Daniel J Mallinson, A Lee Hannah, Policy and Political Learning: The Development of Medical Marijuana Policies in the States, Publius: The Journal of Federalism, Volume 50, Issue 3, Summer 2020, Pages 344–369, https://doi.org/10.1093/publius/pjaa006
Navbar Search Filter Mobile Enter search term Search Navbar Search Filter Enter search term SearchPolicy diffusion studies often infer that learning occurs, but statistical analyses cannot demonstrate it definitively. The spread of medical marijuana offers the opportunity to take a closer look at whether policy and political learning occur during diffusion. An increasing number of states have adopted medical marijuana policies in defiance of federal prohibition and in the space created by federal inactivity. Furthermore, early adopting states have adapted their programs to account for changes in the industry and to coincide with recreational marijuana programs. This article sheds light on how the laws have evolved over time due to policy learning, political learning, and local adaptation. Specifically, we review how states have incorporated best practices from others (policy learning) and how the laws have been repackaged in more politically conservative states (political learning). Finally, we show how states adapt to medical marijuana laws by using precedents from the regulation of other industries.
A purported benefit of the American federal system is that innovative states can experiment with new solutions to prevailing policy problems. Other states then learn from the successes and failures of innovator states when adopting their own version of the policy. Though, policy learning is not the only learning process that occurs as a policy spreads ( Gilardi 2010). States also learn about the political ramifications of the policies that they are considering ( Grossback, Nicholson-Crotty, and Peterson 2004) and legislators further adapt them to match the contours of local demands ( Taylor et al. 2012). Policy diffusion research seeks to understand these dynamics, though quantitative models of diffusion typically test variables that imply a learning process ( Berry and Baybeck 2005) and thus empirical evidence of policy learning is limited ( Volden, Ting, and Carpenter 2008). Without studying the emergence and modification of policies as they are written and implemented, it is difficult to fully understand how, and what kind of, learning is occurring.
Medical marijuana policy is fruitful ground for testing the assumption that learning occurs as a policy diffuses. While there have been changes in U.S. Department of Justice (DOJ) policy and proposed Congressional legislation addressing federal marijuana prohibition, state governments remain the primary engines of policy activity. Given the continued importance of state policymaking in this area and the many challenges that states needed to work through to implement effective medical marijuana programs, it is possible to account for the presence of learning.
Substantial variation exists across the thirty-three state medical marijuana programs ( Klieger et al. 2017; Pacula, Hunt, and Boustead 2014). For example, programs became increasingly restrictive as they were adopted by more states ( Pacula and Smart 2017, Williams et al. 2016). Recent laws limit smoking in favor of vaping ( Hannah 2018), restrict home growing ( Pacula and Smart 2017), establish dispensaries ( Mosher and Akins 2019), and allow localities to further control medical marijuana producers and businesses. What is not clear, however, is what roles policy learning, political learning, and local adaptation played in these changes over time. While existing diffusion research has found patterns of political learning for medical marijuana ( Hannah and Mallinson 2018), it has not examined how all three forces shape specific provisions.
In this article, we examine medical marijuana programs adopted or amended by states between 2009 and 2018, the most recent era of adoption ( Pacula and Smart 2017). The substantial leniency provided to state medical marijuana programs by the federal DOJ in 2009 marked a substantive shift in not only newly developed programs, but also older programs that had to respond to new market pressures. Subsequently, both types of states had to learn new lessons from this market and update their programs. We describe the extent to which new and amended programs reflect policy and political lessons from earlier-adopted programs, as well as the ways in which states adapted medical marijuana programs to fit the proclivities of their local citizens and existing state law. We begin by providing necessary context on policy diffusion theory as well as the history of state medical marijuana adoption. We then address examples of policy learning, political learning, and local adaptation that occurred during the diffusion of medical marijuana. We conclude with the implications of our results for the study of policy learning and diffusion.
It seems that Supreme Court Justice Louis Brandeis’s framing of states as “laboratories of democracy” is a required reference for any policy diffusion article. Brandeis implied that states experiment with new ideas, spreading successes and abandoning failures. While many tip their hat to this notion, identifying when policy learning occurs proves difficult ( Rom 2006). This is often because diffusion research has been dominated by single policy event history models that explain the spatial and temporal patterns of adoption. It is difficult for such models to concretely demonstrate the mechanism underlying a policy’s spread.
Scholars differentiate three types of learning that occur during diffusion: instrumental, social, and political. The form directly implied by Brandeis is rational, or instrumental, learning whereby legislators observe the positive and negative outcomes of policy experiments in other states, accordingly update their prior beliefs regarding the attractiveness of those policies, and then either decide to adopt, abandon, or adjust the innovation ( Meseguer 2005, 2006; May 1992). Such learning is boundedly rational, whereby legislators look to neighboring states for relevant and representative information regarding the policy ( Meseguer 2005). This tends to occur quickly at the beginning of the diffusion process, but resistance to learning from other states can build over time as a state gains experience with the policy problem ( Meseguer 2006; Mooney 2001). Social learning focuses on the success or failure in the social construction of the policy problem, scope of policy, and policy goals ( May 1992; Boushey 2016). Finally, political learning draws lessons about the policy process and expected political prospects of a policy ( May 1992), which can be signaled via the ideological disposition of adopting states ( Grossback, Nicholson-Crotty, and Peterson 2004). Recent evidence suggests that such learning is of increasing relative importance, specifically as regional learning has declined over time (Mallinson forthcoming). Of course, legislators also adapt policies to suit their local constituencies, the distribution of whom may not be identical to other states ( Taylor et al. 2012).
