dentify the independent, dependent, and any intervening variables. Indicate whether the relationships among them are direct or inverse
yasmine
Defiant Innovation: The Adoption of Medical
Marijuana Laws in the American States
A. Lee Hannah and Daniel J. Mallinson
Diffusion research often characterizes the role of the federal government in innovation adoption as a
supportive one, either increasing the likelihood of adoption or its speed. We examine the adoption of
medical marijuana laws (MMLs) from 1996 to 2014 to shed light on what motivates states to adopt
innovations that are in explicit defiance of federal law. Furthermore, we examine whether federal
signals have any influence on the likelihood of adoption. In doing so, we utilize implementation
theory to expand our understanding of how the federal government’s position impacts state policy
innovation adoption. We find mixed evidence for the influence of federal signals on the adoption of
MMLs. The results suggest that medical marijuana policies are much more likely to be adopted in
states when proponents have the political or institutional capital, rather than a medical or fiscal need.
Moreover, this political capital is sufficient independent of the federal government’s real or perceived
position.
KEY WORDS: medical marijuana, policy diffusion, federalism
扩散研究常认为联邦政府在创新采纳中扮演着支持的角色, 它不是提高了政策采纳的可能性,
就是加快了采纳的步伐。我们分析了1996年到2014年间的医用大麻法案, 以此来阐明是什么促
使各州采纳那些明显违背联邦法律的创新。我们还分析了联邦的信号是否影响了政策采纳的可
能性。借此, 我们利用政策执行理论来扩展我们对于联防政府的立场对于州政策创新影响的理
解。我们发现联邦信号对于医用大麻法案采纳的影响复杂且模糊。结果表明, 医用大麻政策更
可能被那些支持者们有政治或制度资源的州, 而非那些有医疗或财政需求的州所采纳。而且, 这
些政治资源足够独立于联邦政府真实的或被感知的立场。
Introduction
The American federal system creates opportunities for sharing and learning
among localities, states, and the federal government. Often touted as laboratories of
democracy,1 the states experiment with new policy ideas, which may spread hori-
zontally to other states, as well as exhibit vertical influence on the federal govern-
ment’s issue agenda. Yet, federalism also generates conflict. For example, states
compete with each other for tax resources and residents (Dye, 1990; Tiebout, 1956).
402
doi: 10.1111/psj.12211
VC 2017 Policy Studies Organization
The Policy Studies Journal, Vol. 46, No. 2, 2018
bs_bs_banner
They also wrestle with the federal government over the proper boundaries of state
sovereignty and the implementation of federal initiatives (Rabe, 2007; Thompson &
Gusmano, 2014). Moreover, from the early doctrine of nullification to the passage of
medical and recreational marijuana policies in the modern era, states sometimes
actively defy federal law. We explore the concept of defiant innovation—a process
whereby states, through initiatives or the legislature, pass laws that not only circum-
vent, but also reimagine federal law. The study of policy diffusion brings insight into
the complex intergovernmental relationships fostered by federalism, yet it does not
address vertical conflict as much as vertical influence. We do so by examining the
adoption of medical marijuana laws. Moreover, we build bridges between theories
of policy implementation and diffusion in order to provide a framework for identify-
ing diffusion patterns that typify defiance.
We use the diffusion of medical marijuana to address the question of what moti-
vates states to adopt innovations that are in explicit defiance of federal law. This is
important for establishing whether the expected determinants of diffusion have the
same effect when states are defiantly adopting. This study also speaks to whether
defiant innovation is motivated differently than the well-studied cases of states
undermining federal mandates (Balla & Deering, 2015; Gormley, 2006; Regan &
Deering, 2009; Shelly, 2008). The two are linked, yet distinct, because in the first case
states are actively innovating in a policy area expressly prohibited by the federal
government and in the second states are refusing incentives and/or mandates to
implement federal initiatives. We seek to determine whether these two acts are moti-
vated differently.
The extent to which medical marijuana can be labeled “defiant” has fluctuated
over time since California first passed medical marijuana in 1996. At times, the fed-
eral government has signaled deference to the states on these issues, providing
implicit support for experimentation. At other times, the federal government has
been proactive and even aggressive in enforcing marijuana laws (O’Keefe, 2013).
Such variation in credible commitments to marijuana prohibition enforcement by the
federal government provides a useful test of how those commitments impact the
willingness of states to innovate in an area of federal–state conflict. Not only does
this address an important gap in policy diffusion research, it is timely given that
eight states have already legalized recreational marijuana and several additional
states are likely to consider legalization in the next 2 years (Steinmetz, 2016). Further-
more, national polling data shows increasing support for marijuana legalization
(Jones, 2015). It is likely that the factors driving state adoption of MMLs are consis-
tent with broader trends in marijuana policy. An understanding of the mechanisms
that drive the adoption of these policies might help to anticipate future development
of recreational marijuana as the federal landscape continues to evolve.
We find mixed evidence for the role of federal commitments to enforcement in
the adoption of MMLs. States were less likely to adopt medical marijuana policies
when the conservative Bush administration was in office. However, while President
Obama’s administration signaled a less interventionist approach from the federal
government, states were no more likely to adopt an MML. While different adminis-
trations signal implicit support or opposition to medical marijuana, we find that
Hannah/Mallinson: Defiant Innovation 403
explicit signals in the form of Drug Enforcement Administration (DEA) marijuana
seizures also do not affect states’ probability of adopting. Thus, the federal govern-
ment had less of an impact on the timing of MML adoptions than would be
expected. Instead, we find that state-level factors such as the availability of the initia-
tive and citizen liberalism increase the likelihood of medical marijuana adoption and
cultural differences, like the number of evangelical adherents, decrease the likelihood
of adoption. Further, states are more likely to adopt if they are ideologically similar
to previous adopters. Taking the results as a whole, the evidence suggests that states
moved forward in defiance with little consideration of the federal government’s
position.
The Logic of State Defiance
States do not always act in accordance with the will of the federal government.
This was most clear during attempts at nullification and interposition prior to the
American Civil War and school integration during the 1950s and 1960s and finally
resolved by the Supreme Court in 1971 (Swann v. Charlotte-Mecklenburg Board of Edu-
cation 402 U.S. 1). The doctrines of nullification, meaning states have the right to nul-
lify, or not implement, federal laws they view as unconstitutional, and interposition,
claiming that states stand as a barrier between citizens and unconstitutional federal
acts, arose from states determining that they were protecting their citizens from the
federal government.2 While the courts have dismissed both doctrines (e.g., Cooper v.
Aaron 358 U.S. 1 [1958]), states still attempt to nullify federal mandates with which
they do not agree. The question remains as to whether acts of defiance spread among
the states in a fundamentally different manner than policies that are supported by
the federal government or those whereby the federal government takes a neutral
stance.
Policy implementation research provides a foundation for developing a cogent
theory of how states respond to federal signals and why they, at times, innovate in
defiance of federal law. Goggin, Bowman, Lester, and O’Toole (1990) sought to pro-
vide testable hypotheses for why states vary in their willingness to implement fed-
eral priorities and subsequent success. Fiscal Federalism, whereby the federal
government induces state implementation through financial incentives, in particular,
created the potential for principal–agent problems (Chubb, 1985). Extant policy diffu-
sion research regarding state responsiveness to federal signals often focuses on
instances when the federal government incentivizes diffusion, meaning when states
respond positively to national coercion (Nicholson-Crotty, 2009; Welch & Thompson,
1980). These are cases where the agent states are working for the principal. Of course,
states can also fail to implement federal priorities (i.e., shirk) or even redesign federal
policy (i.e., sabotage).3 Furthermore, the federal government does not only provide
inducements (i.e., positive signals) for state action, but also utilizes constraints (i.e.,
negative signals). For either federal inducements or constraints to be implemented
by the states, the signals sent to them must be credible, clear, consistent, repeated,
and received (Allen, Pettus, & Haider-Markel, 2004; Goggin et al., 1990).
