of dis-
ited situations. Most studies of MMLs dichotomize
ease. Some states have not approved smoked mari-
the policy variable and place these limited-law states
juana, only edibles, oils, or concentrates. States also
in the control group.
vary in terms of the amount an individual patient
Studies examining the impact of MMLs have gen-
or caregiver can possess at one time, although
erally treated the presence or absence of marijuana
sometimes the amounts (e.g., “60-day supply”) are
laws as a binary function: states with comprehen-
open to gross interpretation. States define methods

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 ! 57
of obtaining the drug, whether through personal istration has approved two synthetic cannabinoid
cultivation or through medical dispensaries, which preparations, dronabinol and nabilone.45 However,
are variably restricted, and also where the drug may until these come into routine use, potency varies
be used. Many states decriminalize possession for widely, complicating the ability to measure effec-
medical use but do not define how individuals are tive increases in use.44 Although it is difficult to
to acquire the marijuana; many specifically allow track trends in potency, it is generally thought that
a certain amount of cultivation or purchase for marijuana available for medical reasons, whether
medical use, while others permit establishment of home grown or distributed through dispensaries,
treatment centers that will dispense marijuana with is of greater potency than marijuana distributed
a specific physician order.37,39 States also vary in illegally for recreational use.23 To date, there is
their requirement for medical marijuana users to heterogeneity, even when just considering mari-
register and obtain an identification card attesting juana consumed legally, posing another challenge
to their status as approved users, which provides to comparisons of the effect of MMLs on general
legal protections if they are found in possession of consumption. If an MML law leads to marijuana of
marijuana. All this variability in MMLs may affect greater THC potency entering the market for recre-
outcomes: for example, Pacula and Sevigny40 found ational use but without affecting the quantity con-
that states requiring annual registration for medi- sumed, this effect may not be detected.23 Such a
cal marijuana had lower recreational marijuana use change would have important public health signif-
rates and lower treatment admissions for marijuana icance: consumption of high-potency marijuana is
abuse compared to states without this requirement, associated with damage to the corpus callosum,46
while states allowing home cannabis cultivation and and THC confers a dose-dependent risk for acute
legal dispensaries demonstrated a positive associa- psychosis,15,47 impairments of executive function
tion with recreational use. and motor control,48 and motor vehicle collisions.49
Pacula et al. also hypothesized that MMLs that Time is another critical factor in evaluating the
simply allow existing cultivation and purchase effects of a new MML. Establishing registries and
routes likely do not affect supply compared with dispensaries may take months or years; finding
those that allow the creation of dispensing centers, providers willing to prescribe marijuana may also
which establish a new supply route. The impact on pose a challenge, particularly as marijuana use for
overall marijuana use has been hypothesized to be any reason remains federally prohibited and pre-
different in these two scenarios: lowering the penal- scribing continues to be monitored by the U.S.
ties for use while supply remains relatively fixed Drug Enforcement Administration,50 the evidence
can be expected to increase demand while driving base for medical marijuana is inadequate, and estab-
up prices, reducing the availability for many users, lished guidelines for physician prescribing are still
while lowering penalties and establishing supply lacking.51 Therefore, studies examining the impact
through dispensaries should allow prices to remain of MMLs in the few years after establishment of a
stable or decrease, increasing prevalence of use.41 law may be premature, contextualizing the repeated
Studying cannabis use brings specific challenges, research finding that adolescent use does not change
as it is unlike other medications with a single active in response to MMLs.27,30,31
component. Cannabis is the source of more than
60 individual pharmacologically active compounds Decriminalization of marijuana possession
called cannabinoids, which include THC, the main Depenalization refers to “any policy that reduces
psychotropic component, and CBD, which lacks the penalties associated with possession or use of
psychotropic activity and has anti-inflammatory, marijuana,”52 whether those policies are criminal or
antioxidant, and neuroprotective properties.42 civil in nature. Decriminalization is a specific type
There are two primary cannabis species, Cannabis of depenalization in which an offense is changed
indica and Cannabis sativa––C. indica has higher from criminal to noncriminal status. Often, stat-
levels of THC––as well as hybrids, and a vari- ues are still in place to regulate the possession of
ety of techniques for cultivation, harvesting, pro- small amounts of marijuana, but with minor penal-
cessing, and drug delivery, all of which can affect ties, such as an infraction and fine, similar to traf-
its potency.43,44 The U.S. Food and Drug Admin- fic violations. Decriminalization has been enacted

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 !
to varying degrees in 20 states and the District Legalization of marijuana for recreational use
of Columbia.52,53 In addition to differences in the Until very recently, studies examining the conse-
type of decriminalization, definitional inconsisten- quences of legalization of marijuana were con-
cies add to the complexity of studying these laws: ducted outside of the United States. However, as of
Pacula et al.52 note that some “decriminalized” states March 2016, Washington,58 Colorado,59 Oregon,60
have not statutorily removed the criminal status of and Alaska61 have approved measures legalizing
minor marijuana possession offences, while other recreational marijuana, providing natural exper-
so-called decriminalized states have. Some states iments for the effects of liberalization, particu-
remove the marijuana-related charge after comple- larly as neighboring states have drastically different
tion of the punishment for the offence, in a pro- approaches to marijuana use. As with MMLs, there
cess called “expungement,” but are not considered is some heterogeneity in the laws (Table 1), with
decriminalized, while others remove jail time as a differences in allowable amounts for possession,
possible punishment without specifically removing regulation, taxation, and timing and means of
the criminal status of marijuana possession, and yet implementation.62
are categorized as decriminalized.54 Potential delays between the adoption of liberal-
Measuring decriminalization alone may not ization laws and the actual implementation of those
capture the effect of the law as well as the actual laws may be substantial. Uruguay passed a mari-
behaviors or perceptions of the behaviors of law juana liberalization law in 2013, allowing marijuana
enforcement agencies. In some cases, decriminal- to be acquired through individual home cultivation
ization may be an attempt to remove sanctions of up to six plants; cultivation within cannabis clubs
against marijuana; they may merely represent laws of 50 members or fewer, who can grow up to 99
catching up to already lenient practices regarding plants in a collective garden; and government sales
detection of small amounts of marijuana, or they of up to 40 g per month per individual through exist-
may be empowering law enforcement to remain ing pharmacies. The latter is the only legal retail
vigilant and follow through with more frequent, if option and is anticipated to be the most popular
smaller, penalties. Marijuana laws have also been means of obtaining marijuana, with 160,000 esti-
observed to be enforced variably, with certain mated customers.63,64 However, as of January 2016,
groups less likely to receive leniency, and without the planned network of government-run dispen-
consistent correlation with the criminal status of saries was not yet in existence.63,65 Ideally, investi-
the drug within the state.52,55,56 gators of such policies should censor data collected
In any of these scenarios, routine and heavy mar- during the implementation period and study the
ijuana users or dealers may be most likely to under- effects of the policy only after full implementation.
stand local/regional law enforcement practices, to be
familiar with law enforcement, and to be calculat- General issues in assessing state marijuana
ing about potential legal risks. Nonusers and casual policy effects
or experimental users, on the other hand, may be Overall, heterogeneity in individual policies
more afraid of legal consequences, and thus may means that investigators must be cautious about
be more influenced by measures that decriminal- dichotomizing states based on the presence or
ize marijuana. Furthermore, the influence of legal absence of a specific type of law. Complete pro-
costs will be balanced with the financial costs of hibition is becoming less common; therefore, the
marijuana. Therefore, changes in these types of laws control-condition states that are used for compari-
can have a variety of effects on marijuana posses- son are increasingly those with some liberalization
sion and cultivation, depending on the context in policy in place. Further, there is little research in
an individual jurisdiction and the type of user in the ways that multiple marijuana laws implemented
question.57 Furthermore, individual municipalities in the same state may affect one another. As part
(e.g., New York City, cities in California) may have of its implementation processes, for example, Ore-
additional laws superseding state laws––in popu- gon allowed immediate sale of marijuana for recre-
lous cities, examining state policies may not take ational purposes from existing medical marijuana
into account the effect this may have on the overall dispensaries, with potential corresponding effects
estimates of drug use. on the medical marijuana distribution network,