Alas, researchers have struggled to effectively account for when learning is occurring, and thus to systematically understand its dynamics. Volden, Ting, and Carpenter (2008) offered a powerful critique of diffusion research, demonstrating that using geographic diffusion patterns alone as the marker of learning potentially results in substantial false positives where states are independently converging on a single policy solution for solving a problem, not learning from each other. They argue that states should be adopting objectively successful policies in order to show learning taking place. Others argue that adaptation due to failed policies is also a hallmark of policy learning ( Adam 2016; May 1992; Braun and Gilardi 2006). Neighbor adoptions due to learning are also difficult to disentangle from diffusion due to regional competition ( Berry and Baybeck 2005).
Recent research pivoted to answering the question of what pathways are necessary for learning to occur. Answers include interest groups serving as vectors of policy transmission ( Boushey 2010); “peer” states, whose definition is based on distance, similarity, and capacity ( Butler et al. 2017; Einstein, Glick, and Palmer 2019); leadership transfer networks resulting from public manager career movement ( Yi and Chen 2019; Yi, Berry, and Chen 2018); site visits ( Ma 2017); joint membership in intergovernmental organizations ( Füglister 2012); and cross-border relationships among technical staff ( Nicholson-Crotty and Carley 2018). All are potential sources of policy, social, and political learning; however, a substantial challenge for many studies is that they can only infer learning from statistical models ( Nicholson-Crotty and Carley 2018). Though some research examines the spread of policy components ( Boehmke 2009), there is little deep forensic analysis of when and how learning occurs during policy diffusion. Medical marijuana policy offers an opportunity to take such a dive and glean insights for our understanding of learning in a federal system.
While the federal government’s position on state marijuana liberalization changes with each presidential administration, new DOJ signals, and limited Congressional action, state governments continue to be the primary engine of marijuana policy activity. Little has changed from the standpoint of the federal government since Kamin (2015) surveyed the “uneasy status quo” of marijuana policy. While proponents feared that the Trump Administration and prohibitionist Attorney General Jeff Sessions would disrupt the spread of marijuana, Sessions’ actions were limited to a change in rhetoric, not enforcement ( Waldrep 2018). The 116th Congress (2019–2020), especially the Democrat-controlled House, has taken up bills to address banking, access, decriminalization, and restorative justice ( Angell 2019), but there is uncertainty that these policies will materialize ( Demko and Fertig 2019).
With this gap in direction from the federal government, states continue to adopt and innovate. Modern medical marijuana laws started diffusing in 1996 with voter approval of Proposition 215 (the Compassionate Use Act) in California. Pacula and Smart (2017) demarcate three periods of adoption: the ballot era (1996–2000), the early legislative era (2000–2009), and the late legislative era (2009–present). Figure 1 displays the states that adopted during each era. The early adoptions occurred in states with direct democracy and liberal populations ( Hannah and Mallinson 2018). Many of the early laws were ambiguous and left patients, caregivers, and doctors uncertain of their rights ( Pacula and Smart 2017). State legislatures began to adopt new medical marijuana policies in the early 2000s and many legislatures in early adopting initiative states revisited laws and added regulations.
Map of States with Comprehensive Medical Marijuana Laws.
Our analysis of state medical marijuana policy from 2009 to 2018 marks the time in which federal prohibition enforcement took a substantial turn. Even though medical marijuana policies spread to five states from 2004 to 2008, the Bush Administration enforced federal prohibition and created a “culture of fear” in the industry and among the states ( Trumble and Kasai 2017). The Obama Administration oversaw a marked shift in the federal government’s stance and enforcement focus. In March 2009, Attorney General Eric Holder announced an end to federal raids on distributors in well-regulated medical marijuana states ( Stout and Moore 2009). Then, in October, the Ogden Memo directed DOJ attorneys to deprioritize prosecuting users and suppliers in states with medical marijuana policies ( Ogden 2009). These actions presented a key shift in medical marijuana policy, as states were previously reluctant to regulate the emergent industry for fear of violating federal prohibition. While the Obama Administration did not take a position on expanding marijuana programs, subsequent states adopted and old adopters moved to regulate the supply of marijuana through dispensaries ( Pacula and Smart 2017). The programs also became increasingly medicalized, mirroring the practices of medical care and pharmaceutical regulation ( Williams et al. 2016). A new marijuana industry took off despite the administration’s efforts to limit the commercialization of medical marijuana supply ( Hollander and Patapan 2016). In contrast, weakly regulated states had to contend with an increase in DOJ enforcement activity ( Cole 2011), even over the Bush Administration ( Trumble and Kasai 2017; Hannah and Mallinson 2018). This pushed old and new adopting states alike to learn from the experiences of early adopters and develop firmer regulations of the industry. The following discussion considers evidence of policy and political learning and local adaptation as these policies spread.