404 Policy Studies Journal, 46:2
States vary in their receptivity to these signals, not only cross-sectionally, but
also temporally. Goggin et al. (1990) hypothesized that states with supportive legisla-
tures and advocacy coalitions will implement federal policy more quickly than those
with weaker, or less supportive, institutions. Ecological capacity (i.e., a combination
of economic, political, and situational capacities) is also important to both implemen-
tation and diffusion theory. States with greater financial and institutional capacity
are in a stronger position to implement federal policy quickly and effectively. In dif-
fusion parlance, slack resources allow states to overcome obstacles to adoption
(Walker, 1969). Conversely, ecological capacity provides the resources to resist fed-
eral signals. Moreover, implementation is temporally dynamic; as a state’s capacity
changes or the legitimacy of the signal sender erodes or improves over time, states
can change their implementation behavior. This temporal dynamic is vital for under-
standing the spread of defiant innovations like medical marijuana policies.
Table 1 establishes a framework for differentiating the observable expectations
of adoption patterns when the federal government and states are either in concor-
dance or discordance over a specific policy. Each can take three positions on a given
policy: pro, anti, or neutral. It is important to note that this table presents the overall
adoption pattern that is observable when a plurality of states falls into each category.
A typology of the macrolevel patterns will allow future research to address how
additional policies fit within each category. After describing these macrolevel pat-
terns, we will shift the unit of analysis to the individual states, so as to establish test-
able hypotheses for our subsequent event history analysis of medical marijuana
adoption.
Perhaps the two most obvious and well-supported categories in Table 1 are
those in which the federal government sends positive signals through financial
incentives (Welch & Thompson, 1980), issue attention (McCann, Shipan, & Volden,
2015), legislative activity (Karch, 2006), or mandates (Woods & Bowman, 2011), and
those in which the federal government is assumed to have a neutral position (e.g.,
Berry & Berry, 1990). For policies like the Civil Defense Compact, the federal govern-
ment increased state financial capacity for implementation, and thus states
Table 1. Expected Diffusion Patterns When the Federal Government Takes a Pro-, Anti-, or Neutral Stance and States Take Pro-, Anti-, or Neutral Stances to an Innovation
Federal Government
State Government
Pro Neutral Anti Pro Rapid adoption
(Civil Defense Compact)
Characteristic- dependent adoption
(Lotteries)
Expanding adoption, declining enforcement
(Medical Marijuana)
Neutral Compliance with partisan reservations
(Common Core)
Highly regionalized issue
(Colorado River Compact)
Limited defiance, varying enforcement
(Medical Marijuana)
Anti Formal opposition (REAL ID)
Limited adoption (Stem Cell Research)
Limited defiance, widespread enforcement
(Medical Marijuana)
Hannah/Mallinson: Defiant Innovation 405
responded quickly in adopting the innovation (Goggin et al., 1990; Nicholson-Crotty,
2009; Savage, 1985; Welch & Thompson, 1980). When the federal government is neu-
tral toward an innovation, like a state lottery, states are prompted to adopt based on
the expected internal determinants and external influences theorized by Berry and
Berry (1990, 2013). Thus, the policy’s overall adoption pattern should resemble a nor-
mal s-curve (Boushey, 2010; Gray, 1973; Mallinson, 2016) with relatively wide adop-
tion over time.
It is also theoretically possible to have a plurality of states take neutral positions
on policies for which the federal government sends positive, negative, or neutral sig-
nals. Policies of highly regional significance seem most likely to fit into the neutral-
neutral category. The protection and management of rivers (e.g., the Colorado River
Compact) for example, may garner little interest from a plurality of the states and
the federal government.
In the case of a positive federal position and a neutral plurality, we would expect
neutral states to experience vocal partisan reservations to the given policy that would
initially limit adoption (e.g., Common Core). Nonetheless, federal incentives can
push states from a neutral to a generally positive stance. The rate of adoption would
increase as this shift occurs, emphasizing the point that a policy can start in one cate-
gory, with a particular expected pattern of spread, and move into another category
via changes in federal signals and/or circumstances in the states. Gay marriage pro-
vides a useful example of this principle. Congruence between state and federal pol-
icy fluctuated across time as states moved in a direction different than the federal
government, until the courts stepped in to settle the law.
There is further evidence of what happens when the federal government sends
positive signals regarding a policy that states simply do not want. For example, Con-
gress passed the REAL ID Act (2005) in response to recommendations by the 9/11
Commission (2004) that the government establish standards for personal identifica-
tion, including those issued by states (e.g., drivers licenses). Maine was the first state
to adopt a law opposing the federal REAL ID requirements and, within 18 months,
20 other states followed suit (Regan & Deering, 2009). In this case, the federal gov-
ernment is trying to induce states to produce a desired policy outcome, but states
refuse. Medicaid expansion under the Affordable Care Act highlights how financial
inducements can erode state resistance over time (Jacobs & Callaghan, 2013). State
resistance is certainly not new. Maryland did not choose to enforce federal alcohol
prohibition in the 1920s and states like New York chose to abandon enforcing alcohol
prohibition prior to the ratification of the Twenty-First Amendment, which repealed
Prohibition. Even further back in American history, states like Pennsylvania refused
to enforce Fugitive Slave Laws (O’Keefe, 2013).
In the case of a policy where the federal government expresses neutrality toward
state action and a plurality of states are opposed, it is likely that very few states will
take up such a cause. Take stem cell research funding, for example. In his early work
on this policy, Karch (2010) identified state stem cell funding as an example of lim-
ited diffusion. Between 1999 and 2008, seven states provided financial support for
stem cell research, three legalized the conduct of this research (without funding),
and six restricted it. In fact, state funding of stem cell research has not substantially
406 Policy Studies Journal, 46:2
expanded further since then (Gugliotta, 2015). Karch (2012) later demonstrated that
while the innovation was not widely adopted, the Bush administration’s banning of
federal research funding prompted many states to consider their own support.
While, on one hand, the federal government communicated its disagreement with
the policy by banning its own source of funding, it did nothing to prevent states
from instituting their own programs. Thus, the federal government is expressing
neutrality toward state innovation. The result was a limited pattern of adoption with
a few concurring anti-innovation states adopting their own bans on research.
The final three categories, which we find exemplified in the spread of MMLs,
are those where the federal government prohibits an activity that states desire (i.e.,
the growth, distribution, and consumption of marijuana). This is distinct from cases
like REAL ID, where the states are refusing to implement a federal program.
Granted, both are principal–agent problems, whereby the states as agents are not
implementing the principal’s desires. The distinction is that mandate resistance is a
case of agent shirking, whereas a defiant innovation, like an MML, is sabotage. In
defiant innovation, states are not only simply allowing a banned activity, but they
are actively promoting the development of a new program even when the federal gov-
ernment bans it. Marijuana is presently classified as a Schedule I drug under the
Controlled Substances Act of 1970 (CSA). Thus, the distinction from mandate refusal;
states are innovating in defiance of federal law by authorizing the distribution of
marijuana for medical purposes.4
States, of course, diverge in their willingness to engage in defiant innovation. A
shrinking, but still substantial, minority is supportive of federal law or is legitimately
dissuaded from acting against federal interests, and thus sees no reason to enact
allowances for marijuana use. Other states are sympathetic, but lack the ecological
capacity to enact defiant innovations and risk retribution (e.g., lawsuits) by the fed-
eral government or peer states (Kamin, 2015). As the plurality of states shifts from
opposed to the innovation, to neutral, to positive, the patterns of both adoption and
enforcement change. We now transition to addressing states as the unit of analysis
and drawing further on implementation and principal–agent theory to derive
hypotheses specific to the predictors of defiant innovation.