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 ! 59
Table 1. State laws legalizing marijuana for recreational use
Implementation/
State Possession Taxation Regulation distribution
Colorado122,123 Adults (21 or older) can 15% excise tax on Four categories of marijuana Effective December
grow up to three the wholesale licenses: retail store, retail product 12, 2012.
immature and three price of manufacturing, retail cultivation,
Commercial sales
mature cannabis plants marijuana, 10% and retail testing facility.
began January 1,
privately in a locked retail tax, 2.9%
Applicants for retail store licenses 2014.
private space and state sales tax;
must:
possess all cannabis additional local
–Be Colorado residents
from the plants they taxes may apply.
"2 years before applying
grow, as long as it stays
–Be at least 21 years of age
where it was grown.
–Not have felony convictions in the
Use of marijuana while previous 5 years
driving is prohibited and –Not have controlled substances
it is illegal to have open felony convictions in the previous
or unsealed containers of 10 years
marijuana in a vehicle. –Not have any criminal history that
suggests “questionable moral
Use of marijuana on federal
character”
land or in public places
–Not employ or be financed by
is illegal. Public use of
anyone with a criminal history
up to 2 oz is a petty
offense punishable by a Local municipalities must also
$100 fine. approve of all licensed marijuana
businesses.
Washington124-126 Adults (21 or older) may 37% retail tax at Three categories of marijuana Effective December
possess up to 1 oz of point of sale (to licenses: producer, retailer, 6, 2012.
usable (smokable) customers). processor.
First retail licenses
marijuana, 16 oz of
Applicants must: issued on July 8,
marijuana-infused
–Be Washington residents for "3 2014.
product in solid form,
months
72 oz of marijuana-
infused product in
–Not have any felony convictions
liquid form, or any
in the previous 10 years
combination of all three.
–Not have a significant criminal
Home cultivation history as determined by the
prohibited, except as Liquor and Cannabis Control
allowed under Board
Washington’s medical –Not have a history of marijuana
marijuana law. law or rule violations
Local municipalities must also
approve of all licensed marijuana
businesses.
Number of retail locations initially
set at 334, ensuring that locations
were distributed across cities and
counties based on population.
Continued

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 !
Table 1. Continued
Implementation/
State Possession Taxation Regulation distribution
Oregon127-129 Adults (21 or older) can 25% sales tax for Four categories of marijuana Effective July 1,
carry up to 1 oz of the limited licenses: retailer, processor, 2015.
marijuana in public, amount of time producer (grower), wholesaler
keep up to 8 oz of (until December Sales started
homegrown, usable 31, 2016) that Applicants must: October 1, 2015
(ready to smoke) recreational –Be at least 21 years of age within existing
marijuana at home, and marijuana will –Be Oregon residents for medical
grow up to four plants be sold in " 2 years marijuana
per household. existing medical –Not have been convicted of dispensaries.
marijuana violating any state, local, or
Recreational retail
Public use of marijuana is federal law, if the conviction is
dispensaries. license
prohibited. deemed “substantially related” to
applications
Recreational the fitness and ability of the
accepted starting
marijuana sold applicant to carry out activities
January 4, 2016.
in recreational under the license
stores will be –Be of “good repute and moral
taxed at a base character;” not in the habit of
rate of 17%. using alcohol, habit-forming
Cities and drugs, marijuana, or controlled
counties may substances to excess
add an
Cities and counties may opt out of
additional tax of
having marijuana businesses in
up to 3%.
their jurisdictions.
No limit to the number of licenses
that will be issued.
Alaska130,131 Adults (21 or older) can $50 per oz Four categories of marijuana 16-month lapse
possess up to 1 oz of wholesale tax. licenses: retail store, cultivation between effective
marijuana, grow up to facility, product manufacturing date of law
six plants, and share/gift facility, testing facility. (February 24,
up to 1 oz or six plants 2015) and start
Applicants must:
to another adult. Only date for first
three of the plants can be commercial
–Be Alaska residents
mature and flowering at marijuana
–Not have any felony convictions
any one time. businesses.
–Not have certain misdemeanor
Public use is a violation convictions related to controlled
punishable by $100 fine. substances, violence against a
Possession of 1–4 oz is a person, or use of a weapon in the
class A misdemeanor previous 5 years
with a punishment of up –Not have a class A misdemeanor
to 1 year in prison conviction related to the sale or
and/or a fine of up to distribution of marijuana in the
$10,000––unless the previous 2 years
possession was for
There is no limit to the number of
personal use and
licenses that will be issued by the
occurred in the
state. Local governments may
offender’s private
impose their own restrictions.
residence, in which case
there is no penalty.
Note: Information up to date as of April 2016.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 ! 61
the neighborhoods around the dispensaries, and strategies. Furthermore, online businesses aggres-
the regulated and unregulated economies related sively market marijuana paraphernalia, including
to the dispensaries. Finally, as the discussion in vaporizers and blunt wraps, via Twitter, Instagram,
this section reflects, the complexity and variabil- and websites, often using celebrities, musicians, and
ity of the laws, the inconsistent use of terminology, models to promote both their products and mar-
and the rapidly changing nature of the laws make ijuana use generally.72 Research about the effects
them a challenging target for accurate characteri- of tobacco marketing provide strong evidence that
zation and study. National nonpartisan or biparti- campaigns that incorporate celebrity endorsements,
san organizations that keep a central repository of popular music, and attractive, youthful models are
current marijuana-related laws, such as the NCSL highly effective at normalizing and even glamoriz-
(www.ncsl.org) or Procon.org (www.procon.org), ing smoking,73,74 leading to increased risk of uptake
are valuable resources for policy researchers. among youth.75–80 Yet to date there is no research
on the effects of such marketing practices for mar-
Commercialization of marijuana
ijuana and related products or the extent to which
At the same time that marijuana-related norms and such marketing interacts with the potential effects
policy are rapidly shifting, there is an emerging legal of legal status.
marijuana economy.66 Marijuana dispensaries and
Research methodological challenges
retail outlets advertise a broad range of products and
and potential pitfalls
devices.67 While some of this advertising and pro-
motion appears in local magazines and newspapers, Defining the using population
much of it takes place online, and includes social Because the various environmental factors likely
media.68,69 For example, companies like WeedMaps, affect subgroups of users differentially, it is impor-
Wikileaf, and Leafly help consumers locate and tant to clearly define subcategories of marijuana
rate products and vendors; countless other sites users. Pacula and Lundberg57 defined four types
offer product and legal information, health claims, of users when considering the impact of a new
how-to advice (ranging from consumption to par- marijuana law: (1) initiators and light users, who
enting), and links to marijuana-related cultural are experimenting or consuming small doses on an
events. This information appears on their websites infrequent basis; (2) regular users, who use mari-
and Twitter, Facebook, and Instagram accounts. juana in small or moderate doses on a more fre-
Some marijuana-related marketing is distinctly quent basis; (3) heavy users, who consume on a near
local, directing individuals to vendors and events daily basis or who meet clinical criteria for depen-
in their community. Other marketing appears in dence or abuse;a and (4) quitters, individuals who
the context of online interest communities—or may endorse current or recent use and be identi-
“stoner” communities. These communities both fied as marijuana users but have already decided to
promote social norms and serve as marketing plat- stop using. Similar subtypes have been described by
forms for other marijuana marketing accounts. other investigators.20,81 These are likely to be over-
Although the actual effect on marijuana purchase simplified, as a great deal of heterogeneity is rep-
and consumption is not yet known, even passive resented within cannabis use disorders, compared
exposure to promarijuana content on social media to alcohol use disorders, which have been observed
has been demonstrated to be associated with higher to have high homogeneity.82 However, using these
odds of marijuana use.70 Based on the number of general subtypes as a general framework, the first
members and followers, the potential influence of
these marketing and stoner community accounts
is significant; the popular promarijuana Twitter a
As an aside, there is a potential measurement bias rep-
account “Weed Tweets” (handle @stillblazingtho), resented here: DSM-IV criteria for drug abuse included
for example, has 1.22 million followers and in 2013 experiencing legal problems due to use of the drug,121
was in the top 10% followed Twitter accounts by meaning that in states with liberalization for recreational
17–19 year olds.71 use, the law could impact rates of drug severity simply by
Little, if anything, is known about the reach being passed, without changing use patterns per se. The
or influence of these digital marijuana marketing legal criterion has been removed in DSM-5.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 !
group has been of particular interest among those marijuana in situations that place them at risk of
studying policy effects, as initiators tend to be ado- harm (e.g., driving, crossing traffic);88 this raises a
lescents, who experience the greatest physical and potential research question of whether policies that
social harms of marijuana. Heavy users, on the other change perceived risk of drug use may have differ-
hand, make up most of the consumption and may be ential effects between genders. The context of drug
more sensitive to changes in price. Survey questions use has also been observed to differ among adoles-
often fail to distinguish between users so as to allow cents based on gender, age, and race, with whites
an understanding of which population is affected by more likely to use drugs in private residences and
a law and to provide empirical evidence of suspected black users more likely to use in school;87 in this
mechanisms of change. For example, if an instru- example, the policy implication for minority groups
ment inquires about daily marijuana use, and this is easy to imagine, as legalization generally applies
use increases over time as a proportion of the total to private use, with persistent prohibition in pub-
population, it may be that the increased availabil- lic places, which are considered subject to federal
ity and decreased cost associated with an influx of law.
marijuana product into the state has led to increased
experimentation among naive individuals, or it may Accurate measurement of relevant outcomes
be that light or regular users with established drug The research community has called for an increase
access are able to increase their use. There are a vari- in the specificity of measurements of drug use.89,90
ety of taxonomies for marijuana use (e.g., by longi- Existing measures tend to ask about days of use
tudinal course83 or by demographic, mental health, and measures of dependence90–93 but rarely quantify
substance use, and risk perception subtypes84 ), but amount per day by asking not only about frequency
which are most relevant as marijuana laws continue or hours spent “high,” but total amounts consumed
to change may depend on the individual law and the using standardized measures of dose and potency.90
population under study; or, as marijuana’s place in Mariani et al.85 estimated marijuana quantity by
society shifts, newer taxonomies may be needed to incorporating a drug substitute into the Timeline
accurately characterize patterns of use. Followback to allow participants to demonstrate
In addition to the challenges of operationaliz- amounts used, similar to how wine or shot glasses
ing different levels or stages of use, it is important may be used to estimate amounts of alcohol con-
to distinguish between the increasing varieties of sumed. Without defining drug potency, however,
methods of cannabis consumption. It is arguable even weight-based amounts do not capture the
that the frequency, amount, and circumstances of effective dose individuals are accessing, informa-
recreational marijuana use may differ substantially tion essential for the evaluation of marijuana policy
from use for medicinal purposes; further, the use of effects (it has been demonstrated, e.g., that avail-
a topical cannabis product is a qualitatively different ability of dispensaries due to changes in MMLs
behavior than smoking marijuana for recreational affects the potency of available marijuana23 ). van der
or medical use. It is likely that the populations using Pol et al.94 found that individual self-assessments
various products and methods are quite distinct. As of dose and potency had poor accuracy; identify-
new types of marijuana consumption emerge, each ing accurate means of capturing these measures in
likely implies different use patterns and health and population-based studies is much needed.
safety profiles. Yet, few, if any, current survey instru- Modes of delivery are neglected in many standard
ments allow researchers to disentangle the wide vari- questionnaires about drug use. Although smoking
eties of products and use patterns.85 marijuana via bowl, pipe, joint, and secondarily
Beyond—but perhaps related to—differences in through the use of a bong, water pipe, or hookah
the frequency and amount of use and the wide vari- remains the most common mode of use, “dab-
ety of marijuana product and devices, distinct mari- bing” (inhalation of concentrated butane-extracted
juana use trends have been observed for age, gender, cannabis products, also called butane hash oil prod-
generational cohort, race, and country of origin.86,87 ucts or BHO), “vaping” (vaporizing cannabis prod-
Policies may, therefore, affect different subpopu- ucts to create an aerosolized mixture of water and
lations in different ways. For example, men have active ingredients, which is inhaled), and edibles
been observed to be more likely than women to use are also common. The use of the latter forms is