States had much to learn when establishing their medical marijuana programs. Though there were state-level decriminalization efforts in the 1970s and 1980s, effective medical programs are more complex than simply reducing penalties for possession. A truly medicalized marijuana program requires states to replicate federal regulations over drug manufacturing, distribution, and marketed use. That is, states had to become their own Food and Drug Administrations (FDA) for marijuana, if they wanted safe and effective programs. Due to prohibition under the Controlled Substances Act, states do not have the regulatory support provided by the federal government for pharmaceuticals.
In the first era of medical marijuana program adoption, when much of the adopting activity was driven by direct initiatives instead of by legislators, programs were sparse in details. As recounted by Governing magazine, “In that time, California gained a reputation as something of the Wild West for weed: no state regulatory model, notoriously lax enforcement and an undefined set of prescription criteria that makes obtaining a medical marijuana card little more than a wink-wink formality” ( Scott 2012). California did not even pass an implementation bill for Proposition 215 (1996) until 2013 (the Medical Marijuana Program Act). Instead, states muddled through early implementation and many advances were shaped by intrastate court cases as well as changes in DOJ policy. For example, Colorado moved both toward and away from allowing for dispensaries (as opposed to home cultivation) before statutorily establishing a regulatory system for dispensaries in 2010. In Washington state, home cultivation was allowed, but dispensaries were not and there was no regulatory control over the marijuana supply chain. This led to two primary means of obtaining medical marijuana in the state: home cultivation or the black market ( Cambron, Guttmannova, and Fleming 2017). Eventually, the Ogden (2009) and Cole (2011) Memos paved the way for states to enact clear regulations. Before these memos were issued, states were careful not to insert themselves too deeply into these new programs. In fact, the DOJ under Bush directly threatened health department workers with federal prosecution in Arizona and Washington ( Cambron, Guttmannova, and Fleming 2017), thereby chilling state involvement in marijuana regulation. New DOJ guidance under Obama encouraged the implementation of state regulatory frameworks as the federal government promised to not prosecute members of the industry in states that were well regulated. Thus for suppliers, the “legal code [became their] only line of defense” ( Scott 2012).
Colorado is a key leader in marijuana policy. After the Ogden Memo, Colorado was the first state to enact comprehensive regulations of medical marijuana. Alcohol and tobacco regulations served as early models for marijuana regulation ( Klieger et al. 2017). Colorado’s regulations then served as a framework used and adapted by other states (e.g., Connecticut) that often adopted stricter regulatory regimes ( Scott 2012). Thus, Colorado is an important reference point when discussing how states engaged in policy and political learning.
Medical marijuana laws are complex and multifaceted ( Pacula and Smart 2017). Therefore, it is not possible to assess all components for the presence of policy learning. However, the following discussion considers three major aspects where states had to learn from each other and their own experiences in order to solve implementation problems: (1) providing marijuana, (2) regulating the supply chain, and (3) packaging and safety.
A substantial challenge in formulating medical marijuana policy is determining how the state will be involved in the production and sale of an illicit substance. The first option is to allow patients to grow their own marijuana plants and then allow them to turn over the plants to processors. This has a host of drawbacks. First, growing marijuana takes a substantial investment in time, effort, and money. Not all patients or their caregivers are well suited to have the grow lights and conditions necessary in their home to produce the product. Hence patients still turned to the black market even after medical marijuana was legalized. Further, home cultivation is susceptible to diversion to the black market ( O’Keefe 2013; Caulkins et al. 2012). Dispensaries, however, can serve an important public health purpose as they are easier to regulate ( Borodovsky and Budney 2017). Furthermore, dispensaries legitimize the use of marijuana as a medicine ( Pacula et al. 2015). They are part of the “medicalization” process, whereby state regulatory regimes begin to reflect components of federal controls on pharmaceuticals. Further, dispensaries provide a stable supply that can meet increasing demand with greater economies of scale, more so than the fragmented supply through home cultivation ( Pacula and Sevigny 2014).
Colorado’s 2010 bill provided legal protections for dispensaries and, after being frightened off by DOJ threats, Washington state paved the way for dispensaries in 2015 ( Cambron, Guttmannova, and Fleming 2017). As of 2017, 89 percent of state medical marijuana programs protect and regulate dispensaries, though half still allow for some form of home cultivation ( Klieger et al. 2017). There is a clear trend towards states instituting greater controls of the supply chain, including regulating dispensaries and restricting home grow. However, this may also be a function of the commercialization of marijuana. Commercial operations that support state legalization efforts will often work against the availability of home grow as it pulls business away from commercial growers and can lead to a collapse in prices ( Gettman and Kennedy 2014; Caulkins et al. 2012).
Patient registries also demonstrate policy learning. Requiring patients to register with the state was uncommon in pre-2009 medical marijuana laws, but now all states require it ( Klieger et al. 2017). This is also part of the medicalization process. States desire to know who patients and doctors are, and registration allows for greater oversight. Relatedly, most states now allow for permit revocation if participating patients or physicians break the rules.