Expectations for the State-Level Predictors of Defiant Innovation
Explicit Federal Government Signals
A state’s decision to adopt a policy innovation is typically a function of their
internal motivations, the obstacles to adoption and availability of resources for over-
coming those obstacles, the effects of other policies and availability of alternative sol-
utions, and the influence of external actors (Berry & Berry, 1990, 2013). The federal
government is one external actor that increases the resources available to states for
innovation (Eyestone, 1977; Welch & Thompson, 1980), as well as the obstacles for
adoption (Karch, 2007). The executive branch, in particular, is dependent on states to
implement a host of domestic policy initiatives. From Medicaid to the Elementary
and Secondary Education Act to the War on Drugs, the states play a major role in
Hannah/Mallinson: Defiant Innovation 407
the success or failure of these programs. Increasingly, federalism scholars have noted
the shift in legislatively driven cooperative federalism to executive-driven coopera-
tive federalism (Gais & Fossett, 2005). Meaning, the executive branch has been taking
an increased role in shaping and reshaping intergovernmental implementation in the
face of gridlock in Congress and diminished legislative control (Bulman-Pozen &
Metzger, 2016; Gais & Fossett, 2005). The use of waivers for No Child Left Behind
and the Affordable Care Act clearly demonstrates the practical effects and limitations
of executive federalism (Dinan, 2014; Saultz, McEachin, & Fusarelli, 2016; Thompson
& Gusmano, 2014).
The dependence of the federal government on the states for policy imple-
mentation provides ample opportunity for states to shirk or sabotage their role.
The federal government, of course, uses the tools at its disposal (grants, waivers,
cross-over sanctions, among others) to encourage states to faithfully implement
the law. These tools send explicit signals to the states regarding the executive’s
expectations regarding proper policy implementation. Federal signals must be
credible and legitimate in the eyes of state officials for “implementation to pro-
ceed promptly and without modifications” (Goggin et al., 1990, p. 175). While
prior diffusion research corroborates the importance of credible and legitimate
signals (Allen et al., 2004), much of the extant research focuses on instances
where the federal government is promoting an innovation. In contrast, defiant
innovation modifies dissonant federal policies.
We argue that the federal government sends two kinds of signals—explicit
and implicit—to which the states respond. Explicit commitments come in the
form of executive branch enforcement activities. In the case of No Child Left
Behind, the Obama administration sent explicit signals regarding its desire to
reform the implementation of that law through the use of waivers (Saultz et al.,
2016). In the present case of MMLs, the Department of Justice (DOJ) and Drug
Enforcement Agency (DEA) have authority over the enforcement of federal drug
laws. This enforcement power includes the ability to seize marijuana. We
hypothesize that increased seizure of marijuana sends explicit signals regarding
the executive branch’s support of federal marijuana prohibition. There are, of
course, stronger signals at the president’s disposal, namely the millions of dol-
lars in funds and equipment sent to local police departments for the purpose of
executing the drug war. To our knowledge, the president has yet to use this
major stick against states acting in defiance of federal law. Thus, we use the mea-
sure of seizures as the clearest signal of the executive branch’s commitment to
marijuana prohibition. Given that states vary in their ecological capacity to defy
federal law, stronger signals from the executive branch should raise the obstacles
to adoption to the point that the diffusion of a defiant innovation like medical
marijuana would slow, but not stop (Rabe, 2007). Conversely, the weakening of
explicit signals should prompt increased innovation activity.
Explicit Federal Commitments Hypothesis: Increased federal enforcement
decreases the likelihood that states will adopt a defiant innovation and
decreased enforcement has the opposite effect.
408 Policy Studies Journal, 46:2
When campaigning for office, presidents also send implicit signals regarding
their favorability toward a host of policies under their jurisdiction. Even after the
president is able to use their powers to send explicit signals regarding state imple-
mentation, some presidents maintain an image of promarijuana or antimarijuana
that may not directly match with their enforcement of the law. For example, Presi-
dent Obama has received criticism for increasing marijuana enforcement activities
during his presidency (Tau, 2012). This sends conflicting signals, as states strategi-
cally anticipate how the federal government will receive their defiant actions.
Changes in presidential administration provide one implicit signal of agreement or
disagreement with state defiance. It is possible that an administration can have a pos-
itive or negative impact on state innovation, even if the perception of their views on
a policy deviates from more explicit signals sent to the states.
Implicit Federal Commitments Hypothesis: Changes in presidential
administrations alter the likelihood of adopting a defiant innovation.
Demand
Demand for the innovation should have a particularly strong effect, providing
an impetus for defying federal law. This demand could be for the products of the
policy itself, medical marijuana for example, or state fiscal demand if the innovation
has a substantial influence on the state’s fiscal health. States regularly propose alter-
native means for raising revenue in order to avoid less politically palatable increases
to income and consumption taxes (Berry & Berry, 1990; Nelson & Mason, 2007). For
example, in the case of medical marijuana, states not only levy taxes on individuals,
they also tax dispensaries and distributors (Hickey, 2014). In many states, this has
led to substantial windfalls (Oosting, 2014; Stern, 2015).
Usage Demand Hypothesis: The likelihood of a state innovating defiantly
increases as internal demand for that innovation concomitantly increases.
Fiscal Demand Hypothesis: States facing fiscal stress are more likely to
adopt a defiant innovation that can raise additional revenue.
Ecological Capacity
Ecological capacity “encompasses state wealth, state institutional capabilities,
and state situational contexts” (Goggin et al., 1990, p. 135). Such capacity is necessary
for rapid and effective implementation (Goggin et al., 1990) and innovation adoption
(Walker, 1969). A state’s ecological capacity should thus condition its ability to
respond to federal signals in defiance. The popular initiative is an important institu-
tional capability for defiance, as it provides a means for citizens that are dissatisfied
Hannah/Mallinson: Defiant Innovation 409
with federal policy to circumvent state legislators reticent to defy the federal govern-
ment. Thus, if public opinion is growing increasingly incongruent with federal policy
and usage demand is increasing, state defiance actually increases opinion-policy con-
gruence and improves citizen perceptions of state responsiveness (Ferraiolo, 2008).
Moreover, the initiative serves as a legitimizer of innovative policies and sets the
agenda for other states, including legislatures in states that do not have the initiative
(Magleby, 1988). Finally, while initiatives are sponsored by outside groups, their
inclusion on the ballot still indicates a degree of state defiance, since they must be
certified by a state’s executive branch. In the case of MMLs, the first five laws
(Alaska, California, Maine, Oregon, and Washington) were adopted through direct
initiative between 1996 and 1999. Hawaii became the first state, in 2000, to pass an
MML through the state legislature.5
Initiative Availability Hypothesis: States with an available initiative pro-
cess are more likely to adopt an MML.
Other institutional, political, and social factors that should impact the likelihood
of MML innovation include legislative professionalism (McCann et al., 2015), voter
liberalism (Berry, Ringquist, Fording, & Hanson, 1998), and prohibitionist religious
culture (Frendreis & Tatalovich, 2010; Jensen, 2003). Many of these relate to a state’s
political culture, which shapes the policy choices made by states (Elazar, 1966) and
intergovernmental implementation (Goggin et al., 1990). These components of a
state’s ecological capacity influence its willingness to defy or support federal law.
Thus, while we expect federal signals to be weak, internal characteristics should be
prevalent predictors of adoption.
Ecological Capacity Hypothesis: Internal characteristics of the state, like
ideology, religious and political cultures, and institutional capacity (e.g.,
legislative professionalism) condition the likelihood that states will react
positively or negatively to federal signals in deciding to adopt a defiant
innovation.
Diffusion
Diffusion scholars have long found that innovations may spread outward from
an innovator state to geographically contiguous states (Berry & Berry, 1990; Walker,
1969). In short, states that share borders are more likely to follow the political hap-
penings of their neighbors than those across the country. States that share borders
are also likely to be more similar in terms of their internal characteristics than states
that are farther away. Finally, neighbors are in direct competition (Berry & Baybeck,
2005; Dye, 1990), which is relevant given the large sums of tax revenue marijuana
can yield. Thus, it is important to test whether neighbor states are in fact influencing
each other.