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 ! 63
increasing and is expected to increase further with earmarked for substance use prevention, treatment,
the legalization of marijuana. Asking about modes and research.103,104 Measurements related to level of
of use, therefore, is critical to the accuracy of mea- utilization of these services must be carefully con-
surements of use.89 sidered, as increases in drug use treatments provided
National surveys, including the National Sur- may reflect increased funding, availability, or quality
vey on Drug Use and Health (NSDUH), have of services rather than increased severity or preva-
allowed some evaluation of the effect of chang- lence of drug use disorders.
ing marijuana laws on a population level. Self- Drugs are not used in isolation. When marijuana
report has demonstrated good concordance with becomes more available, it may serve as a com-
objective measures,95,96 although context pertains plement to other substances, making use of these
to the accuracy of self-reported drug activity: for substances increase in parallel. Alternatively, as
example, self-report performs poorly when mea- marijuana supply increases, it may decrease use
sured after driving under the influence.97 Gender of other drugs by effectively serving as a sub-
and nationality may also affect accurate endorse- stitute for them. A topic of ongoing debate is
ment of marijuana use and consequences.98 Further, whether cannabis and alcohol are complements or
there are inherent limitations to this type of data. substitutes.34,105–107 Observations of fewer opioid-
Datasets like the Centers for Disease Control and related deaths in states with MMLs have suggested
Prevention’s Youth Risk Behavioral Study (YRBS) that marijuana may serve as a substitute for opi-
are centrally designed and administered by individ- oid use and misuse.108 Chaloupka et al.109 examined
ual states. In the case of YRBS, states have discretion Monitoring the Future data and found that increases
about including some “optional” modules, includ- in cigarette prices were associated with reductions
ing more detailed assessments related to drug use in the average level of marijuana use, supporting
and other health risk behaviors. States may not par- the theory that these substances are economic com-
ticipate in given years, whether by election or lack of plements; Farrelly et al.110 found a similar relation-
resources, so that the data are often not continuous. ship between cigarette sales taxes and intensity of
Furthermore, the data fail to be robust once response marijuana use in the National Household Survey
rates fall below a certain level, so years of low on Drug Abuse. The notion that cigarette smoking
participation lead to further discontinuity in avail- forms a gateway to marijuana use has been exten-
able data. Surveys such as YRBS and the adult health sively studied, with the underlying assumption that
survey Behavioral Risk Factor Surveillance Survey, young people typically begin experimenting with
which contain some of the most consistently col- a legal drug and subsequently progress to using
lected national health data in the United States, are an illegal one.111 However, more recent evidence
designed to measure a wide range of health indices. has also supported that marijuana use may lead to
Thus, the substance use questions are brief and gen- tobacco smoking, nicotine dependence, and prevent
eral, allowing only assessment of lifetime and past- cessation.112–115 The phenomenon of “blunting”–
month use episodes. These types of surveys often do –i.e. smoking little cigars and cigarillos (LCCs)
not allow detection of changes in use, particularly emptied of tobacco and filled with marijuana––
among established regular or heavy users. Such sur- has been identified as expanding the LCC market
veys often lack information about the broad variety among marijuana smokers and partially explain-
of outcomes related to marijuana use that are rele- ing how blunt use can contribute to nicotine
vant to understanding the impact of policies, such dependence.116 Given the grave long-term health
as negative personal consequences,91 direct health effects of tobacco, longitudinal studies are needed
effects,33 healthcare utilization,36,99 and economic to assess how marijuana laws may affect tobacco
and societal costs,100,101 including effects on school consumption, if these substances are indeed mutual
performance and employment.32 complements. Determining if and how increasing
Among the arguments for legalization of cannabis marijuana availability affects the use of alcohol and
is the possibility that government resources can be other drugs––each with their own associated health
shifted from law enforcement to substance-related risks and costs––is an important part of the calcula-
health services and prevention;102 indeed, in some tion of the consequences of marijuana-related state
states, taxes from cannabis markets are specifically laws.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 !
Figure 2. Hypothetical interrupted time series analysis of marijuana law. β1 , slope of line preintervention; β2 , change in intercept;
β3 , change in slope from preintervention to postintervention.
Quasi-experimental study designs for repeated measures.118 The ITS model is represented
research on marijuana policy effects as (presuming a linear trend):b
The gradual adoption of marijuana laws by individ-
ual states provides a series of natural experiments Ratet = β0 + β1 timet + β2 programt
that can be studied with quasi-experimental meth- + β3 time after programt + et
ods. It is not effective or helpful to simply compare
prevalence in states with and without laws, as states ITS also lends itself well to graphical represen-
with higher use are more likely to pass laws that are tation of changes in outcomes pre- and postinter-
more lenient toward marijuana possession, cultiva- vention (Fig. 2), including in samples stratified by
tion, use, and sale; laws require significant public characteristics of interest (age, gender, race, etc.) to
support before they are passed. Such studies do find show variable effects of policies on specific subpop-
higher prevalence of marijuana use associated with ulations. The ability to illustrate the effect of a policy
liberalization policies,28 but this finding is expected with clarity is valuable for communicating with the
and likely to reflect baseline prevalence of drug use public and policymakers.117
before changes in the law rather than a causal associ- Although ITS is an established and popular quasi-
ation. Simple before-and-after analyses, too, may fail experimental approach, there are a number of
to account for underlying trends occurring before potential pitfalls. Studies may fail to justify the time
policy implementation. points or spacing between data-collection points,
Time-series analysis is used to evaluate the influ- including failing to account for the expected lag
ence of policy interventions on behavior at the pop- during implementation of an intervention, before
ulation level.117 Compared to simple before–after effects of intervention are expected, leading to
analyses, an interrupted time-series (ITS) analysis undermeasurement of its effects. Ignoring autocor-
uses segmented regression, calculating the rate of relation, whether between sequential data points
the outcome (intercept) and the changes in the rate
(slope) in the outcome separately before and after
the studied intervention. This corrects for potential b
More detailed information about how to set up data and
biases in simple before-and-after comparisons, conduct ITS can be found at: Interrupted time series (ITS)
which may include secular trends, seasonal varia- analyses. Cochrane Effective Practice and Organisation of
tion, random fluctuations, and autocorrelation in Care (EPOC) Group. 2013.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 ! 65
or higher-order correlations, such as those occur- the same way as the primary outcome variable.
ring with seasonal variation, or failing to include In a popular example of the NEDV, Ross et al.119
enough time points to adjust for autocorrelation, examined the effect of a law authorizing police to
may result in incorrect estimates of effect size. Inves- administer on-the-road breathalyzer testing on the
tigators must also consider possible effects of the outcome of traffic injuries and fatalities. The authors
intervention on measurement of the outcome (e.g., used traffic accidents during high-volume commut-
if social desirability or fear about divulging drug ing hours when pubs were closed as a NEDV: this
use decreases due to more lenient policy changes, variable might be expected to be responsive to other
this may lead to increased reporting of marijuana laws and campaigns that affected driving safety, but
use, rather than or in addition to increased use should not change in response to the breathalyzer
of marijuana). Adequate data points––particularly law. Demonstrating a large decrease in injuries and
preintervention––are necessary in order to estab- fatalities during times that pubs were open but not
lish the underlying trends and fluctuations in drug during times when pubs were closed was strong evi-
using behavior and for accurate model fit. The true dence that the law had an effect.
postintervention period, as discussed above, may An alternative design is the difference-in-
begin sometime after policy implementation, par- difference analysis, which is the average difference in
ticularly if there is an initial pilot period or delays in outcome(s) of interest in one state (with the policy
actual implementation. Therefore, obtaining ade- change of interest), less the average difference
quate data points after the intervention entails wait- among the comparison group (the state without the
ing many months or years, which may be frustrating policy change).
from a public policy standpoint.
δ̂1 = (ȳB,2 − ȳ B,1 ) − ( ȳ A,2 − ȳA,1 )
Marijuana regulation is very much in the public
eye, particularly around rapid changes in legislation. The first difference reflects the change in the
Therefore, important events, including other pro- primary outcome (drug use) that occurs after the
posed or approved marijuana laws in nearby states, policy implementation. By subtracting the second
public health campaigns, or–– increasingly––mass difference, the change that occurs in the comparison
marketing efforts, may influence popular attitudes group, secular changes that may have occurred
and behavior, as would any other policy put in place for reasons not related to the policy are excluded
within the “test” state that might be mistaken as an from the analysis. Any remaining differences
effect of the intervention; these competing forces in outcome––the difference in difference––are
are the biggest threat to validity of ITS analysis. An attributed to the policy change. Compared to
ITS with a control group improves upon the simple the controlled, ITS, the difference in difference is
ITS, particularly the threats to validity due to other simpler, as it looks at average change; therefore, it
events that may have affected the marijuana-related has less stringent data requirements.120 Difference-
outcome (Fig. 3). The impact of a program is evalu- in-difference analyses carry the assumption that the
ated by looking at whether the behavior trend within underlying trends are parallel between intervention
the treatment group changes from its preinterven- and control groups; thus, the average change in
tion trend by a greater amount than in the compari- the control group after the intervention is what
son group; the counterfactual outcome (what would would have happened in the intervention group
have happened in the absence of the intervention) is had there been no intervention. However, if there
estimated using the control condition, rather than were preintervention differences in trends between
the state’s own preintervention trend data. Utiliz- the two groups, or if some outside factor unrelated
ing a control group addresses the threats to internal to the policy produces differential effects across
validity discussed above, including history, instru- the two groups, results may be biased.132,c The
mentality, and regression to the mean.
Another solution is to include a nonequivalent
dependent variable (NEDV). An NEDV is one that c
Other approaches to endogeneity using difference-in-
is not expected to change in response to the inter- difference analysis, including difference-in-difference-
vention, but is expected to respond to the other in-difference and instrumental variable procedures, are
contextually important internal validity threats in discussed in detail in Ref. 133.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 !
Figure 3. Hypothetical interrupted time series analysis of marijuana law, with control state.
difference-in-difference is increasingly being used likely to be some imbalance, it is important to con-