As the marijuana industry grew, states exerted more control over the system distributing the product ( Chapman et al. 2016). There are roughly six links in the supply chain for medical marijuana: the seeds, planting, harvesting, processing, formulating into final products, and dispensing. In the early days of home cultivation, patients or their caregivers took care of the first three stages and then processors took care of the rest. As the industry grew and commercialized, states exerted more control over this process. Some states, for example, require or encourage testing to ensure product quality. Many have also developed worksite regulations such as requiring security systems, locked shopping spaces, hand washing stations, ventilation systems, and proper waste disposal ( Klieger et al. 2017). These are all hallmarks of the institutionalization of the industry and the medicalization of state programs. As time progressed, new states tended to adopt these provisions and older adopting states updated their laws to include them.
The clear trend in states is toward greater control as they have learned lessons about safety and security in the industry, though this is not uniform. Ohio and Pennsylvania are good examples. Pennsylvania chose to integrate the planting, growing, harvesting, processing, and producing functions into a single license. Dispensing is then a separate license. By integrating the whole supply chain, except for dispensing, into a single grower-processor, Pennsylvania’s program is less complicated to oversee and control. Ohio, which adopted medical marijuana in the same year as Pennsylvania (2016), chose to have separate licenses. Ohio allows businesses to be vertically integrated but does not require it. Of the twelve large-scale provisional licensees, all completed separate applications for processor licenses and eight for dispensaries ( Borchardt 2017). 1 Neighboring West Virginia had to revisit its medical marijuana bill, passed in 2017, after studies showed that businesses could only be profitable if they were vertically integrated ( Beard 2019). Thus, while some states have learned from oversight challenges in earlier adopting states, others locally adapt due to political concerns, which we discuss below.
Another hallmark of the medicalization process is the effort to protect consumer safety. When there were few controls over production and dispensing, there was likewise little control over the final product. But as states moved to formalize their programs they also had to grapple with the threat of unsafe or mislabeled products ending up in the hands of patients. Again, in this respect, the states were forced to replace the role of federal regulators like the FDA. To that end, states turned to a variety of requirements including plain labels, opaque containers, childproofing, tamper proofing, re-closable containers, a warning against operating heavy machinery, a warning for medical use only, a list of health risks, potency, and proof of testing ( Borodovsky and Budney 2017; Klieger et al. 2017). These changes are like the difference between being handed a few pills with no information and a standard orange plastic pharmaceutical tube. States essentially moved over time towards requiring standard pharmaceutical practices as they learned how to be drug regulators. Most states have also instituted product safety testing throughout the supply chain ( Klieger et al. 2017, Ghosh et al. 2015).
Another area of policy learning is in the realm of advertising. As medical, and eventually recreational, programs were adopted, new products flooded the market. Companies used standard methods to advertise these products—flashy packaging, pictures of marijuana, catchy names (e.g., Pot Tarts), and more. States quickly noted increasing trends in emergency room visits, particularly by children who unknowingly consumed marijuana-filled products ( Wang et al. 2016). In response, states have experimented with a variety of controls including restricting anything designed to attract minors, cartoons, depictions of minors, pictures of marijuana, resemblance to candy, and making false claims. Data on these requirements demonstrate a trend toward greater packaging and safety regulation ( Klieger et al. 2017).
While much of the preceding discussion considers how trends in specific policy components suggest the occurrence of policy learning, it is difficult to pinpoint exactly who is learning from whom. Official delegations, however, offer glimpses into how governments learn from each other ( Ma 2017). Over the last six years, a broad array of states visited Colorado to learn from the state’s adoption and regulation of medical and recreational marijuana ( Wallace 2016). Such visits often include tours of dispensing facilities, meetings with regulators, and even sampling of products ( Maulbetsch 2018; Hefler 2016). Pennsylvania Senator Daylin Leach made an official visit during his push to adopt medical marijuana ( Shumway 2014). Colorado officials also travel to other states to share lessons learned. The learning purpose was well captured by Colorado’s Director of Marijuana Enforcement Division of the Department of Revenue during his visit to a Virginia Cannabis Summit in 2019: “Virginia has an opportunity to look at states like Illinois and Colorado that are here at the table to understand some of those lessons learned, to capture those best practices and ideally develop practices that are informed by what we’ve done well and where we faced some challenges” ( Austermuhle 2019). In fact, Colorado’s former Director of the Marijuana Enforcement Division, Andrew Freedman, is now a much sought out consultant on marijuana policy for state and local governments ( Wogan 2017).
Interest groups also push specific lessons with model legislation. Model legislation from the Marijuana Policy Project (MPP) includes the statement, “Thirty-three states and the District of Columbia have removed state-level criminal penalties from the medical use and cultivation of cannabis. [Insert State] joins in the effort for the health and welfare of its citizens.” 2 Americans for Safe Access goes a step further and notes the merits of policy learning: “… more than 20 years of state-level experimentation provides a guide for state and federal law and policy related to the medical use of cannabis.”
