Do arrests of youths for possession of cannabis change when cannabis policy focuses on adults?
This quasi-experimental study suggests that arrest rates of youths significantly decreased in states that decriminalized cannabis possession for everyone but did not decrease in states that legalized adult use.
It appears to be important to consider what happens to youths when cannabis policy treats adults and youths differently.
Civil liberty advocates typically support legalization of cannabis, which targets adult use, rather than decriminalization, which can affect both adults and youths. However, it is unknown how arrests of youths for cannabis possession change when adult use of cannabis is legalized.
To model changes in arrest rates of adults and youths after decriminalization and legalization of cannabis.
Design, Setting, and Participants
This quasi-experimental study used the publicly available Uniform Crime Reporting Program Data: Arrests by Age, Sex, and Race administrative data set to examine arrest rates in 38 states from January 1, 2000, to December 31, 2016. Adult (age, ≥18 years) and youth (age, <18 years) arrests for possession of cannabis were examined. States were excluded if they did not report complete arrest data or if a policy was implemented that reduced penalties for possession of cannabis but fell short of decriminalization. Fixed-effects regression was used in an extended difference-in-differences framework. The analyses in their final form were conducted between January 17 and February 28, 2019.
Living in a state with a cannabis decriminalization policy (ie, making the penalty for cannabis possession similar to the small fine for a traffic violation) or legalization policy (ie, creating a legal supply of cannabis along with the removal of penalties for possession of a small amount of cannabis for recreational use).
Main Outcome and Measures
State cannabis possession arrest rate per 100 000 population.
Data from 38 states were examined, including 4 states with cannabis legalization policies and 7 states with cannabis decriminalization policies. The adult arrest rate decreased by 131.28 (95% CI, 106.23-154.21) per 100 000 population after the implementation of decriminalization and 168.50 (95% CI, 158.64-229.65) per 100 000 population after the implementation of legalization. The arrest rate for youths decreased by 60 (95% CI, 42-75) per 100 000 population after decriminalization but did not significantly change after legalization in a state (7 per 100 000 population; 95% CI, −15 to 30).
Conclusions and Relevance
Legalization, as implemented through 2016, did not appear to reduce arrests for cannabis possession among youths, despite having benefited adults. The study’s findings suggest that decriminalization reduces youth arrests in most cases, but these findings also suggest that any benefit for youths could be lost when adult use has also been legalized. To address this problem, it appears that state decriminalization policies should take the additional step to explicitly describe when youths can be arrested for possession of small amounts of cannabis.
A 2015 American Academy of Pediatrics (AAP) policy statement supported the decriminalization of marijuana possession for youths and young adults, noting that the “illegality of marijuana has resulted in the incarceration of hundreds of thousands of adolescents.”1(p586) The AAP opposed outright legalization owing to anticipated increases in cannabis use, but the 2015 policy statement was a marked shift from their 2004 statement, in which they chose not to recommend any change to cannabis policy.2
The preference of the AAP for decriminalization is in contrast with other advocacy groups interested in reducing the negative legal consequences of cannabis possession, which have tended to support legalization (eg, the American Civil Liberties Union).3 However, there are important differences between these 2 types of policies that could influence how young people are affected. Decriminalization refers to efforts to reduce the penalties associated with possession of a small amount of cannabis from criminal to civil infractions (ie, from a crime for which one is typically arrested to something more similar to a traffic violation for which one receives a noncriminal citation). Legalization refers to policies that allow for the legal sale of cannabis as well as the removal of all penalties for possession of a small amount of cannabis for recreational use. Both adult and youth possession can be decriminalized, but legalization is specifically aimed at adult recreational use (typically age 21 years or older, as for alcohol).
Conceptually, the desire to limit youths’ access to cannabis while also reducing harms for those who do use it is related to similar concerns about youths’ use of tobacco and alcohol.4 As Wagenaar and Toomey5 note, it is inappropriate to blame only young people for their use of a substance that is legal for adults to use because adults created the environment that provided easier access for youth. Furthermore, criminal penalties for youths who have already used cannabis (or alcohol) are not effective when viewed through a harm-reduction lens.