to examine effects of policy changes.134 sider how to control for key characteristics thought
For both the controlled, ITS and difference-in- to be influential on the outcome, whether by adjust-
difference analysis, selection of an appropriate con- ing the analysis for such factors, matching subjects
trol is a challenge. Ideally, in the case of state policy by the characteristic(s), or otherwise standardizing
analysis, a neighboring state may provide political the study samples.
and cultural similarity and be subject to the same Harper et al.30 provide a helpful example of how
factors, other than the intervention in question, that the approach to analysis of state-level data may
may influence year-to-year marijuana use.d At the affect the conclusions of studies on marijuana pol-
same time, a neighboring state may be most likely icy. The authors first present a previously published
to be influenced by events of the test state, including study by Wall et al.,29 which examined NSDUH
a spillover of attitudes about marijuana use. In the data using a linear regression with fixed effects
case of marijuana laws, the high degree of national for year, MML status (present or not), and an
interest in liberalization makes it hard to assume that interaction between year and MML status. This
any state is shielded from effects in another state, allowed comparison between states with and with-
particularly when a mechanism of increased recre- out MMLs before and after the implementation of
ational use is thought to be changing social norms the MML and found an association between MMLs
and attitudes. Further, the state-to-state compar- and higher adolescent marijuana use. However, the
isons that are often used for studying marijuana laws authors then reexamined the same data using the
often ignore underlying demographic differences difference-in-difference analysis (selected due to
between the state populations. Given that there is limited pre/post time points). By using the differ-
ence in difference, the authors were able to compare
within-state changes in marijuana use before and
d
Neighboring states with differences in state laws, of after a new law to within-state changes in states
course, may in fact be polar opposites culturally, polit- without the law. This model did not find an impact
ically, and demographically: Nevada and Utah are stark of MMLs on adolescent marijuana use on the basis
examples of this. of the NSDUH data.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 ! 67
Table 2. Overview of cannabis policy research challenges