States also learn about the political ramifications policies and adapt them to match the contours of citizen demands ( Grossback, Nicholson-Crotty, and Peterson 2004). Identifying political learning can be difficult because two factors occurred simultaneously: (1) attitudes about medical and recreational marijuana quickly liberalized due to intracohort changes ( Felson, Adamczyk, and Thomas 2019) and (2) policy entrepreneurs worked to reframe marijuana to increase the likelihood of adoption ( Hannah 2018; Kim and Kim 2018). Moreover, as more states adopt medical marijuana and recreational policies, there may be bandwagon effects. It has been long noted in the diffusion literature that an innovation gains legitimacy as it spreads and eventually becomes “something states ought to have” ( Walker 1969, 890). A recent editorial in the Kansas City Star put it bluntly, “While the medical marijuana bus left the station long ago in most states, Kansas lawmakers are blithely telling sick Kansans to walk on without it” ( Kansas City Star Editorial Board 2020). There is no uniform threshold for when this effect occurs but the fact that a bevy of conservative states (mainly in the South and Midwest) are seriously considering medical marijuana programs suggests that the tipping point may have been reached for medical marijuana. It is a general pattern that as more ideologically similar states adopt the innovation, it places greater pressure on late adopters and laggards, particularly, to do likewise. Political learning occurred both terms of the institutional strategies used for passage and how bills were written.
Early medical marijuana policies were largely passed via ballot initiative. This allowed interest groups to shape the policy environment without stonewalling from state legislatures and governors ( Ferraiolo 2007, 2008). Seven of the first eight medical marijuana laws were passed this way: Alaska, California, Colorado, Maine, Nevada, Oregon, and Washington. Over time, legislatures started to pick up the issue, but with the most liberal states in the lead ( Hannah and Mallinson 2018). The initiative process has also been used to facilitate adoption in more conservative states recently (Oklahoma, Utah, and Missouri in 2018) and an initiative threat pushed legislators to act in Ohio in 2016 ( Hannah 2018). As Ferraiolo (2017, 390) notes, direct democracy is effective as a policy tool for “pushback federalism.” Figure 2 displays the cumulative frequency of adoptions over time. The figure shows that early adoptions largely occurred via ballot initiative. Meanwhile, starting in 2009, most adoptions occurred via legislatures. After 2016, adoptions were carried out equally in both venues. Crossing over from initiative adoption to legislative is a learning task for both interest groups and state legislators. Successful ballot initiatives in early adopting states send clear political signals regarding the issue’s palatability, but do not provide information for how to frame the issue in such a way that it can be guided through a compromise-oriented legislative process. States like Hawaii paved the way for future legislative adoptions of medical marijuana by working out the thorny details.
Cumulative Frequency of Medical Marijuana Law Adoptions Over Time.
The framing that advocates adopt in supporting medical marijuana policy adoption has also changed. Hannah (2018) found that the newspaper framing of conditions associated with medical marijuana laws has changed substantially. In the early adopting years, coverage focused on AIDS and cancer patients. Over time, other conditions became salient, such as marijuana’s impact on veteran populations with PTSD and children with epilepsy. Meanwhile, perhaps due to the geographical spread of the policy and the onset of the opioid crisis, opioids also became a prominent condition discussed in newspaper coverage of medical marijuana. Four states have permitted doctors to recommend medical marijuana instead of opioids for pain: New Mexico, New Jersey, New York, and Pennsylvania. In another study that looked at framing around marijuana policy—both medical and recreational— Kim and Kim (2018) find that early newspaper articles focused on law enforcement challenges and the risks of youth drug use. However, coverage began to focus on the economic effects of marijuana starting in 2009, which coincides with the Ogden Memo. While the tax revenue from medical marijuana has garnered some attention ( Cooper 2012), recreational marijuana which tends to draw more coverage regarding the economic benefits. In the case of medical marijuana, legislators and proponents must be cautious about discussing tax revenue considering the other purpose of treating serious medical conditions.
This new language around marijuana is also utilized by lawmakers. Ohio State Senator Kenny Yuko noted these changes from the Ohio Senate floor:
Well a person my age, when you heard of marijuana back in those days, back in 2003, When I heard two words associated with marijuana, it wasn’t “medical” and ‘marijuana”. It was more like “Cheech” and “Chong”, “Snoop” and “Dog”, and “Willie” and “Nelson”… The reality of it is, once I did a little research, I found out how wrong I was…This bill is not perfect folks, but it’s what Ohio patient [sic] needs. It’s not acceptable to make them wait any longer. If we can give just one veteran comfort, if we can ease one patient’s horrible pain, if we can maybe improve or even save a child’s life, this bill will be worth it. (Ohio Senate, 2016, Third Consideration Hearing for H.B. 523, 131 st General Assembly 2 nd sess., May 24. Quoted in Hannah 2018, 50)
Former Speaker of the House, John Boehner, reversed his position on marijuana stating, “I have concluded descheduling the drug is needed so that we can do research and allow [Veteran’s Affairs] to offer it as a treatment option in the fight against the opioid epidemic that is ravaging our communities” ( Horton and Ingraham 2018). Additionally, advocates within specific communities have worked to change the narrative, including veterans ( Jonsson 2018), parents of children with seizure disorders ( Cha 2014), and those tackling opioid abuse ( Jones 2019). In these cases, advocates are leveraging positively constructed groups, some of which have high power (veterans) and others low power (children and parents) and even a negatively constructed group with lower power (opioid drug users) to expand the scope of conflict and argue for conferring the benefits of medical marijuana onto the groups ( Schneider and Ingram 1993).