Past work has determined that decriminalization of cannabis resulted in decreased arrests for adults and youths in Massachusetts, Connecticut, Rhode Island, Vermont, and Maryland through 2015.6 Decriminalization was also not associated with any increases in cannabis use by youths in these states. In the present study we examine the outcome of both decriminalization and legalization of cannabis on adults and youths. This is important because policy change that targets adult behavior can result in unintended consequences for youths, especially when there are no specific provisions for them (eg, legalization of cannabis for adults without accompanying decriminalization of cannabis for minors). The stakes are high for young people—being arrested during adolescence can have a profound influence on a young person’s life trajectory and health. Arrest can be traumatizing and in extreme cases can result in physical harm.7 Being arrested also has a pervasive socioeconomic influence. Teens who are arrested are more likely to drop out of high school than their peers8 and those who do stay in school may find the college admissions process more involved9 and could become ineligible for college financial aid.10 Having an arrest on one’s record can also affect employment.11 In these ways, adolescent arrest can interact with other social determinants of health. For example, lower educational level and lower income are known to be associated with poorer health outcomes.12
Our first goal is to examine changes in arrest rates of adults and youths after decriminalization of cannabis. These findings will confirm that decriminalization had its intended outcome and also determine whether adults and youths benefited to the same degree. Our second goal is to examine changes in arrest rates of adults and youths after legalization of recreational use of cannabis for adults. It is currently unknown whether arrest rates of youths are affected by legalization of cannabis for adults. Policy recommendations that focus on the legalization of cannabis currently ignore its potential outcome on youths. By identifying important differences in the implementation of decriminalization and legalization of cannabis, these findings can help inform recommendations for future policy.
Counts of arrests were obtained from the publicly available Uniform Crime Reporting Program Data: Arrests by Age, Sex, and Race (UCR) files from January 1, 2000, to December 31, 2016 (the most recent year available).13 The Federal Bureau of Investigation’s UCR program captures crime data from more than 18 000 law enforcement agencies nationally.14 State population estimates were obtained from the National Cancer Institute’s county-level US Population Data—1990-2016.15 Using these data, we generated separate adult (≥18 years of age) and youth (<18 years of age) panel data sets with the arrest rate for each year and state combination. There are several important limitations with the UCR data. First, the UCR does not contain complete data for Illinois, Florida, and the District of Columbia throughout the whole period of observation. Denver, Colorado, also determined that it misreported arrests after legalization of cannabis in that state, meaning that UCR data for Colorado were incorrect. To address this limitation, we obtained counts of arrests for Denver directly from the Denver Police Department and generated corrected state-level counts and arrest rates for Colorado. After discovering this error, we attempted to locate other instances of potential misreporting but were unable to do so. The Eastern Virginia Medical School Institutional Review Board reviewed this project and determined that it did not involve human research participants and did not require Institutional Review Board approval.
States with laws on decriminalization and legalization of cannabis were identified using several sources, which were then verified by examining legislative databases for each state (Table 1). Both policies were coded based on date of implementation, rather than passage. When separate features of a state’s policy were phased in at different times, we based our coding scheme on the time when the penalty for possession of cannabis changed (eg, possession was legalized in Washington State in December 2012 and commercial sales began in July 2014, so our effective date of implementation for Washington was December 2012). Policy variables were coded as a binary 0 or 1 except during transition years, for which fractional values were used (eg, 0.5 for policy change during June). There were also several other relevant policy changes that occurred during the analysis period. Two states, Arkansas and North Dakota, increased penalties for possession of cannabis. Several states already had at least some sort of cannabis decriminalization (Maine, Nebraska, and Ohio), while others decreased penalties during our period of analysis but fell short of actual statewide decriminalization (Kansas, Kentucky, Louisiana, Nevada, and New York). All these states, in addition to those that did not report complete arrest data, were excluded from our analyses. After exclusions, we analyzed data from 4 states with cannabis legalization policies, 7 states with cannabis decriminalization policies, and 27 control states in which the penalty for cannabis possession did not change. Delaware, which decriminalized adult possession of cannabis but kept criminal penalties for possession by youths, was considered a decriminalization state in the adult analyses and a control state in the youth analyses.