410 Policy Studies Journal, 46:2
Neighbor Hypothesis: A state is more likely to adopt a defiant innovation
as the proportion of its contiguous neighbors that previously adopted also
increases.
While spatial relationships are important for policy diffusion, geography does
not provide the only means of similarity through which policies can diffuse. Ideolog-
ical similarity is a potential pathway of policy learning (Grossback, Nicholson-
Crotty, & Peterson, 2004) and likely relevant for acts of defiance. If states are choos-
ing to defy federal policy, they may be doing so based on ideological differences,
meaning the set of states willing to defy are connected more through ideology than
geography. Thus, it will be ideological peers, not contiguous neighbors, which influ-
ence defiant innovation.
Relative Ideology Hypothesis: A state is more likely to adopt a defiant
innovation if the ideology of past adopters falls closer to its own.
We now turn to describing our operationalization of these concepts and analyti-
cal technique for testing the above hypotheses.
Data and Analytical Technique
Dependent Variable
Data were collected on the 50 states from 1996 through 2014. Of these states, 23
passed MMLs, either by voter initiative or statute (see Figure 1 for a map of these
states). We employ event history analysis (EHA) to test the hypotheses presented
above (Berry & Berry, 1990). The time period covered in the model starts in the year
of the first passage of medical marijuana legislation (1996) and advances to 2014. In
1996, California voters approved Ballot Proposition 215, making California the first
state to adopt and drop out of the dataset. By 2014, 27 states remain in the risk set.6
Independent Variables
Credible commitments for federal marijuana prohibition enforcement by the
executive branch have varied over time, resulting in uncertainty for states, dispensa-
ries, and other businesses that support the marijuana industry. As medical marijuana
legislation emerged in the states, the Clinton administration made the War on Drugs
a priority in order to avoid being seen as soft on crime (National Research Council,
2014). It was not until the waning days of his administration that President Clinton
argued for marijuana decriminalization (Wenner, 2000). During the George W. Bush
administration, funding was increased for global drug interdiction and the federal
government remained committed to marijuana prohibition enforcement (Gaumond,
Davis, & Hill, 2009). The commitment to enforcement wavered during the Obama
administration. In 2009, DOJ issued the Ogden (2009) memo, which suggested that
Hannah/Mallinson: Defiant Innovation 411
DOJ would not expend resources prosecuting individuals that were acting in compli-
ance with state MMLs. DOJ, however, stepped back from this assurance in 2011 by
asserting their commitment to enforcing the CSA (Cole, 2011). Finally, in 2013, DOJ
announced that it would leave primary enforcement of marijuana law to states with
“strong and effective regulatory and enforcement systems to control the cultivation,
distribution, sale, and possession of marijuana” (Cole, 2013). It also threatened to
pursue enforcement action if state enforcement mechanisms are not robust enough.
While this paves the way for states to operate medical and recreational marijuana
programs with less threat of federal government involvement, there are no guaran-
tees that the government’s “wait and see” approach will remain as administrations
change.
There is an arresting gap, however, between the rhetoric and practices of all
three presidents. Figure 2 displays one marker of this by plotting the total amount of
marijuana seized domestically (1,000 kilograms) by the DEA from 1986 through
2014. It also demarcates the first year in office of Clinton (1993), Bush (2001), and
Obama (2009). Strikingly, marijuana seizures declined rapidly prior to Clinton taking
office, stayed relatively steady during the Clinton administration, declined again
during the Bush administration, and increased rapidly under the Obama administra-
tion to levels not seen since Ronald Reagan’s expansion of the War on Drugs. This,
of course, is not the only way in which the executive branch enforces federal mari-
juana prohibitions, but seizures are highly meaningful to states, like California, that
Figure 1. Map of States with Medical Marijuana Laws as of 2014 (Shading Based on Years Since Law Was First Passed).
412 Policy Studies Journal, 46:2
encourage the development of a network of private suppliers for their medical mari-
juana programs. Thus, the states have received different implicit and explicit signals
regarding their policy experiments with marijuana. In fact, President Obama has
been criticized for these conflicting messages (Scherer, 2012).
Thus, we examine both the implicit and explicit signals of administrative com-
mitment to federal marijuana prohibition. We use two dichotomous indicators
denoting the Bush Administration and the Obama Administration as implicit mea-
sures of the federal government’s preferences. We expect that different administra-
tions signal changes in federal drug policy to the states. We anticipate that the more
conservative Bush administration discouraged states from adopting medical mari-
juana policies. Meanwhile, we expect that states were encouraged to adopt with less
fear of federal intervention during the Obama administration. The Clinton adminis-
tration serves as the reference category for each of these comparisons. We also model
explicit changes in the federal government’s stance by controlling for the Federal
Seizures of marijuana in a given year by the DEA.
We consider usage and demand in the state by including Marijuana Treatment
Admission Data (TEDS),7 Cancer Incidence Rate, and Glaucoma Prevalence. We
expect that higher rates of marijuana use will remove the stigma associated with
marijuana use and increase support for adopting an MML. We also anticipate that
demand for the program will increase in states with a higher number of potential
patients that could benefit from the drug. In 22 of the 23 laws we examined, cancer
and glaucoma are listed as debilitating medical conditions in the text of the laws;
and in most cases they are listed first. We that expect states with a higher preponder-
ance of cancer or glaucoma may experience greater pressure from citizens to increase
access to this medical treatment. To examine fiscal opportunity, we included a
lagged measure of Fiscal Health measured as the ratio of total-state-revenue-minus-
total-state-spending to total-spending (Berry & Berry, 1990). We expect that states
with lower year-end balances or deficits are more likely to adopt an MML.
Along with these measures based on our hypotheses, we include a number of
internal characteristics as controls that we expect to affect the probability of a state
Figure 2. Marijuana Seized by the U.S. Drug Enforcement Administration (in 1,000 kilograms), 1986– 2014.
Hannah/Mallinson: Defiant Innovation 413
adopting an MML. We control for the role of states’ institutional capacity by includ-
ing an indicator for Initiative Availability and Legislative Professionalism. The
ability to use the direct initiative should increase the likelihood of a state passing a
law. We also expect that more professional legislatures will be able to devote the
time and resources to taking up this controversial and complex policy (Huber & Shi-
pan, 2002; Huber, Shipan, & Pfahler, 2001). Next, we consider characteristics of the
state electorates. We use a citizen ideology measure (Citizen Liberalism) developed
by Berry and colleagues to test the relationship between a state’s ideology and its
adoption of medical marijuana policy (Berry, Fording, Ringquist, Hanson, & Klarner,
2010; Berry et al., 1998). We expect that more liberal states are more likely to pass
MMLs. While there is an ideological dimension to support for medical marijuana
(Pew Research Center, 2015), there is also a moral dimension that we capture with
the statewide Evangelical Rate, which captures the size of prohibitionist religious
cultures in a state (Frendreis & Tatalovich, 2010).
Finally, we consider the role of state-to-state diffusion. We created a Proportion
of Neighbors with MML to account for the fraction of neighboring states that have
previously adopted an MML. We expect that states are increasingly likely to adopt a
law if their neighbors are also passing them. We also construct the Relative Ideology
of States Adopting MMLs and expect that states are more likely to adopt MMLs if
states that are ideologically similar have already passed the law (Grossback et al.,
2004). Finally, we include the Time After the First Adoption to account for remain-
ing duration dependence (Beck, Katz, & Tucker, 1998). Table A1 in the appendix fur-
ther describes each of the independent variables and provides summary statistics
along with the data source.
Results
To illustrate the consistent spread of this innovation, regardless of changes in the
federal government’s position on the issue, Figure 3 displays the cumulative total
number of states that adopted medical marijuana programs between 1996 and 2014.
It also displays separate cumulative adoption curves for states adopting via a legisla-
ture and those adopting via a popular initiative. Immediately evident is that many of
the innovators and early adopters took up MMLs through the initiative, as it pro-
vided early cover for legislators that did not want to be identified with marijuana
policy. Figure 3 also shows that the expansion of medical marijuana programs
remained consistent after the initial period of rapid adoption and that many of these
later adoptions were passed by legislatures. In fact, initial ballot success and growing
public support further pushed legislatures to consider and adopt MMLs (Ferraiolo,
2008).