Such tactics have helped to open policy windows in later adopting conservative states. Spetz et al. (2019) found that the Marijuana Policy Project (MPP) donated primarily to Democrats prior to 2014 but then transitioned to sending a greater share of their donations to Republicans. Other cannabis industry leaders have increased donations to Republicans ( Hughes 2018) while also packaging marijuana as a states’ rights issue ( Friedersdorf 2018). Thus, advocates have shifted from a party-specific donation strategy to one that better resembles the bipartisan access-buying that occurs in other policy areas ( Lowery and Brasher 2004). Doing so is important for gaining legislative access in more conservative states where policies may be driven by Democrats, but require substantial Republican support for success (e.g., Pennsylvania). Thus, changes in framing have occurred as both legislators and advocates learned what strategies are most successful.
As noted above, most states permitted patients to grow medical marijuana at home prior to 2009. Many early adopting states have tried to rein in lax home grow rules. For example, Colorado Governor John Hickenlooper signed a law in June of 2017 that would curb the “Wild West” of home grow and limit the number of plants a home grower can cultivate for medical purposes from ninety-nine per patient (up to 600 in a home) to no more than twelve for a residence ( Vestal 2017). Oregon and Washington followed suit. Meanwhile, late adopters have largely prohibited home grow—particularly when adopting through the legislature. Of the laws passed prior to 2018, home cultivation was permitted in 85 percent of states that passed MMLs by initiative and only 27 percent of states that passed by legislature ( PDAPS 2019). This is reflective of legislators responding to the growing commercialization of marijuana. Commercial producers do not want home grow as it cuts into their market share ( Jaeger 2019). State officials have had to respond to these pressures, and many have limited home grow.
It appears that trends in restricting home cultivation are specific to medical marijuana. States that have taken up recreational marijuana have been more permissive of home cultivation (Marijuana Policy Project n.d.). Also, of the three states that adopted medical marijuana laws in 2018, each included home cultivation (ProCon n.d.). However, Utah legislators rewrote the law to remove home cultivation and thus exert greater central control over marijuana with a state-run cannabis dispensing system, which would have been unique among all the states ( Rodgers 2019a). Ultimately, Utah backed away from this approach in favor of privately owned pharmacies. The reasoning harkened back to Washington state’s hesitancy over state health department engagement with medical marijuana ( Cambron, Guttmannova, and Fleming 2017), namely the concern that the federal government would pursue department employees as drug dealers ( Rodgers 2019b). When the Illinois legislature approved recreational marijuana in 2019, it also amended their medical marijuana law to permit home cultivation, a model that states like New York are also considering ( Associated Press 2019).
States regulate where and how patients can consume marijuana. Because medical marijuana policy is a morality policy ( Hollander and Patapan 2016), and given marijuana’s legal and moral status in recent history ( Mosher and Akins 2019), restrictions on individual use seem to be conditioned more by optics than by substance. As an Ohio legislator reasoned when explaining why the state was not adopting home cultivation or smoking, “smoking dope and growing dope, that’s not an Ohio thing” ( Balmert 2016). Recently, state laws have allowed for vaporizing, but not smoking marijuana. Smoking has been prohibited in recent legislative adoptions (New York, Minnesota, Ohio, Pennsylvania, and West Virginia) and Florida’s legislature tried to amend a ballot initiative to restrict smoking, only to relent after losing in the courts ( Griffin 2019). It appeared that vaping allowed legislators to have it both ways: provide medical access to an increasingly popular product without approving of behaviors associated with the war on drugs (i.e., smoking). However, this trend might change in light of recent concerns about vaping-related products ( Richtel 2019). As Belluz (2019) notes, “states haven’t been able to protect the public from dangerous vape products, and so it may be time for agencies experienced in regulating consumer products, like food and drugs, to take over, they say.” Thus, the recent crisis around vaping could serve as a backdoor to federal action on marijuana. Current legislatures considering adoption have also shed light on the concerns over vaping. In Kansas, legislators created a special committee to study Ohio’s medical marijuana program after citing reservations about the programs in neighboring states of Missouri and Oklahoma—which permit smoking marijuana and home cultivation. But legislators also voiced support for going a step further than Ohio and prohibiting vaping ( Shorman 2019).
Due to remaining stigma, states also vary in the extent to which they protect marijuana patients from discrimination. Employment protections vary across the states. According to Americans for Safe Access, Ohio’s law “includes the worst employment language in the country for patients” in giving employers discretion to screen for and prohibit medical use, even cutting off access to unemployment benefits to workers fired for medical marijuana use ( Hannah 2018). State courts have weighed in on employment discrimination cases, but there is little consensus. Ten states have enacted language that bars employment discrimination ( Massie 2019). This may change as stigma continues to decline and states learn how to develop employment regulations that balance the interests of employers and employees.
In any policy diffusion process, legislators not only respond to external signals regarding policy and political success, but they must also respond to internal signals from their constituencies. This means that even if a state adopts an innovation, it adjusts to ensure the policy is palatable to citizens, interest groups, businesses, etc. Adjustments also arise from differing states approaches to local government power and state regulatory control. Local adaptation is evident in the spread of medical marijuana laws. There are few components of these laws that have been incorporated by all adopting states ( Klieger et al. 2017). In some cases, variation in the laws is due to policy or political learning that occurred in some jurisdictions, but not all. In other cases, states adapted the law for their own proclivities.