Our outcome measure for all analyses was state arrest rate for possession of cannabis. Arrests for possession of cannabis are their own category in the UCR data. Reporting follows what is known as the hierarchy rule, in which a multi-offense incident with a Part I offense is reported solely as the most severe crime that was committed (Part I offenses include crimes such as homicide, rape, and robbery). Part II offenses (such as possession of cannabis) are reported only if they occurred during an incident without a Part I offense.16 This rule means that counts of arrests for possession of marijuana are nonviolent drug offenses. Arrest rates were calculated by dividing the total number of arrests for each age group in a state by the total population for that age group and then multiplying by 100 000.
Our quasi-experimental approach used a fixed-effects, difference-in-differences analysis to compare preimplementation and postimplementation differences in arrest rates for states with a policy change with arrest rates in unexposed comparison states.17,18 Linear regression was used to estimate how arrest rates were associated with the timing of the implementation of a decriminalization or legalization policy in each state. This approach uses unordered categorical variables for state and year to control for invariant state and time effects provided that policy change and arrest rates are not both correlated with unobserved factors. Furthermore, a linear state-specific time trend was used to control for unmeasured state-level factors that might vary at a constant rate. We also included several state-level covariates that have been associated with cannabis use or arrest in past work: percentage of individuals of various age and race/ethnic groups, median income, per capita number of police, unemployment rate, poverty rate, percentage of population with a college degree, smoke-free air policy score, presence of a medical cannabis policy, and a measure of citizen political ideology.6,19-21 Sources for these covariates are listed in eTable 1 in the Supplement. Two-way cluster-robust standard errors were used to address possible correlation of observations across time and between states.22 All analyses were performed with R, version 126.96.36.199 The analyses in their final form were conducted between January 17 and February 28, 2019.
We also estimated the outcome of cannabis policy change when penalties were decreased but fell short of actual decriminalization. In this analysis, we estimated the outcome of policy change in 5 states (Kansas, Kentucky, Louisiana, Nevada, and New York). All other methods were the same as in our main analyses (including the same controls).
Table 1 lists the policies for each state, their date of implementation, and differences between provisions for adults and minors. Decriminalization policies typically applied to both youths and adults with only slight differences (eg, requiring attendance at a drug education program). Delaware is a notable exception; youth possession of cannabis remained a criminal offense even after adult decriminalization. Rhode Island deviates somewhat in that possession by youths younger than 17 years is automatically treated as a status offense (ie, “wayward” child) handled by the juvenile court system. Vermont also decriminalized only the first 2 possession offenses for youths until May 31, 2016, after which subsequent offenses were civil infractions.
Most states that legalized adult cannabis use treated youths very differently. Alaska, Colorado, and Washington continued to treat all youth possession of cannabis as a criminal offense after adult use was legalized (although these are typically among the least-severe criminal penalties in each state—especially in Colorado, which has also decreased penalties for youths over time). Youth possession of 1 oz or less of cannabis is a civil infraction in Oregon. However, the penalties for more than 28 g (1 oz) are different for adults and minors; possession of between 28 and 56 g (1-2 oz) is a civil infraction for adults but is treated as a misdemeanor for minors.
There was an abrupt decrease in the adult arrest rate associated with the timing of legalization of cannabis in Alaska, Colorado, Oregon, and Washington, but the youth arrest rate essentially stagnated after legalization (Figure, A). However, youth and adult arrest rates seem to have responded similarly to decriminalization (Figure, B). The youth arrest rate in states with cannabis legalization policies at 3 years after implementation also appears to be quite similar to the youth arrest rate in control states by 2016 (Figure, C). The states with legalization policies are also the only states in which the adult arrest rate decreases below the youth arrest rate.