! Tremendous heterogeneity in marijuana-related laws
! Categorization of states based on types of laws implemented is inconsistent
Understanding the laws
! Relationship between legal status, actual law enforcement, and public perception of
! Complex drug with many compounds, variability in psychotropic potency

existing cannabis laws is inconsistent
! Potency difficult to measure

Defining cannabis use
! Need to capture greater detail in frequency, amount, type of product, and mode of
!
delivery
!
Need for data on a broad variety of related health, social, economic outcomes
!
Defining the using population Need to better characterize and understand typologies of use
!
Need to better understand patterns of use among demographic subgroups
!
Analytic considerations Simple before-and-after analyses fail to capture secular trend
Sufficient time points are needed to establish underlying trends and adjust for
! Difference-in-difference analysis may not need as many data points as ITS analysis, but
autocorrelation
! Design needs to consider threats to validity due to influential events other than
has a strong parallel trends assumption
! Difficult to identify appropriate control group

marijuana law
Summary and future directions Cannabis itself is a complex drug, and the impreci-
sion and inaccuracy of current methods of measur-
The potential challenges to studying the effects of
ing drug quantity, frequency, potency, and mode
marijuana policy are summarized in Table 2. A fea-
and method of use may limit our ability to examine
sible and critical first step is to improve the accuracy
the effects of new laws. Understanding mechanisms
and precision of definitions and measurement tools
of drug use change will enable investigators to
for cannabis consumption, as no amount of ana-
measure relevant proximal and distal outcomes.
lytic sophistication can compensate for inaccurate
There is much to be learned about the impact of
data. A more nuanced approach to marijuana laws–
legalization on marketing; the impact of marketing
–instead of evaluating them as simply present ver-
on norms and utilization; and how use, abuse, and
sus absent––is also needed to gain understanding of
marijuana-related health and social outcomes will
how policy changes affect marijuana use and related
be affected by commercialization of marijuana mar-
outcomes. Examining the effect of new laws on
kets. Policy research must also examine the complex
demographic subgroups and recognizing any inher-
interplay of overlapping and variable policies and
ent biases, particularly along racial or gender lines,
implementation of policies, as well as the impact of
should be an early concern for the public and pol-
marijuana laws on the use of other legal and illegal
icymakers. Given the wide-ranging potential areas
substances. ITS with control groups and difference-
of research, however, achieving consensus among
in-difference analyses are useful approaches for
scholars with expertise in marijuana policy research
isolating the effect of a policy on state drug use
and policy makers may be a valuable first step in
outcomes.
order to prioritize the research that will best inform
and guide public policy. Conflicts of interest
The authors declare no conflicts of interest.
Conclusions
The current landscape of marijuana legislation rep-
resents a great opportunity for natural policy exper- References
iments. Like many “real life” experiments, studying 1. Drug Enforcement Administration: Drug Schedul-
marijuana laws is complex, owing to the variability ing. Accessed February 1, 2016. http://www.dea.gov/
in the individual laws and their implementation. druginfo/ds.shtml.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 !
2. Hawken, A., J. Caulkins, B. Kilmer & M. Kleiman. 2013. fMRI response during spatial working memory. Psychiatry
Quasi-legal cannabis in Colorado and Washington: local Res. 163: 40–51.
and national implications. Addiction 108: 837–838. 19. Centers for Disease Control and Prevention. 2010.
3. Lorber, L. 2005. Gonzales v Raich: the US Supreme Court’s Alcohol and other drug use and academic achieve-
consideration captured the public policy debate about the ment. Accessed February 1, 2016. https://www.cdc.gov/
medical use of marijuana. Gend. Med. 2: 124–130. healthyyouth/health_and_academics/pdf/alcohol_other_
4. California Department of Public Health. 1996. Medical drug.pdf.
marijuana program: Compassionate Use Act. Proposition 20. Homel, J., K. Thompson & B. Leadbeater. 2014. Trajectories
215. Accessed March 28, 2016. https://www.cdph.ca.gov/ of marijuana use in youth ages 15–25: implications for
programs/MMP/Pages/CompassionateUseact.aspx. postsecondary education experiences. J. Stud. Alcohol Drugs
5. Pew Research Center. 2015. In debate over legalizing 75: 674–683.
marijuana, disagreement over drug’s dangers. Washington, 21. Fergusson, D.M. & J.M. Boden. 2008. Cannabis use and
DC: Pew Research Center. Accessed February 9, 2016. later life outcomes. Addiction 103: 969–976; discussion
http://www.people-press.org/2015/04/14/in-debate-over- 977–978.
legalizing-marijuana-disagreement-over-drugs-dangers/. 22. Ammerman, S., S. Ryan, W.P. Adelman; Committee on
6. Berg, C.J., E. Stratton, G.L. Schauer, et al. 2015. Perceived Substance Abuse, Committee on Adolescence. 2015. The
harm, addictiveness, and social acceptability of tobacco impact of marijuana policies on youth: clinical, research,
products and marijuana among young adults: marijuana, and legal update. Pediatrics 135: e769–e785.
hookah, and electronic cigarettes win. Subst. Use Misuse 50: 23. Sevigny, E.L., R.L. Pacula & P. Heaton. 2014. The effects of
79–89. medical marijuana laws on potency. Int. J. Drug Policy 25:
7. Macdonald, S., P. Erickson, S. Wells, et al. 2008. Predicting 308–319.
violence among cocaine, cannabis, and alcohol treatment 24. Morris, R.G., M. TenEyck, J.C. Barnes & T.V. Kovandzic.
clients. Addict. Behav. 33: 201–220. 2014. The effect of medical marijuana laws on crime:
8. Gilbert, L., N. El-Bassel, M. Chang, et al. 2012. Substance evidence from state panel data, 1990–2006. PLoS One 9:
use and partner violence among urban women seeking e92816.
emergency care. Psychol. Addict. Behav. 26: 226–235. 25. Moore, S.C. 2010. Substitution and complementarity in
9. Korcha, R.A., C.J. Cherpitel, J. Witbrodt, et al. 2014. the face of alcohol-specific policy interventions. Alcohol
Violence-related injury and gender: the role of alcohol and Alcohol. 45: 403–408.
alcohol combined with illicit drugs. Drug Alcohol Rev. 33: 26. Petry, N.M. 2001. A behavioral economic analysis of poly-
43–50. drug abuse in alcoholics: asymmetrical substitution of alco-
10. Winstock, A. The Global Drug Survey 2015 Findings hol and cocaine. Drug Alcohol Depend. 62: 31–39.
Accessed January 25, 2016. http://www.globaldrugsurvey. 27. Lynne-Landsman, S., M. Livingston & A. Wagenaar. 2013.
com/the-global-drug-survey-2015-findings/. Effects of state medical marijuana laws on adolescent mar-
11. Centers for Disease Control and Prevention. 2013. 1997– ijuana use. Am. J. Public Health 103: 1500–1506.
2009 Youth Risk Behavior Survey. Accessed February 3, 28. Cerdá, M., M. Wall, K.M. Keyes, et al. 2012. Medical
2016. www.cdc.gov/yrbss. marijuana laws in 50 states: investigating the relationship
12. United Nations Office on Drugs and Crime. 2012. World between state legalization of medical marijuana and mar-
drug report. Accessed January 10, 2016. http://www.un ijuana use, abuse and dependence. Drug Alcohol Depend.
odc.org/documents/data-and-analysis/WDR2012/WDR_ 120: 22–27.
2012_web_small.pdf. 29. Wall, M.M., E. Poh, M. Cerda, et al. 2011. Adolescent mar-
13. Nutt, D., L.A. King, W. Saulsbury & C. Blakemore. 2007. ijuana use from 2002 to 2008: higher in states with medi-
Development of a rational scale to assess the harm of drugs cal marijuana laws, cause still unclear. Ann. Epidemiol. 21:
of potential misuse. Lancet 369: 1047–1053. 714–716.
14. Whiting, P.F., R.F. Wolff, S. Deshpande, et al. 2015. 30. Harper, S., E.C. Strumpf & J.S. Kaufman. 2012. Do medical
Cannabinoids for medical use: a systematic review and marijuana laws increase marijuana use? Replication study
meta-analysis. JAMA 313: 2456. and extension. Ann. Epidemiol. 22: 207–212.
15. Di Forti, M., A. Marconi, E. Carra, et al. 2015. Proportion 31. Choo, E.K., M. Benz, N. Zaller, et al. 2014. The impact
of patients in south London with first-episode psychosis of state medical marijuana legislation on adolescent mari-
attributable to use of high potency cannabis: a case–control juana use. J. Adolesc. Health 55: 160–166.
study. Lancet Psychiatry 2: 233–238. 32. Kepple, N.J. & B. Freisthler. 2012. Exploring the ecological
16. Kelley, M.E., C.R. Wan, B. Broussard, et al. 2016. Marijuana association between crime and medical marijuana dispen-
use in the immediate 5-year premorbid period is associated saries. J. Stud. Alcohol Drugs 73: 523–530.
with increased risk of onset of schizophrenia and related 33. Salomonsen-Sautel, S., S.-J. Min, J.T. Sakai, et al. 2014.
psychotic disorders. Schizophr. Res. 171: 62–67. Trends in fatal motor vehicle crashes before and after
17. Schweinsburg, A.D., S.A. Brown & S.F. Tapert. 2008. The marijuana commercialization in Colorado. Drug Alcohol
influence of marijuana use on neurocognitive functioning Depend. 140: 137–144.