Table 2 provides results from a logistic EHA model predicting MML adoption.
The model includes traditional policy diffusion variables accounting for ecological
capacity and regional diffusion as well as signals from the federal government and
ideological diffusion. Finally, the second column reports odds ratios for the variables
that are statistically significant. Federal signals appear to have had little impact on
defiant adoption of an MML. The explicit signal of DEA seizures has no effect.
414 Policy Studies Journal, 46:2
Figure 3. State Adoptions of Medical Marijuana Laws, 1996–2014.
Table 2. Results of Logistic Regression Model Predicting Medical Marijuana Policy Adoption in the U.S. States (1996–2014)
Coefficients Odds Ratios
Federal Signals Marijuana Seizures in kilograms (10k) 20.02 (0.02) Bush Administration –3.85* (1.73) 0.021
[0.001, 0.408] Obama Administration –4.811 (2.77) 0.008
[0.000, 1.211] Ecological Capacity Initiative Availability 1.11* (0.52) 3.029
[1.084, 8.464] Legislative Professionalism 0.07 (2.77) Citizen Liberalism 0.09* (0.03) 1.097
[1.031, 1.167] Evangelical Rate –0.011 (0.00) 0.989
[0.974, 1.003] Fiscal Healtht-1 20.64 (3.27) Demand Marijuana TEDS 0.00 (0.00) Glaucoma Percent 25.09 (3.11) Cancer Incidence Rate
(per 100,000) 0.00 (0.00)
Diffusion Proportion of State Neighbors with MML 0.16 (1.50) Relative Ideology of States Adopting MML –0.13* (0.05) 0.881
[0.799, 0.965] Time After First Adoption 0.51** (0.19) Constant 1.78 (4.93) N 768 Log Likelihood 274.48
Wald v2 33.3** 1 indicates statistical significance at p < 0.10 (two-tailed); *p < 0.05, **p < 0.01. Robust standard errors reported in parentheses. 95% confidence intervals reported in brackets.
Hannah/Mallinson: Defiant Innovation 415
Additionally, while the likelihood of adoption is lower during the conservative Bush
administration, it remains so during the more implicitly, though less explicitly, sup-
portive Obama administration. In terms of the odds of adoption, they declined by
nearly 100 percent in the Bush and Obama administrations (each relative to the Clin-
ton administration).8
Regarding state ecological capacity, not only the initiative, but also citizen liber-
alism and Protestant fundamentalism are important. A one standard deviation
(14.99) increase in citizen liberalism results in a 145 percent increase in the odds of
adopting an MML. Additionally, the odds decrease by 130 percent with a one stan-
dard deviation (117.8) increase in evangelical adherents. Having an initiative process
available increases the odds of adopting an MML by 203 percent. The results for
medical usage demand are not significant (cancer and glaucoma incidence). Mean-
while, there is no evidence that increased marijuana use motivates a state to act on
passing an MML. Finally, fiscal health does not have a statistically significant effect
on the probability of adoption.
Perhaps most intriguing, from the standpoint of traditional expectations of
diffusion patterns, is that MMLs did not experience regional diffusion through
contiguous neighbor adoptions. Instead, MMLs have an important ideological
component, whereby states that are less ideologically similar to those that
already adopted are less likely to adopt an MML.9 Substantively speaking, a one
standard deviation (8.17) increase in the ideological distance between a state and
those that adopted previously results in a 97 percent reduction in the odds of
adopting an MML.
Furthermore, the combination of the ideological spread of this policy with
the declining likelihood of adoption after the initial innovation and early adop-
tion phases suggests that MML adoption may be reaching a saturation point,
whereby most of the states that have demand for an MML already have it and
those that support federal marijuana prohibition are less likely to consider adopt-
ing without fundamental changes in their internal political or social context.
Thus, while many states still have no medical marijuana statute, the policy may
have already saturated the states that are willing to act in defiance of federal
law. States in the South and Midwest (see Figure 2), for example, are likely reti-
cent to adopt such a policy, and are more likely to be supportive of the federal
prohibition in the first place. Consequently, these laggard states may be less
likely to adopt moving forward, short of the emergence of substantial internal
pressures due to fiscal stress and gains in cultural acceptability. The results sug-
gest that medical marijuana policies are much more likely to be adopted when
proponents have the political or institutional capital, rather than a medical or fis-
cal need. Moreover, this political capital is sufficient independent of the federal
government’s real or perceived position.
Discussion and Conclusion
What motivates states to innovate in defiance of federal law? Past research dem-
onstrates that internal demand, institutional capacity, and financial constraints are
416 Policy Studies Journal, 46:2
associated with state decisions to shirk the implementation of federal policy (Balla &
Deering, 2015; Palazzolo, Moscardelli, Patrick, & Rubin, 2008; Regan & Deering,
2009); however, we take this a step further to test whether these dynamics hold in
the case of states sabotaging federal law by defiantly innovating new policy. We
find that, in the case of medical marijuana policy, policy adoption was most
clearly associated with components of a state’s ecological capacity (the initiative,
evangelical rate, and citizen liberalism). Furthermore, this policy exhibits an
ideological, not a geographical, pattern of spread. The states most willing to act
in defiance did so early on and adoption became increasingly less likely over
time. This was the case not only during the conservative Bush administration,
but also after the Obama administration implicitly signaled greater openness to
these programs. Early adoptions via the initiative gave way to a mixture of legis-
lative and initiative adoptions, thus corroborating the legitimizing effect of the
popular initiative (Magleby, 1988).
The decline in the likelihood of adoption during both the Bush and Obama
administrations appears to be less likely due to the clarity of federal signals regard-
ing marijuana prohibition enforcement, and more to the saturation of adoption in
states that were willing, and had the necessary capacity, to defy federal law. One
way to view this is that the policy is becoming less defiant over time. This appears to
be true, as the tide of public opinion is shifting in support of loosening marijuana
restrictions. Adoption by a now majority of the states further legitimizes this defiant
position, while eroding the legitimacy of the federal government’s position. That
being said, President Obama’s increased enforcement efforts and the election of Don-
ald Trump, and particular the confirmation of Senator Jeff Sessions as Attorney Gen-
eral, demonstrates that even as the federal government’s view loses legitimacy, state
action today is not less defiant. It remains to be seen whether the Trump administra-
tion will enhance domestic interdiction efforts, but the appointment of Jeff Sessions
will put in place an Attorney General that is a staunch opponent of drug legalization.
Pennsylvania, Florida, North Dakota, and Arkansas demonstrated in 2016, however,
that as demand grows, cultural norms change, and states face increasingly difficult
fiscal challenges (Kiewiet & McCubbins, 2014), laggard states will still follow, but
they may do so sparingly.
The combination of both the significant and null results in this study paint a
picture of how defiant innovation progressed in the case of medical marijuana
laws, but in a way that likely captures other acts of state statutory defiance in an
era of gridlock and polarization. The popular initiative and ideology were impor-
tant driving forces of MML adoption in the states, regardless of mixed and vary-
ing implicit and explicit federal signals regarding prohibition enforcement. While
the expression of changing popular opinion initially occurred through popular
initiatives, ideologically similar states appear to have been emboldened by these
actions and thus the policy spread among those states that were most likely to
agree with the early defiant actors, and thus disagree with the federal govern-
ment. Over time, as these like-minded states followed suit, the pool of potential
adopters dwindled. Nonadopters, who are themselves more ideologically similar
to each other than to adopters, will likely not adopt without a shift in internal
Hannah/Mallinson: Defiant Innovation 417
support for the policy or substantially increased external influence. This is impor-
tant for the broader study of policy diffusion, as there is limited attention to how
and why policies do not fully diffuse (Karch, 2010) and how the motivations for
adopting change as diffusion unfolds (Rogers, 2003). While ecological capacity is
important for both mandate refusal and this case of defiant innovation, the ideo-
logical bifurcation of adopters and non-adopters and the importance of citizen
initiatives for spurring and legitimizing diffusion are key distinctions. These find-
ings provide an important starting point for differentiating types of state defiance
and developing a fuller picture of how vertical diffusion operates in the United
States.