Decriminalization was associated with large, statistically significant decreases in the arrest rate for both youths and adults, while legalization benefited only the adult arrest rate. On average, the adult arrest rate decreased by 131.28 (95% CI, 106.23-154.21) per 100 000 population after the implementation of decriminalization and 168.50 (95% CI, 158.64-229.65) per 100 000 population after the implementation of legalization (Table 2; see eTable 2 in the Supplement for full models). The arrest rate for youth decreased by 60 (95% CI, 42-75) per 100 000 population after decriminalization; the reduction of 7 (95% CI, –15 to 30) per 100 000 after legalization was not statistically significant. Several covariates were also significantly associated with the arrest rate in the analyses of youths: state poverty rate (35.27; 95% CI, 7.07-63.51), political ideology (0.66; 95% CI, 0.09-1.22), and per capita police rate (2.67; 95% CI, 1.35-3.98). In the adult analyses, the percentage of 30- to 49-year-olds (–37.49; 95% CI, –71.16 to –3.81) and the state poverty rate (52.32; 95% CI, 0.87-103.77) were significant covariates. In our supplemental analyses, other policy changes were not significantly associated with arrests for youths (–18.38; 95% CI, –2.42 to 39.18) or adults (–39.11; 95% CI, –83.39 to 5.18).
Our results suggest that legalization, as implemented in Alaska, Colorado, Oregon, and Washington through 2016, did not reduce arrests for nonviolent possession of cannabis among youths, despite having benefited adults. As depicted in Figure, A, there were sharp reductions in the adult arrest rate in states that legalized adult use of cannabis without a similar change in arrests for minors. Figure, B, shows a similar response in both youth and adult arrest rates after decriminalization of cannabis. This finding is further supported by our regression analyses, which, after controlling for a variety of state-level factors known to be associated with risk for arrest, failed to show a significant policy effect for youth after legalization. Of the covariates that we examined, state poverty rate exhibited a strong association with rates of arrest for possession of cannabis for both adults and youths, even after controlling for cannabis policy changes. For youths, the number of per capita police in a state was also meaningfully associated with the arrest rate, but we did not observe this finding for adults.
Our findings are not necessarily surprising—legalization of cannabis does explicitly target adult behavior and, in most cases, youth possession remained a criminal offense. However, focusing on adults to the exclusion of youths also runs counter to the spirit of legalization, which is motivated by the desire to reduce criminal penalties for cannabis use. This issue raises an important question: Why did Alaska, Colorado, and Washington keep criminal penalties for youths after legalization when most other states’ policies decriminalized adult and youth possession? It is possible that this is an artifact of the ballot initiative process, which has a history of preferencing the interests of the majority over those of minority groups (in this case, adult interests over those of young people under the voting age).24 Another possible explanation is that these penalties were left in place to appease political opposition to legalization, which was often based on the potential outcome on youths.
Similarly, cannabis possession was only decriminalized for adults in Delaware. This situation is comparable to that in Alaska, Colorado, and Washington, in which adults and youths are treated fundamentally differently by the law. If the goal is to reduce the legal consequences of cannabis possession for everyone, the obvious policy answer would be to specifically decriminalize youth possession arrests legislatively.
However, there were also no changes in youth arrests in Oregon, where possession of less than 28 g (1 oz) of cannabis was decriminalized for youths after adult use was legalized. This finding suggests a possible lack of benefit for youths even when they were included in the policy change. Other policy features could be to blame; for example, there is a clear disparity in how adults and youths are punished for possession of 28 to 56 g (1-2 oz) of cannabis in Oregon. But legalization could also simply be a different phenomenon with respect to youths. It is possible that adult legalization refocuses police attention on youths by reinforcing the prohibition on youth use, even when possession by youths has been decriminalized. This possibility is consistent with prior research suggesting that police interact with youths and adults differently and that arrest is the preferred strategy of officers when interacting with juvenile suspects.25 It is likely that other, less-obvious policy recommendations would be needed to address this disparity. For example, police typically retain the discretion to arrest if they suspect criminal activity, which could be a factor if it is difficult for an officer to differentiate between criminal and noncriminal possession of cannabis (eg, penalties increase with multiple offenses, or if the threshold amount for noncriminal possession is low). In this case, states should specify that individuals should not be arrested for possession of cannabis under the decriminalized amount, such as in Rhode Island.