in adolescents. Curr. Drug Abuse Rev. 1: 99–111. 34. Anderson, D.M. & D.I. Rees. 2013. Medical marijuana laws,
18. Schweinsburg, A.D., B.J. Nagel, B.C. Schweinsburg, et al. traffic fatalities, and alcohol consumption. J. Law Econ. 56:
2008. Abstinent adolescent marijuana users show altered 333–369.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 ! 69
35. Thies, C.F. & C.A. Register. 1993. Decriminalization of mar- 54. Pacula, R., J. Chriqui & J. King. 2003. Decriminaliza-
ijuana and the demand for alcohol, marijuana and cocaine. tion in the United States: what does it mean? National
Soc. Sci. J. 30: 385–399. Bureau of Economic Research, Cambridge. Working paper
36. Davis, J.M., B. Mendelson, J.J. Berkes, et al. 2015. Public No. 9690. Accessed January 8, 2016. http://www.nber.org/
health effects of medical marijuana legalization in Col- papers/w9690.
orado. Am. J. Prev. Med. 50: 373–379. 55. Alexander, M. 2012. The New Jim Crow: Mass Incarceration
37. National Conference of State Legislatures. 2016. State in the Age of Colorblindess. New York: The New Press.
medical marijuana laws. Accessed January 27, 2016. http:// 56. Johnson, B.D., A. Golub, E. Dunlap & S.J. Sifaneck. 2008.
www.ncsl.org/research/health/state-medical-marijuana- An analysis of alternatives to New York city’s current
laws.aspx. marijuana arrest and detention policy. Policing 31: 226–
38. Hoffmann, D.E. & E. Weber. 2010. Medical marijuana and 250.
the law. N. Engl. J. Med. 362: 1453–1457. 57. Pacula, R.L. & R. Lundberg. 2014. Why changes in price
39. Procon.org. 24 Legal medical marijuana states and DC. matter when thinking about marijuana policy: a review of
Accessed April 18, 2016. http://medicalmarijuana.procon. the literature on the elasticity of demand. Public Health Rev.
org/view.resource.php?resourceID=000881. 35: 1–18.
40. Pacula, R.L. & E.L. Sevigny. 2014. Marijuana liberaliza- 58. Washington State Liquor and Cannabis Board. Accessed
tion policies: why we can’t learn much from policy still in January 20, 2016. http://www.liq.wa.gov/mj2015/faqs_
motion. J. Policy Anal. Manage. 33: 212–221. i-502.
41. Pacula, R.L., D. Powell, P. Heaton & E.L. Sevigny. 2015. 59. Colorado Department of Revenue, Enforcement Divi-
Assessing the effects of medical marijuana laws on mari- sion: Marijuana enforcement. Accessed January 20, 2016.
juana use: the devil is in the details. J. Policy Anal. Manage. https://www.colorado.gov/pacific/enforcement/marijuanae
34: 7–31. nforcement.
42. Pertwee, R.G. 2006. Cannabinoid pharmacology: the 60. Oregon.gov: Recreational Marijuana. Accessed January
first 66 years. Br. J. Pharmacol. 147(Suppl. 1): S163– 20, 2016. http://www.oregon.gov/olcc/marijuana/Pages/
S171. default.aspx.
43. Hillig, K.W. & P.G. Mahlberg. 2004. A chemotaxo- 61. Alaska Department of Health and Social Services, Divi-
nomic analysis of cannabinoid variation in Cannabis sion of Public Health. Know the laws about marijuana.
(Cannabaceae). Am. J. Bot. 91: 966–975. Accessed January 21, 2016. http://dhss.alaska.gov/dph/
44. Burgdorf, J.R., B. Kilmer & R.L. Pacula. 2011. Heterogeneity Director/Pages/marijuana/law.aspx.
in the composition of marijuana seized in California. Drug
62. Oregon.gov. Oregon/Washington/Colorado Comparison.
Alcohol Depend. 117: 59–61.
Accessed January 28, 2016. http://www.oregon.gov/olcc/
45. FDA and Marijuana. 2015. Accessed January 8, 2016.
marijuana/Documents/Measure91_sidebysidecomparison.
http://www.fda.gov/NewsEvents/PublicHealthFocus/ucm
pdf.
421163.htm.
63. Carless, W. 2016. Two years later, Uruguay still has no legal
46. Rigucci, S., T.R. Marques, M. Di Forti, et al. 2016. Effect of
marijuana for sale. Global Post. January 9, 2016. Accessed
high-potency cannabis on corpus callosum microstructure.
April 18, 2016. http://www.globalpost.com/article/
Psychol. Med. 46: 841–854.
6716408/2016/01/07/uruguay-legal-marijuana-sale.
47. Di Forti, M., C. Morgan, P. Dazzan, et al. 2009. High-
64. Boidi, M.F., J.M. Cruz, R. Queirolo & E. Bello-Pardo.
potency cannabis and the risk of psychosis. Br. J. Psychiatry
2015. Marijuana legalization in Uruguay and beyond.
195: 488–491.
Accessed February 9, 2016. https://www.wilsoncenter.
48. Ramaekers, J.G., G. Kauert, P. van Ruitenbeek, et al. 2006.
org/sites/default/files/Marijuana%20Legalization%20in%
High-potency marijuana impairs executive function and
20Uruguay%20and%20Beyond_brief%20report_v3%20
inhibitory motor control. Neuropsychopharmacology 31:
(2).pdf.
2296–2303.
65. Cher, L. 2016. Uruguay’s half-baked marijuana experiment.
49. Li, M.-C., J.E. Brady, C.J. DiMaggio, et al. 2012. Mari-
Foreign Policy. Accessed April 18, 2016. http://foreignpolicy.
juana use and motor vehicle crashes. Epidemiol. Rev. 34:
com/2016/03/21/uruguay-marijuana-legalization.
65–72.
50. Lazar, K. & S. Murphy. 2014. DEA targets doctors linked to 66. Ferner, M. 2015. Pioneer pot states have collected more
medical marijuana. Boston Globe. June 6, 2014. than $200 million in marijuana taxes. The Huffington Post.
51. California Medical Association. 2011. Physician rec- August 2015. Accessed April 18, 2016. http://www.huff
ommendation of medical Cannabis guidelines of the ingtonpost.com/entry/washington-colorado-marijuana-
Council on Scientific Affairs Subcommittee on Medical taxes_us_55d6133be4b07addcb45e65d.
Marijuana Practice Advisory. Accessed January 8, 2016. 67. Ferner, M. & R.J. Reilly. 2014. Marijuana industry sets its
http://www.mbc.ca.gov/Licensees/Prescribing/medical_ sights on the mainstream. The Huffington Post. November
marijuana_cma-recommend.pdf. 2014. Accessed April 18, 2016. http://www.huffingtonpost.
52. Pacula, R.L., R. MacCoun, P. Reuter, et al. 2005. What com/2014/11/24/marijuana-industry-mainstream_n_
does it mean to decriminalize marijuana? A cross-national 6195194.html.
empirical examination. Adv. Health Econ. Health Serv. Res. 68. Sloane, G. 2015. Pot marketers must get creative as digital
16: 347–369. ad networks just say no. Adweek. Accessed April 18, 2016.
53. Joffe, A. & W.S. Yancy. 2004. Legalization of marijuana: http://www.adweek.com/news/technology/pot-marketers-
potential impact on youth. Pediatrics 113: e632–e638. must-get-creative-digital-ad-networks-just-say-no-162682.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 !
69. Jao, C. 2015. How companies are trying to bring mari- blunt, and non-blunt marijuana users from the National
juana ads online. Entrepreneur. Accessed April 18, 2016. Survey of Drug Use and Health. Drug Alcohol Depend. 160:
https://www.entrepreneur.com/article/246896. 105–111.
70. Cabrera-Nguyen, E.P., P. Cavazos-Rehg, M. Krauss, et al. 85. Mariani, J.J., D. Brooks, M. Haney & F.R. Levin. 2011.
2016. Young adults’ exposure to alcohol- and marijuana- Quantification and comparison of marijuana smoking
related content on Twitter. J. Stud. Alcohol Drugs 77: 349– practices: blunts, joints, and pipes. Drug Alcohol Depend.
353. 113: 249–251.
71. Cavazos-Rehg, P., M. Krauss, R. Grucza & L. Bierut. 86. Kerr, W.C., T.K. Greenfield, J. Bond, et al. 2007. Age–
2014. Characterizing the followers and tweets of a period–cohort influences on trends in past year marijuana
marijuana-focused Twitter handle. J. Med. Internet Res. 16: use in the US from the 1984, 1990, 1995 and 2000 National
e157. Alcohol Surveys. Drug Alcohol Depend. 86: 132–138.
72. Emery, S.L., H. Tran, G. Szczypka, et al. 2015. “Check out 87. Goncy, E.A. & S. Mrug. 2013. Where and when adoles-
that #rollgame:” amount and themes of little cigar and cents use tobacco, alcohol, and marijuana: comparisons
cigarillo content on Twitter. In Annual Meeting of the Amer- by age, gender, and race. J. Stud. Alcohol Drugs 74: 288–
ican Public Health Association, Chicago, IL. 300.
73. Cummings, K.M., C.P. Morley, J.K. Horan, et al. 2002. 88. Gillespie, N.A., M.C. Neale, L.N. Legrand, et al. 2012. Are
Marketing to America’s youth: evidence from corporate the symptoms of cannabis use disorder best accounted
documents. Tob. Control 11(Suppl. 1): i5–i17. for by dimensional, categorical, or factor mixture models?
74. Pucci, L.G. & M. Siegel. 1999. Features of sales promotion in A comparison of male and female young adults. Psychol.
cigarette magazine advertisements, 1980–1993: an analysis Addict. Behav. 26: 68–77.
of youth exposure in the United States. Tob. Control 8: 29– 89. Asbridge, M., C. Duff, D.C. Marsh & P.G. Erickson. 2014.
36. Problems with the identification of “problematic” cannabis
75. Distefan, J.M., J.P. Pierce & E.A. Gilpin. 2004. Do favorite use: examining the issues of frequency, quantity, and drug
movie stars influence adolescent smoking initiation? Am. use environment. Eur. Addict. Res. 20: 254–267.
J. Public Health 94: 1239–1244. 90. López-Pelayo, H., A. Batalla, M.M. Balcells, et al. 2015.
76. Hafez, N. & P.M. Ling. 2006. Finding the Kool Mixx: Assessment of cannabis use disorders: a systematic review
how Brown & Williamson used music marketing to sell of screening and diagnostic instruments. Psychol. Med. 45:
cigarettes. Tob. Control 15: 359–366. 1121–1133.
77. Mead, E.L., R.N. Rimal, R. Ferrence & J.E. Cohen. 2014. 91. Bashford, J., R. Flett & J. Copeland. 2010. The Cannabis
Understanding the sources of normative influence on Use Problems Identification Test (CUPIT): development,
behavior: the example of tobacco. Soc. Sci. Med. 115: 139– reliability, concurrent and predictive validity among ado-