American federalism is not neat and tidy, in that the federal government encour-
ages innovation through issue attention and financial incentives while also taking up
and nationalizing successful innovations from the states. There is an important ele-
ment of struggle within federalism. There are costs for states and municipalities that
defy the federal government. Cities and universities across the United States are
presently facing the possibility of federal sanctions for refusal to report undocu-
mented immigrants to U.S. Immigration and Customs Enforcement (Arrieta-Kenna,
2016).10 The federal government could conceivably do the same thing to states with
recreational and medical marijuana provisions by withholding billions of dollars in
federal money committed to the War on Drugs. Additionally, the Trump administra-
tion could take the states to court and test the Supreme Court’s support for federal
preemption (Kamin, 2015). This is why the merging of diffusion and implementation
studies is so important. States, particularly when choosing to defy federal law, must
respond to changes not only in their ecological capacity to innovate, but also in the
overall federal ecosystem. If federal signals clarify and intensify in the next adminis-
tration, states may make different choices in either (a) defying in the first place or (b)
how they choose to defy.
Future work building on the theoretical foundation established here can help
explicate not only the when of defiance, but also the how. State approaches to mari-
juana are not identical and that variation can be used to understand the extent to
which choices are dependent on internal capacities or external forces from peer states
and/or the credibility and legitimacy of federal commitments. The study of policy
diffusion can speak to the practical effects of this struggle in a powerful way by iden-
tifying how patterns of policy adoption vary as the relationship between the federal
government and the states varies across policies. State and local defiance in policy
areas that are either expressly prohibited by the federal government or where the
government has made unclear legal commitments is unlikely to subside, particularly
in an era of fragmented federalism.
A. Lee Hannah is an assistant professor of political science in the School of Public
and International Affairs at Wright State University. His research focuses on feder-
alism and marijuana policy, the diffusion of public policy, and education policy.
Daniel J. Mallinson is an assistant professor of public policy and administration in
the School of Public Affairs at Penn State Harrisburg. His research focuses on pol-
icy diffusion, elite behavior, pedagogy, and several policy topics.
418 Policy Studies Journal, 46:2
Notes
The authors would like to thank Liam Anderson, Michael Berkman, Charles Crabtree, Christopher Ojeda,
Julianna Pacheco, and Eric Plutzer for thoughtful and helpful feedback on earlier manuscripts. And we
would like to thank the editor and the three anonymous reviewers of this journal for their valuable com-
ments. Replication files available at: https://dataverse.harvard.edu/dataverse/LeeHannah
1. Supreme Court Justice Louis Brandeis, dissent, New State Ice Co. v. Lieberman, 285 U.S. 262 (1932).
2. The concepts of nullification and interposition originated with Thomas Jefferson and James Madi-
son’s Kentucky and Virginia Resolutions of 1798 and 1799, which were in response to the federal pas-
sage of the Alien and Sedition Acts.
3. These terms are adapted from the study of street-level bureaucracy (Brehm & Gates, 1997).
4. The Tenth Amendment and Supreme Court precedent (see New York v. United States 505 U.S. 144 [1992]
and Printz v.United States 521 U.S. 898 [1997]) prohibit the federal government from commandeering
state resources to enforce federal law. Thus, states can decide to not use their own resources in support
of federal law and leave implementation solely to the federal government. Put another way, permitting
behavior that the federal government prohibits does not constitute preemption, as the federal govern-
ment can invest the resources required to enforce its own policy (Garvey, 2012). This is the principle
upon which states are choosing to remove their own resources from the pursuit of the federal War on
Drugs, at least for marijuana (Kamin, 2015). Legal uncertainty arises from the additional steps taken to
promote the marijuana industry through medicinal, and especially recreational, programs.
5. From 1996 to 2014, 11 of the 23 states that adopted medical marijuana policies did so through a direct
initiative. Twenty-three states have access to the popular initiative. Of these 23 initiative states, 12
states passed MMLs and 11 of these states did so by initiative. Of the states that passed an MML in the
legislature, only one state had the initiative as an option (Illinois), a state where the law nominally
grants the option of the initiative but initiatives rarely qualify for the ballot (Bowler & Donovan,
2004).
6. There is a limitation in our analysis that warrants mentioning. The subset of states that have an initia-
tive process available to citizens (n 5 23) faces competing risks of either initiative adoption or legisla- tive adoption. Given the fact that there is likely dependence in these risks, and the relatively small
nature of our dataset, we did not model the competing risks in this article. Future work that pools
within-state marijuana policy (i.e., examines different marijuana policy choices) and among a variety
of defiant innovations will be better positioned to address this problem. The competing risks most
likely affect the estimates for initiative availability and legislative professionalism. It is also likely that
by not addressing the potential for competing risks, that we are using a more conservative test for
those two effects. The effect of initiative availability is ultimately strong enough to achieve statistical
significance in our pooled sample, but legislative professionalism is not. It is important to keep this in
mind when interpreting our results.
7. TEDS provides data on individuals that are admitted to treatment for excessive consumption of mari-
juana/hashish. These data provide the most objective measure of marijuana use for the time period
under study (Pacula, Powell, Heaton, & Sevigny, 2015). The Substance Abuse and Mental Health
Services Administration (SAMHSA) collects marijuana use data in the National Survey on Drug Use
and Health (NSDUH) survey. However, NSDUH has determined that state-level estimates for 2002
and years later are not comparable with data prior to 2002 due to methodological changes (O’Keefe &
Earleywine, 2011). While measures of TEDS include some who are sent to treatment as part of a plea
bargain (and may not be dependent), the data provide “relatively consistent coverage of states in all
years, reducing the problem of sampling over time” (Pacula et al., 2015, p. 14).
8. Both Bush and Obama Administrations are significantly distinguishable from the Clinton administra-
tion. The Bush and Obama administrations are not statistically distinguishable from one another.
9. While the results regarding vertical and horizontal influences on state adoption behavior run counter
to the typical expectations, it does not disprove that regional diffusion occurred. The standard limita-
tions of this type of macrolevel observational analysis apply, in that we cannot isolate the specific
causal pathway of diffusion. That being said, evidence for the observable implications of neighbor-
based learning and the influence of federal signals is lacking.
10. See, for example, H.R. 83, the Mobilizing Against Sanctuary Cities Act, in the 115th Congress.
Hannah/Mallinson: Defiant Innovation 419
References
9/11 Commission. 2004. Report. Washington DC: National Commission on Terrorist Attacks Upon the United States.
Allen, Mahalley D., Carrie Pettus, and Donald P. Haider-Markel. 2004. “Making the National Local: Spec- ifying the Conditions for National Government Influence on State Policymaking.” State Politics & Policy Quarterly 4 (3): 318–44.
Arrieta-Kenna, Ruari. 2016. “Sanctuary Cities Stand Firm against Trump.” Politico Magazine. http:// www.politico.com/story/2016/12/sanctuary-cities-trump-immigration-232449. Accessed January 10, 2017.
Balla, Steven J., and Christopher J. Deering. 2015. “Salience, Complexity and State Resistance to Federal
Mandates.” Journal of Public Policy 35 (3): 459–76.
Beck, Nathaniel, Jonathan N. Katz, and Richard Tucker. 1998. “Taking Time Seriously: Time-Series- Cross-Section Analysis with a Binary Dependent Variable.” American Journal of Political Science 42
(4): 1260–88.
Berry, Frances Stokes, and William D. Berry. 1990. “State Lottery Adoptions as Policy Innovations: An Event History Analysis.” American Political Science Review 82 (4): 395–415.