Our study has several limitations. Most important, we assume that any changes in arrest rates are primarily owing to policy change. This assumption is supported by the individual state trends, which in most cases show a clear association between the timing of policy implementation and changes in arrest rate. We also assume that changes in arrest rates reflect differences in police behavior, rather than a systematic change in use of cannabis, but this assumption is supported by past work suggesting that decriminalization of cannabis has not led to increases in youth use.6 That arrest rates decreased so sharply for adults under both policies and for youths in decriminalization states offers further support for this assumption (ie, it is unlikely that cannabis use decreased drastically in response to policy change while police behavior remained unchanged).
Our findings suggest that, while youths strongly benefit from cannabis decriminalization when the same (or similar) provisions apply to both youths and adults, that benefit could be lost when adult use has been legalized. Our results are based on past implementation in specific states and might not reflect experiences for youths when adult use is legalized elsewhere, but there are still important implications for future policy in other states. The possibility that decriminalization could be associated with youth arrests differently when adult cannabis use is legalized should not be overlooked, especially if there are relatively straightforward solutions. For example, the AAP could safely stipulate in its recommendations that decriminalization of cannabis for youths should accompany the legalization of recreational use for adults (although the AAP has a specific recommendation for decriminalization, other language, which “strongly recommends strict enforcement of rules and regulations that limit access and marketing and advertising to youth” in states that have legalized cannabis likely implies the opposite).1(p586) There is also likely little cost to society associated with a policy that explicitly states when someone can be arrested for possession of cannabis under the decriminalized threshold (eg, only when someone refuses to provide identification or his or her name, address, and date of birth). These provisions would ultimately be in keeping with the goals of reducing criminal penalties associated with cannabis use and would also likely offer far greater protection for youths.
Accepted for Publication: March 12, 2019.
Corresponding Author: Andrew D. Plunk, PhD, MPH, Department of Pediatrics, Eastern Virginia Medical School, 855 Brambleton Ave, E. V. Williams Hall, Norfolk, VA 23510 (email@example.com).
Published Online: June 17, 2019. doi:10.1001/jamapediatrics.2019.1539
Author Contributions: Dr Plunk had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Plunk, Grucza.
Acquisition, analysis, or interpretation of data: Plunk, Peglow, Harrell.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: Peglow, Harrell, Grucza.
Statistical analysis: Plunk.
Obtained funding: Plunk.
Administrative, technical, or material support: Plunk, Peglow, Grucza.
Conflict of Interest Disclosures: Dr Plunk reported receiving grants from the National Institutes of Health during the conduct of the study. Dr Harrell reported receiving grants from the National Institute on Drug Abuse during the conduct of the study and personal fees from Moffitt Cancer Center outside the submitted work. Dr Grucza reported receiving grants, personal fees, and honoraria from the National Institutes of Health, and holding stock in Allergan outside the submitted work. No other disclosures were reported.
Funding/Support: This work was supported by grants DA046757, DA042195, DA040411, and DA031288 from the National Institute on Drug Abuse.
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Committee on Substance Abuse, Committee on Adolescence; Committee on Substance Abuse Committee on Adolescence. The impact of marijuana policies on youth: clinical, research, and legal update. Pediatrics
. 2015;135(3):584-587. doi:10.1542/peds.2014-4146PubMedGoogle ScholarCrossref
A; American Academy of Pediatrics Committee on Substance Abuse; American Academy of Pediatrics Committee on Adolescence. Legalization of marijuana: potential impact on youth. Pediatrics
. 2004;113(6):1825-1826. doi:10.1542/peds.113.6.e632PubMedGoogle ScholarCrossref
JM. Econometric Analysis of Cross Section and Panel Data. 2nd ed. Cambridge, MA: MIT Press; 2010.
DL. Robust inference with multi-way clustering. NBER technical working paper no. 327. https://www.nber.org/papers/t0327
. Published September 2006. Accessed May 8, 2019. .
R Development Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2018.
D. Structuring the ballot initiative: procedures that do and don’t work. Univ Colo Law Rev
. 1994;66:47.Google Scholar