143. lescents and adults. Addiction 105: 615–625.
78. National Center for Chronic Disease Prevention and Health 92. Piontek, D., L. Kraus & D. Klempova. 2008. Short scales to
Promotion (US) Office on Smoking and Health. 2012. Pre- assess cannabis-related problems: a review of psychometric
venting tobacco use among youth and young adults: a properties. Subst. Abuse Treat. Prev. Policy 3: 25.
report of the surgeon general. Atlanta: Centers for Disease 93. Annaheim, B., J. Rehm & G. Gmel. 2008. How to screen
Control and Prevention. for problematic cannabis use in population surveys: an
79. Sargent, J.D. & R. Hanewinkel. 2015. “Impact of media, evaluation of the Cannabis Use Disorders Identification
movies and TV on tobacco use in the youth.” In The Test (CUDIT) in a Swiss sample of adolescents and young
Tobacco Epidemic. Progress in Respiratory Research. 2nd ed. adults. Eur. Addict. Res. 14: 190–197.
Vol. 42. R. Loddenkemper & M. Kreuter, Eds.: 171–180. 94. van der Pol, P., N. Liebregts, R. de Graaf, et al. 2013. Vali-
Basel: Karger AG. dation of self-reported cannabis dose and potency: an eco-
80. Slater, M.D. & A.F. Hayes. 2010. The influence of youth logical study. Addiction 108: 1801–1808.
music television viewership on changes in cigarette use 95. Hjorthøj, C.R., A. Fohlmann, A.-M. Larsen, et al.
and association with smoking peers: a social identity, 2012. Correlations and agreement between delta-9-
reinforcing spirals perspective. Communic. Res. 37: 751– tetrahydrocannabinol (THC) in blood plasma and timeline
773. follow-back (TLFB)-assisted self-reported use of cannabis
81. de Dios, M.A., B.J. Anderson, D.S. Herman, et al. 2010. of patients with cannabis use disorder and psychotic illness
Marijuana use subtypes in a community sample of young attending the CapOpus randomized clin. Addiction 107:
adult women. Womens Health Issues 20: 201–210. 1123–1131.
82. Blanco, C., E. Ogburn, J. Pérez de Los Cobos, et al. 2008. 96. Robinson, S.M., L.C. Sobell, M.B. Sobell & G.I. Leo.
DSM-IV criteria-based clinical subtypes of cannabis use 2014. Reliability of the timeline followback for cocaine,
disorders: results from the National Epidemiological Sur- cannabis, and cigarette use. Psychol. Addict. Behav. 28: 154–
vey on Alcohol and Related Conditions (NESARC). Drug 162.
Alcohol Depend. 96: 136–144. 97. Van der Linden, T., P. Silverans & A.G. Verstraete. 2014.
83. Kandel, D.B. & K. Chen. 2000. Types of marijuana users by Comparison between self-report of cannabis use and toxi-
longitudinal course. J. Stud. Alcohol 61: 367–378. cological detection of THC/THCCOOH in blood and THC
84. Cohn, A., A. Johnson, S. Ehlke & A.C. Villanti. 2015. Char- in oral fluid in drivers in a roadside survey. Drug Test. Anal.
acterizing substance use and mental health profiles of cigar, 6: 137–142.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 ! 71
98. Delforterie, M., H. Creemers, A. Agrawal, et al. 2015. Func- ijuana use, and tobacco dependence. Addict. Behav. 33:
tioning of cannabis abuse and dependence criteria across 1491–1499.
two different countries: the United States and the Nether- 115. Ford, D.E., H.T. Vu & J.C. Anthony. 2002. Marijuana
lands. Subst. Use Misuse 50: 242–250. use and cessation of tobacco smoking in adults from
99. Kelly, E. & I. Rasul. 2014. Policing cannabis and drug related a community sample. Drug Alcohol Depend. 67: 243–
hospital admissions: evidence from administrative records. 248.
J. Public Econ. 112: 89–114. 116. Sifaneck, S.J., B.D. Johnson & E. Dunlap. 2005. Cigars-
100. Morris, R.G., M. TenEyck, J.C. Barnes & T.V. Kovandzic. for-blunts: choice of tobacco products by blunt smokers.
2014. The effect of medical marijuana laws on crime: J. Ethn. Subst. Abuse 4: 23–42.
evidence from state panel data, 1990–2006. PLoS One 9: 117. Penfold, R.B. & F. Zhang. 2013. Use of interrupted time
e92816. series analysis in evaluating health care quality improve-
101. Shanahan, M. & A. Ritter. 2014. Cost benefit analysis of two ments. Acad. Pediatr. 13(Suppl. 6): S38–S44.
policy options for cannabis: status quo and legalisation. 118. Ramsay, C.R., L. Matowe, R. Grilli, et al. 2003. Interrupted
PLoS One 9: e95569. time series designs in health technology assessment: lessons
102. Blumenauer, E. & J. Polis. 2012. The path forward: rethink- from two systematic reviews of behavior change strategies.
ing federal marijuana policy. Accessed February 8, 2016. Int. J. Technol. Assess. Health Care 19: 613–623.
http://polis.house.gov/uploadedfiles/the_path_forward. 119. Ross, H.L., D.T. Campbell & G.V. Glass. 1970. Determining
pdf. the social effects of a legal reform: the British “breathalyser”
103. State of Washington. 2011. Initiative measure No. 502. crackdown of 1967. Am. Behav. Sci. 13: 493–509.
An act relating to marijuana. Accessed January 10, 2016. 120. Somers, M.-A., P. Zhu, R.T. Jacob & H. Bloom 2013. The
http://sos.wa.gov/_assets/elections/initiatives/i502.pdf. validity and precision of the comparative interrupted time
104. State of Colorado. Proposition BB: retain revenue in excess series design and the difference-in-difference design in
of blue book estimate. Accessed February 9, 2016. http:// educational evaluation. MDRC working paper in research
www.leg.state.co.us/LCS/InitiativeReferendum/1516initre methodology. MDRC, New York.
fr.nsf/b74b3fc5d676cdc987257ad8005bce6a/f4d9ff58ad46b2 121. Hasin, D.S., C.P. O’Brien, M. Auriacombe, et al. 2013.
5687257e310079be83/$FILE/Final Draft Packet Proposi- DSM-5 criteria for substance use disorders: recom-
tion BB.pdf. mendations and rationale. Am. J. Psychiatry 170: 834–
105. Williams, J., R. Liccardo Pacula, et al. 2004. Alcohol and 851.
marijuana use among college students: economic comple- 122. Colorado Official State Web Portal. Accessed August 2,
ments or substitutes? Health Econ. 13: 825–843. 2016. https://www.colorado.gov/pacific/marijuana/laws-
106. DiNardo, J. & T. Lemieux. 2001. Alcohol, marijuana, and about-marijuana-use.
American youth: the unintended consequences of govern- 123. Colorado Official State Web Portal. Accessed August
ment regulation. J. Health Econ. 20: 991–1010. 2, 2016. https://www.colorado.gov/pacific/enforcement/
107. Pacula, R.L. 1998. Does increasing the beer tax reduce mar- business-license-applicants-retail-marijuana-2.
ijuana consumption? J. Health Econ. 17: 557–585. 124. Washington State Liquor and Cannabis Board. Fact Sheet.
108. Bachhuber, M.A., B. Saloner, C.O. Cunningham & C.L. Accessed August 2, 2016. http://www.liq.wa.gov/mj2015/
Barry. 2014. Medical cannabis laws and opioid analgesic fact-sheet.
overdose mortality in the United States, 1999–2010. JAMA 125. Washington State Liquor and Cannabis Board. Frequently
Intern. Med. 174: 1668–1673. Asked Questions about Implementing Initiative 502.
109. Chaloupka, F.J., R.L. Pacula, M.C. Farrelly, et al. 1999. Accessed August 2, 2016. http://www.liq.wa.gov/mj2015/
Do higher cigarette prices encourage youth to use mar- faqs_i-502#Retail-Stores.
ijuana? Accessed April 18, 2016. http://www.nber.org/ 126. Washington Administrative Code. Title 314 Chapter 55 Sec-
papers/w6939.pdf. tion 15. General information about marijuana licenses.
110. Farrelly, M.C., J.W. Bray, G.A. Zarkin & B.W. Wendling. Accessed August 2, 2016. http://apps.leg.wa.gov/wac/
2001. The joint demand for cigarettes and marijuana: default.aspx?cite=314-55&full=true#314-55-015.
evidence from the National Household Surveys on Drug 127. What’s Legal Oregon. Accessed August 2, 2016.
Abuse. J. Health Econ. 20: 51–68. http://whatslegaloregon.com.
111. Kandel, D. 1975. Stages in adolescent involvement in drug 128. Oregon Liquor Control Commission. Frequently Asked
use. Science 190: 912–914. Questions – Taxes.” Accessed August 2, 2016. http://www.
112. Timberlake, D.S., B.C. Haberstick, C.J. Hopfer, et al. 2007. oregon.gov/olcc/marijuana/Pages/Frequently-Asked-
Progression from marijuana use to daily smoking and nico- Questions.aspx#Taxes.
tine dependence in a national sample of U.S. adolescents. 129. Oregon Liquor Control Commission. Frequently Asked
Drug Alcohol Depend. 88: 272–281. Questions – Licensing – General. Accessed August 2, 2016.
113. Patton, G.C., C. Coffey, J.B. Carlin, et al. 2005. Reverse http://www.oregon.gov/olcc/marijuana/Pages/Frequently-
gateways? Frequent cannabis use as a predictor of tobacco Asked-Questions.aspx#Licensing_-_General.
initiation and nicotine dependence. Addiction 100: 1518– 130. Alaska Alcohol and Marijuana Control Office. Mari-
1525. juana Licensing Frequently Asked Questions. Accessed
114. Okoli, C.T.C., C.G. Richardson, P.A. Ratner & J.L. Johnson. August 2, 2016. https://www.commerce.alaska.gov/web/
2008. Adolescents’ self-defined tobacco use status, mar- amco/MarijuanaLicensingFAQs.aspx.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 !
131. AK Stat § 43.61.010. 2014. Excise Tax on Marijuana. 133. Besley, T. & A. Case. 2000. Unnatural experiments? Esti-
Accessed August 2, 2016. https://www.commerce. mating the incidence of endogenous policies. Econ. J. 110:
alaska.gov/web/portals/9/pub/MCB/StatutesAndRegulat F672–F694.
ions/MJ_BallotMeasure2.pdf. 134. Han X., K.R. Yabroff, A.S. Robbins, et al. 2014. Dependent
132. Dimick J.B. & A.M. Ryan. 2014. Methods for Evaluating Coverage and Use of Preventive Care under the Affordable
Changes in Health Care Policy. JAMA. 312: 2401. Care Act. N. Eng.l J. Med. 371: 2341–2342.