———. 2013. “Innovation and Diffusion Models in Policy Research.” In Theories of the Policy Process, 3rd
ed., ed. Paul Sabatier, and Christopher M. Weible. Boulder, CO: Westview Press, Chapter 9.
Berry, William D., and Brady Baybeck. 2005. “Using Geographic Information Systems to Study Interstate Competition.” American Political Science Review 99 (4): 505–19.
Berry, William D., Richard C. Fording, Evan J. Ringquist, Russell L. Hanson, and Carl E. Klarner. 2010. “Measuring Citizen and Government Ideology in the U.S. States: A Re-Appraisal.” State Politics & Policy Quarterly 10 (2): 117–35.
Berry, William D., Evan J. Ringquist, Richard C. Fording, and Russell L. Hanson. 1998. “Measuring Citi-
zen and Government Ideology in the American States, 1960–93.” American Journal of Political Science 42 (1): 327–48.
Boushey, Graeme. 2010. Policy Diffusion Dynamics in America. New York, NY: Cambridge University
Press.
Bowler, Shaun, and Todd Donovan. 2004. “Measuring the Effect of Direct Democracy on State Policy: Not All Initiatives Are Created Equal.” State Politics & Policy Quarterly 4 (3): 345–63.
Brehm, John, and Scott Gates. 1997. Working, Shirking, and Sabatoge: Bureaucratic Response to a Democratic Public. Ann Arbor, MI: University of Michigan Press.
Bulman-Pozen, Jessica, and Gillian E. Metzger. 2016. “The President and the States: Patterns of Contesta-
tion and Collaboration Under Obama.” Publius: The Journal of Federalism 46 (3): 308–36.
Chubb, John E. 1985. “The Political Economy of Federalism.” The American Political Science Review 79 (4): 994–1015.
Cole, James M. 2011. Guidance Regarding the Ogden Memo in Jurisdictions Seeking to Authorize Marijuana for Medical Use. Washington, DC: Department of Justice.
———. 2013. Guidance Regarding Marijuana Enforcement. Washington, DC: Department of Justice.
Dinan, John. 2014. “Implementing Health Reform: Intergovernmental Bargaining and the Affordable
Care Act.” Publius: The Journal of Federalism 44 (3): 399–425.
Dye, Thomas R. 1990. American Federalism: Competition among Governments. Lexington, MA: Lexington Books.
Elazar, Daniel J. 1966. American Federalism: A View from the States. New York, NY: Thomas Y. Crowell Company.
Eyestone, Robert. 1977. “Confusion, Diffusion, and Innovation.” The American Political Science Review 71 (2): 441–47.
Ferraiolo, Kathleen. 2008. “State Policy Innovation and the Federalism Implications of Direct Democracy.” Publius: The Journal of Federalism 38 (3): 488–514.
420 Policy Studies Journal, 46:2
Frendreis, John, and Raymond Tatalovich. 2010. “’A Hundred Miles of Dry’: Religion and the Persistence
of Prohibition in the U.S. States.” State Politics & Policy Quarterly 10 (3): 302–19.
Gais, Thomas, and James Fossett. 2005. “Federalism and the Executive Branch.” In The Executive
Branch, ed. Joel D. Aberbach, and Mark A. Peterson. New York, NY: Oxford University Press,
486–522.
Garvey, Todd. 2012. Medical Marijuana: The Supremacy Clause, Federalism, and the Interplay Between State
and Federal Laws. Washington, DC: Congressional Research Service, Library of Congress.
Gaumond, Peter, Carolyn Davis, and Tom Hill. 2009. “Public Policy, Treatment.” In Encyclopedia of Sub-
stance Abuse Prevention, Treatment, and Recovery, ed. Gary L. Fisher, and Nancy A. Roget. Washing-
ton, DC: SAGE, 732–42.
Goggin, Malcolm L., Ann O’M. Bowman, James P. Lester, and Laurence J. O’Toole, Jr. 1990. Implementa-
tion Theory and Practice: Toward a Third Generation. Glenview, IL: Scott, Foresman/Little, Brown
Higher Education.
Gormley, William T., Jr. 2006. “Money and Mandates: The Politics of Intergovernmental Conflict.” Publius
36 (4): 523–40.
Gray, Virginia. 1973. “Innovation in the States: A Diffusion Study.” American Political Science Review 67
(4): 1174–85.
Grossback, Lawrence J., Sean Nicholson-Crotty, and David A. M. Peterson. 2004. “Ideology and Learning
in Policy Diffusion.” American Politics Research 32 (5): 521–45.
Gugliotta, Guy. 2015. “Why Many States Now Have Stem Cell Research Programs.” Governing. http://
www.governing.com/topics/health-human-services/last-decades-culture-wars-drove-some-states-
to-fund-stem-cell-research.html. Accessed May 31, 2016.
Hickey, Walt. 2014. “Medical Marijuana Is Still the Best Deal on Pot in Colorado.” FiveThirtyEight.
http://fivethirtyeight.com/features/medical-marijuana-is-still-the-best-deal-in-colorado/. Accessed
July 2, 2016.
Huber, John D., and Charles R. Shipan. 2002. Deliberate Discretion? The Institutional Foundations of Bureau-
cratic Autonomy. New York, NY: Cambridge University Press.
Huber, John D., Charles R. Shipan, and Madelaine Pfahler. 2001. “Legislatures and Statutory Control of
Bureaucracy.” American Journal of Political Science 45 (2): 330–45.
Jacobs, Lawrence R., and Timothy Callaghan. 2013. “Why States Expand Medicaid: Party, Resources, and
History.” Journal of Health Politics, Policy and Law 38 (5): 1023–50.
Jensen, Jason L. 2003. “Policy Diffusion through Institutional Legitimation: State Lotteries.” Journal of Pub-
lic Administration Research and Theory: J-Part 13 (4): 521–41.
Jones, Jeffrey M. 2015. “In U.S., 58% Back Legal Marijuana Use.” Gallup. http://www.gallup.com/poll/
186260/back-legal-marijuana.aspx. Accessed May 31, 2016.
Kamin, Sam. 2015. “The Battle of the Bulge: The Surprising Last Stand against State Marijuana Legal-
ization.” Publius: The Journal of Federalism 45 (3): 427–51.
Karch, Andrew. 2006. “National Intervention and the Diffusion of Policy Innovations.” American Politics
Research 34 (4): 403–26.
———. 2007. Democratic Laboratories: Policy Diffusion among the American States. Ann Arbor, MI: University
of Michigan Press.
———. 2010. Examining Episodes of Limited Diffusion: The Politics of Embryonic Stem Cell Research. Paper
presented at the Annual Meeting of the American Political Science Association, Washington, DC.
———. 2012. “Vertical Diffusion and the Policy-Making Process: The Politics of Embryonic Stem Cell
Research.” Political Research Quarterly 65 (1): 48–61.
Kiewiet, D. Roderick, and Mathew D. McCubbins. 2014. “State and Local Government Finance: The New
Fiscal Ice Age.” Annual Review of Political Science 17 (1): 105–22.
Magleby, David B. 1988. “Taking the Initiative: Direct Legislation and Direct Democracy in the 1980’s.”
PS: Political Science & Politics 21 (3): 600–11.
Mallinson, Daniel J. 2016. “Building a Better Speed Trap: Measuring Policy Adoption Speed in the Ameri-
can States.” State Politics & Policy Quarterly 16(1): 98–120.
Hannah/Mallinson: Defiant Innovation 421
McCann, Pamela J. Clouser, Charles R. Shipan, and Craig Volden. 2015. “Top-Down Federalism: State Pol-
icy Responses to National Government Discussions.” Publius: The Journal of Federalism 45 (4): 495–525.
National Research Council. 2014. “The Underlying Causes of Rising Incarceration: Crime, Politics, and
Social Change.” In The Growth of Incarceration in the United States, ed. Jeremy Travis, Bruce Western,
and Steve Reburn. Washington, DC: The National Academies Press, 104–29.