The regulation of locally unwanted land uses (LULUs) tends to demonstrate a substantial amount of local adaptation. Restrictions on the siting of vice (e.g., liquor stores, strip clubs) and other disturbances (e.g., oil and gas drilling rigs) often include setback requirements that keep them away from sensitive public spaces (e.g., schools), commercial zoning that keeps them out of residential neighborhoods, and density controls that limit their concentration ( Németh and Ross 2014). These controls, however, tend to push LULUs into areas with high densities of persons of color and the poor. Medical marijuana production and dispensary facilities are often regulated as LULUs and thus dispensary rules exhibit similar patterns of variation due to local adaptation.
States have moved in two separate directions regarding whether they allow local governments to regulate dispensaries. Some exert greater state-level control, whereas others allow local governments to decide whether they want a dispensary in their jurisdiction. Florida, for instance, has preempted local-level regulations of dispensaries. Meanwhile, activists in Missouri were hopeful that a recent constitutional amendment passed by voters incorporated lessons learned from the slow implementation process in Illinois. In particular, the amendment centralized the regulation of dispensaries and allowed for the licensing of 192 dispensaries, nearly 3.5 times as many as Illinois ( Regnier 2018).
Pennsylvania and Ohio serve as informative cases for how internal forces shape the amount of local control over dispensary siting. Ohio allows local governments the option of banning the presence of growers, processors, and dispensaries. The state law prohibits marijuana facilities within 500 feet of schools, churches, public parks, public playgrounds, and public libraries. Twenty-one states have similar restrictions, though the specific setback rules vary greatly ( Klieger et al. 2017). Ohio allows its local governments to add restrictions on top of those in the state law. This creates a regulatory floor, as opposed to a ceiling beyond which the local government cannot act. Cities like Akron have used this authority to create approval processes involving multiple actors (such as the City Planning Commission, City Council, and the Mayor), requirements to coordinate with the local police department on security, expansive inspection powers, and requirements for providing business information to the city (City of Akron n.d.). Under state law, however, local governments cannot restrict the rights of registered patients to use approved products and physicians to recommend marijuana within city limits.
Pennsylvania, which adopted medical marijuana in the same year as Ohio (2016), has exerted more centralized control over dispensary siting. Its law provides a 1,000-foot barrier between marijuana facilities and schools and daycare centers, which local governments can extend. The law also prevents dispensaries and grower/processors from inhabiting the same physical location and requires operation in an enclosed facility. Local governments are required, however, to regulate dispensaries like any other commercial facility and grower/processors as any other manufacturing, processing, and production facility ( Koresnoski 2017). Under the Pennsylvania Municipalities Planning Code, all municipalities must accommodate all legal uses in at least part of their territory ( Kraus 2016). This prevents municipalities from banning marijuana facilities, though they do have some control over where those facilities are sited.
The two neighboring states are more generally distinct in their approaches to local government. Ohio is a Dillon’s Rule state, meaning local governments are only empowered to do what the state specifically allows. Pennsylvania is highly fragmented with over 5,000 local governments, some of which have opted for home rule. Pennsylvania also provides wide latitude in zoning and other powers to local government. This was evident in the issue of hydraulic fracturing for natural gas where the Pennsylvania Supreme Court invalidated state efforts to supersede local regulation of natural gas development. Like marijuana, however, local governments are not allowed to ban oil and gas drilling, as it is a legal enterprise ( Clark 2017). It is in the nuances of these local-state relationships where at least part of the explanation for variation in dispensary regulations lies. In Ohio, which at the time of passage had single-party Republican control of government, local governments were empowered to expand their regulations of marijuana facilities, including adopting outright bans. In Pennsylvania, where the law was passed by the Republican General Assembly and a Democratic governor, the state restricted a typically expansive local zoning power by requiring municipalities to treat marijuana operations like any other commercial or industrial pursuit. Local governments also cannot ban marijuana facilities, as they are a legal enterprise under state law.
States differ vastly in how they determine the geographic dispersion of dispensaries. Some, like Pennsylvania and Ohio, divide the state into sectors and then distribute licenses for dispensaries, growers, and processors within those divisions. The goal of such a design is to ensure an even distribution of facilities across the state. Other states, like Colorado, do not place a cap on the number of dispensaries and thus their dispersion is market driven. California left the question of geographic distribution up to local governments who consequently struggled to regulate the industry. Spatial analyses confirm that dispensaries in California tend to cluster in commercial zones, poorer areas, areas with higher density of minorities, and close to alcohol shops ( Thomas and Freisthler 2016; Morrison et al. 2014).
The liberalization of marijuana policy has significant implications for intergovernmental relationships in the United States. It typifies the concern that such relations have become more contentious ( Rose and Goelzhauser 2018). In this case of defiant innovation ( Hannah and Mallinson 2018), states not only enable, but directly regulate and legitimize an industry that remains expressly prohibited by the federal government. Like any innovative policy, however, states had to work their way through both policy and political problems that emerged during the diffusion process. When considering simply the temporal ordering and spatial distribution of policy adoptions, diffusion scholars have a difficult time distinguishing when causal processes like learning, competition, coercion, and social contagion occur. The critique by Volden, Ting, and Carpenter (2008) that states could simply converging on a common solution to a problem without really being influenced by one another remains powerful. We, however, have sought to demonstrate that examining the component parts of a policy can illuminate whether learning occurs.