Ann. N.Y. Acad. Sci. 1394 (2017) 55–73 ! 73
Copyright of Annals of the New York Academy of Sciences is the property of Wiley-
Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.

Clearing The Haze - The Complexities And... Es of Research On State Marijuana Laws

Uploaded by

Copyright:

Available Formats

Clearing The Haze - The Complexities And... Es of Research On State Marijuana Laws

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Clearing The Haze - The Complexities And... Es of Research On State Marijuana Laws

Uploaded by

Copyright:

Available Formats

Ann. N.Y. Acad. Sci.

ANNALS OF THE NEW YORK ACADEMY OF SCIENCES

Clearing the haze: the complexities and challenges of

Keywords: cannabis; marijuana; policy; law; legislation

Introduction has emerged as a medical treatment for refractory

56 C 2016 New York Academy of Sciences.

C 2016 New York Academy of Sciences.

58 C 2016 New York Academy of Sciences.

C 2016 New York Academy of Sciences.

Table 1. State laws legalizing marijuana for recreational use

60 C 2016 New York Academy of Sciences.

Note: Information up to date as of April 2016.

C 2016 New York Academy of Sciences.

62 C 2016 New York Academy of Sciences.

C 2016 New York Academy of Sciences.

64 C 2016 New York Academy of Sciences.

C 2016 New York Academy of Sciences.

66 C 2016 New York Academy of Sciences.

difference-in-difference is increasingly being used likely to be some imbalance, it is important to con-

C 2016 New York Academy of Sciences.

Table 2. Overview of cannabis policy research challenges

! Complex drug with many compounds, variability in psychotropic potency

! Potency difficult to measure

! Difficult to identify appropriate control group

68 C 2016 New York Academy of Sciences.

C 2016 New York Academy of Sciences.

70 C 2016 New York Academy of Sciences.

C 2016 New York Academy of Sciences.

72 C 2016 New York Academy of Sciences.

C 2016 New York Academy of Sciences.

You might also like