Nelson, Michael, and John Lyman Mason. 2007. How the South Joined the Gambling Nation: The Politics of
State Policy Innovation. Baton Rouge, LA: Louisiana State University Press.
Nicholson-Crotty, Sean. 2009. “The Politics of Diffusion: Public Policy in the American States.” Journal of
Politics 71 (1): 192–205.
Ogden, David W. 2009. Investigations and Prosecutions in States Authorizing the Medical Use of Marijuana.
Washington, DC: U.S. Department of Justice.
O’Keefe, Karen. 2013. “State Medical Marijuana Implementation and Federal Policy.” Journal of Health
Care Law and Policy 16 (1): 39–58.
O’Keefe, Karen, and Mitch Earleywine. 2011. Marijuana Use by Young People: The Impact of State Medical
Marijuana Laws. Washington, DC: Marijuana Policy Project.
Oosting, Jonathan. 2014. Michigan Medical Marijuana Law Leads to Record Revenues, Sharp Drop in Licensed
Growers. http://www.mlive.com/politics/index.ssf/2014/01/michigan_medical_marijuana_pro.html.
Accessed October 6, 2015.
Pacula, Rosalie Liccardo, David Powell, Paul Heaton, and Eric L. Sevigny. 2015. “Assessing the Effects of
Medical Marijuana Laws on Marijuana Use: The Devil Is in the Details.” Journal of Policy Analysis
and Management 34 (1): 7–31.
Palazzolo, Daniel, Vincent G. Moscardelli, Meredith Patrick, and Doug Rubin. 2008. “Election Reform after
HAVA: Voter Verification in Congress and the States.” Publius: The Journal of Federalism 38 (3): 515–37.
Pew Research Center. 2015. In Debate Over Legalizing Marijuana, Disagreement Over Drug’s Dangers.
http://www.people-press.org/2015/04/14/in-debate-over-legalizing-marijuana-disagreement-over-
drugs-dangers/, Accessed April 5, 2016.
Rabe, Barry. 2007. “Environmental Policy and the Bush Era: The Collision between the Administrative
Presidency and State Experimentation.” Publius: The Journal of Federalism 37 (3): 413–31.
Regan, Priscilla M., and Christopher J. Deering. 2009. “State Opposition to Real ID.” Publius: The Journal
of Federalism 39 (3): 476–505.
Rogers, Everett M. 2003. Diffusion of Innovations, 5th ed. New York, NY: Free Press.
Saultz, Andrew, Andrew McEachin, and Lance D. Fusarelli. 2016. “Waivering as Governance: Federalism
During the Obama Administration.” Educational Researcher 45 (6): 358–66.
Savage, Robert L. 1985. “When a Policy’s Time Has Come: Cases of Rapid Policy Diffusion, 1983–1984.”
Publius: The Journal of Federalism 15 (3): 111–25.
Scherer, Michael. 2012. What Is President Obama’s Problem with Medical Marijuana? http://swampland.time.
com/2012/05/03/what-is-president-obamas-problem-with-medical-marijuana/. Accessed October
29, 2016.
Shelly, Bryan. 2008. “Rebels and Their Causes: State Resistance to no Child Left Behind.” Publius: The
Journal of Federalism 38 (3): 444–68.
Squire, Peverill. 1992. “Legislative Professionalization and Membership Diversity in State Legislatures.”
Legislative Studies Quarterly 17 (1): 69–79.
Steinmetz, Katy. 2016. “These States Just Legalized Marijuana.” Time. http://time.com/4559278/mari-
juana-election-results-2016/. Accessed January 9, 2017.
Stern, Ray. 2015. “Arizona Medical-Marijuana Users Consumed 10 Tons of Legal Cannabis Products in
2014.” Phoenix New Times. http://www.phoenixnewtimes.com/news/arizona-medical-marijuana-
users-consumed-10-tons-of-legal-cannabis-products-in-2014-6663992. Accessed January 9, 2017.
Tau, Byron. 2012. “Obama’s Post Promise a Pipe Dream?” Politico Magazine. http://www.politico.com/
story/2012/04/obamas-pot-promise-a-pipe-dream-075421. Accessed December 20, 2016.
Thompson, Frank J., and Michael K. Gusmano. 2014. “The Administrative Presidency and Fractious Fed-
eralism: The Case of Obamacare.” Publius: The Journal of Federalism 44 (3): 426–50.
422 Policy Studies Journal, 46:2
Tiebout, Charles M. 1956. “A Pure Theory of Local Expenditures.” Journal of Political Economy 64 (5): 416–24.
Walker, Jack L. 1969. “The Diffusion of Innovations among the American States.” American Political Sci- ence Review 63 (3): 880–99.
Welch, Susan, and Kay Thompson. 1980. “The Impact of Federal Incentives on State Policy Innovation.” American Journal of Political Science 24 (4): 715–29.
Wenner, Jann S. 2000. “Bill Clinton: The Rolling Stone Interview.” Rolling Stone (December 28). http:// www.rollingstone.com/politics/news/the-rolling-stone-interview-bill-clinton-20001228. Accessed July 15, 2016.
Woods, Neal D., and Ann O’M. Bowman. 2011. “Blurring Borders: The Effect of Federal Activism on Interstate Cooperation.” American Politics Research 39 (5): 859–84.
Appendix
Table A1. Variable Descriptions, Summary Statistics, and Sources
Variable Description Mean St. Dev.
Marijuana Seizures in kilogramsa
DEA domestic drug seizures in 10k kilograms. 34.65 16.64
Initiative Availabilityb Dummy 5 1 if the direct initiative is available in the state. 0.40 0.49 Legislative
Professionalismb Squire’s (1992) updated professionalism index—Average
of 1996–2005 values 0.18 0.11
State Citizen Ideologyc Ideology score for state government and citizens based on Berry et al. (1998) revised 1960–2013 citizen ideology series.
48.17 14.99
Evangelical Rated Rates of adherents in evangelical churches per 1,000 population.
167.00 117.78
Fiscal Healthe Fiscal health is calculated as a ratio of ((Total Revenuestate2Total Expendituresstate)/Total Expendituresstate)
0.05 0.12
Marijuana TEDSf Marijuana Treatment Episode Data Set (TEDS) collected annually by state substance abuse agencies. Admissions per 100,000 aged 12 and over.
125.23 69.99
Glaucoma Prevalenceg
Estimated state-by-state prevalence rates of glaucoma aged 40 and over (%).
1.87 0.16
Cancer Incidence Rateh
Age-Adjusted Invasive Cancer Incidence Rates by State. Rates are per 100,000 persons and are age-adjusted to the 2,000 U.S. standard population (SL Interpolation of 1999–2012 estimates)
485.60 43.43
Proportion of Neighbors with MMLsb
Proportion of geographic neighbors with MMLs at start of year.
0.09 0.18
Relative Ideologybc Distance between a state’s ideology (Berry et al., 1998) and the average ideology of previously adopting states (Grossback et al., 2004)
15.52 8.17
Time from first adoptionb Number of years since first law passed in 1996. 8.26 5.43
aConstructed by authors based on data from Drug Enforcement Agency (DEA) – http://www.dea.gov/ resource-center/statistics.shtml. bConstructed by authors. cConstructed by authors based on data from https://rcfording.wordpress.com/state-ideology-data/. dConstructed by authors based on data from the Association of Religion Data Archives. http://www. thearda.com eConstructed by authors from U.S. Census Annual Survey of State and Local Finances - https://www. census.gov/govs/local/. fConstructed by authors from Substance Abuse & Mental Health Data Archive from https://datafiles. samhsa.gov/study/treatment-episode-data-set-admissions-teds-1992-2012-nid13582. gConstructed by authors based on Prevent Blindness America from http://visionproblemsus.org/glau- coma/glaucoma-map.html. hConstructed by authors based on data from Centers for Disease Control and Prevention. https://nccd. cdc.gov/uscs/cancersrankedbystate.aspx.
Hannah/Mallinson: Defiant Innovation 423