Our survey of the evolution of medical marijuana legislation suggests that both policy and political learning occurred in the diffusion of state medical marijuana policies. It also suggests that local adaptation occurred for aspects of policies that were related to existing state regulations. Even with examining in greater detail how states responded to policy and political problems over time, our analysis remains limited in providing definitive proof that learning occurred. Statutes and rulemakings can provide such evidence, but only if they explicitly denote when another state served as a model. Marijuana policy is replete with anecdotes to this effect, including state legislators discussing other states’ experiences with the policy, but a further systematic analysis of such indicators will take time and effort. Conversely, the absence of such explicit signaling does not mean that learning has not occurred.
We would argue that by identifying clear problems, how states solved those problems, and demonstrating how states appear to be converging on particular solutions to those problems with specific components of their policies is a stronger indicator of learning than is possible in standard event history diffusion models that assume all policies are equal in their content. When convergence occurs in policy components that address identifiable policy or political problems it is unlikely that this is simply due to chance. It is more likely that states have identified workable solutions to those problems and the solutions are spreading. Furthermore, learning is the only way to understand why so many states have sent delegations or individual legislators to Colorado, or invited Colorado officials to their state, to draw lessons from the implementation of medicinal and recreational marijuana. Visits and meetings include tours of dispensaries and discussions of Colorado’s regulatory framework. The state has become a commonly recognized “mentor” to states considering marijuana liberalization ( Wallace 2016).
By looking at the components of a policy, it is also undeniable that learning during the diffusion of an innovation is not a linear and additive process. There are few individual components that appear in every state medical marijuana law. Some are nearly ubiquitous, like some form of packaging regulation, but even with safety there are myriad differences in state rules. It is not determinative that even if states worked to solve a problem and learned from each other, subsequent adopters would adopt those practices. That said, convergence clearly occurs as states deal with practical issues in innovative polices. This is particularly stark in the area of medical marijuana because states cannot rely on federal regulations and agencies for support in ensuring access to safe and effect pharmaceuticals. Instead, states struggled through figuring out how to serve those roles and converged on solutions to related problems.
It also appears that proponents and opponents of medical marijuana learned political lessons from earlier adopting states. A major threshold that groups had to cross in medical marijuana was the shift from passing bills through direct initiatives to guiding them through the legislative process. As the sputtering advancement of recreational marijuana programs in New York and New Jersey showed in 2019, the dynamics of legislative adoption differ from direct initiatives. As time progressed and medical marijuana was taken up in increasingly conservative legislatures, advocates changed their lobbying strategy from one of partisan support to bipartisan access buying. Advocates also learned how different frames brought together increasingly diverse advocacy coalitions for medical marijuana. Moving beyond the focus on AIDS and cancer patients in earlier adopting states, advocacy was increasingly embraced by powerfully and positively constructed groups (i.e., veterans) and weak, but positively constructed, groups (children and parents) ( Schneider and Ingram 1993). The shift in framing legitimized the conferring of benefits, like access to medical marijuana, on these groups.
While this survey of medical marijuana policy adoption suggests that learning occurred, researchers should consider additional strategies for making more definitive statements about the presence of policy learning during diffusion. We suggest three methods: A broader use of text analysis, deeper qualitative interviews with interest groups strategists and legislators, and further exploration of the marijuana industry. Text analysis is promising for the study of policy diffusion ( Jansa, Hansen, and Gray 2019; Linder et al. forthcoming). Analyses of the corpus of statehouse news coverage on an issue like marijuana would help to systematically identify instances where state actors refer to learning from source states when considering legislation. At times, these references make it into the bills themselves, but rules for writing legislation differ across the states and make analysis of the laws a conservative indicator of learning. Expansion to additional documents beyond the laws themselves is promising for identifying diffusion mechanisms ( Train and Snow 2019). Interest groups like MPP and NORML serve as vectors in the transmission of policy ideas ( Boushey 2010). Thus, deeper interviews with these groups can reveal the linkages between states that are not captured in news articles and government documents, as could interviews with key legislators and their staff ( Wesley and Salomons 2019). Finally, large cross-state corporations are forming and increasing their lobbying efforts as marijuana markets grow. These companies leverage their size and purported successes in other states to involve themselves in policymaking and budding industries in newly adopting states, thus creating another vector for policy transmission.
While much of the policy diffusion literature seeks to understand policy learning at the stage of adoption by observing and explaining spatial and temporal adoption patterns, this misses a lot of the learning that occurs throughout the policy process. Agenda setting involves a public negotiation over what is possible in a piece of legislation. Both political and policy learning result in choosing frames as well as policy components that set the legislative agenda. Additionally, as much as adoption is about political success (i.e., policy viability, demand, and popularity), implementation is about policy success and is shaped by local governments, interest groups, and consultants. Given that laws are often not highly prescriptive, the locus of policy learning shifts from legislative adoption to the regulatory rulemaking process. By delving more deeply into a single policy like medical marijuana, we can illustrate how and why policy and political learning occur.
