The extent to which these two measures may differ is dependent on enforcement

Visual cues to marijuana use such as billboard and magazine advertising for cannabis are strongly associated with adolescents’ intentions to use marijuana and eventual use . The presence of dispensaries may be analogous to advertising because many dispensaries in LA County use their exterior walls as advertising space like any other store . It is therefore possible that repeatedly seeing dispensaries located near their school will have an impact on high school students’ likelihood to use marijuana, even if they are not able to obtain it directly from these outlets. Furthermore, among people who have already used a psychoactive substance, visual reminders of that substance activate a chemical response that triggers a craving for the substance, increasing their propensity to use substances to which they are frequently exposed to reminders of . This means that among high school students who have already tried marijuana, the sight of dispensaries may trigger cravings for marijuana and thus increase their propensity to use it. Although measuring individual enforcement efforts by city or county police or code enforcement officers was beyond the scope of my analysis, enforcement is nevertheless an important construct in the conceptual model for this dissertation and the analyses that follow. In the conceptual model below, the effects of city policies banning or enacting stricter regulations on storefront dispensaries are hypothesized to be dependent on effective enforcement. For example, the impact of a city policy allowing dispensaries on adolescent substance, such as dispensary density in a city,cannabis curing is determined not just by how many dispensaries the city ordinance allows, but also on how many dispensaries are actually in operation.

Similarly, city policies that allow dispensaries often require them to be located a specific distance away from schools, but dispensaries have often been found located near schools in violation of these policies. Keeping dispensaries away from sensitive areas is therefore also dependent on effective enforcement.Key informant interviews conducted with city officials as part of the LA County Department of Public Health Cannabis Health Impact Evaluation indicate that preventing unlicensed outlets is a central goal for city dispensary ordinances. Therefore, the number of unlicensed outlets per 10,000 city residents was included as a proxy measure of the effectiveness of the city dispensary ordinance, with a higher proportion of unlicensed dispensaries per residents indicating less effective enforcement. Figure 3.2 presents the conceptual model of this dissertation. At the individual level, the focal relationship is between city dispensary bans and students’ self-reported marijuana use . The additional variables that explain and influence the focal relationship are described in the research questions and hypotheses that follow. The conceptual model also presents the backdrop of potentially confounding external influences, which include a general trend toward greater acceptance of marijuana use in American society, changes in state laws that have seen the majority of U.S. states enact laws that allow some level of access to marijuana, and changes in the Federal government’s stance on enforcement priorities concerning marijuana. Although these changes occur outside the scope of this dissertation, they are relevant from the standpoint of the Social Ecological Model and the Drug Normalization Framework and are accounted for in the study design wherever possible.

This study draws upon diverse data sources and uses several different methodological approaches to arrive at a greater understanding of the impact that the dispensary bans enacted throughout Los Angeles County over the past decade have had on high school students’ marijuana use. In the five descriptive and explanatory data analyses for this dissertation that follow, I first used primary and secondary data sources to construct an administrative data set that documented which cities in LA County have enacted medical marijuana dispensary bans. Second, I used school-based CHKS survey data to measure marijuana use among 9th and 11th grade students in each city. Third, I geocoded school addresses from the California Department of Education school directory and mapped their locations within city and county boundaries. Fourth, I linked the CHKS data set to the geographic location of the schools using the unique ID assigned to each school by the California Department of Education. Lastly, I used street addresses from commercial listings of marijuana businesses to establish the location of dispensaries in cities and near schools. These data sources and methods were required to compare long term trends in student marijuana use by whether a city bans or allows dispensaries and to test different ways that city bans may influence high school students’ marijuana use . The population of interest for this dissertation was adolescents living within LA County. The study population was 9th and 11th grade students at public high schools that participated in the CHKS survey between the 2005/2006 and 2016/2017 school years. Students’ demographic and socio-economic characteristics and their marijuana use behavior were recorded using restricted-use secondary data from a school-based survey of student health and school climate, the California Health Kids Survey. Dispensary policies for each of the 88 cities in Los Angeles were obtained from online municipal code databases and categorized by whether they allowed or banned dispensaries. City dispensary policies were linked to student behavior by the city where their high school was located, which according to California public school residency requirements is most often the city where they live .

Addresses of dispensaries were then downloaded from commercial listings of marijuana businesses and mapped to determine their location and density within cities and near schools. Each of these data sources were required to address the central question of this dissertation; whether dispensaries bans prevent adolescent marijuana use. The dependent variables,drying weed lifetime and recent marijuana use, are self-reported data from the CHKS survey. The independent variable is whether each city had a dispensary ban. I also conducted a mediation analysis using measures of marijuana density within cities and relative to schools as well as student perceptions of the health risks of marijuana use to test whether the effectiveness of MMD bans was dependent on any of these variables. The survey sample used for this dissertation was comprised of students who completed the California Healthy Kids Survey at LA County public high schools between the 2005/2006 school year and the 2016/2017 school year. The CHKS is a statewide survey that covers a range of health perceptions and behaviors and is administered annually in school districts throughout the state. The initial population was 532,200 LA County high school students who participated in the CHKS survey between 2005/2006 and 2016/2017. In Los Angeles County during the school years studied, most high schools administered the CHKS every other year to 9th and 11th grade students. However, about 10% of the surveys each year were administered to 10th and 12th grade students or students who chose categories of “don’t know” or “ungraded/other” for grade. These students were excluded to draw more precise conclusions about the behavior of students in 9th and 11th grade and for comparability with the other research published using CHKS survey data, which focuses on these grades. After excluding a handful of remaining students who attended special education schools or who were missing important data, the population available for analysis over the 12 years of the study period numbered 487,354. Criteria for exclusion from the study sample are presented below in Table 4.2. The study period spanning the 2005/2006 school year and the 2016/2017 school year was chosen for several reasons. An original motivation for this study was to learn whether rates of marijuana use among LA County high school students increased overall as the number of cities in the County that allowed dispensaries increased after medical marijuana entered the formal marketplace after SB 420 in 2004 allowed medical marijuana collectives to operate as businesses.

The endpoint for the study period, the 2016/2017 school year, preceded the licensure of non-medical marijuana storefronts throughout the state of California and LA County that began in January of 2018. Ending data collection in 2017 allowed this analysis to focus on the impacts of medical marijuana dispensaries and to serve as a comparison point for non-medical marijuana sales after 2018. As noted above, the data presented in this dissertation were drawn from multiple sources. These sources are reviewed in more detail below. The data source used to measure high school students’ perceptions of the health risks of marijuana use and marijuana use behaviors is a restricted-use secondary data set obtained from a state-level survey of California middle and high school students; the California Healthy Kids Survey . I documented whether the 88 incorporated cities within Los Angeles County had ordinances that banned or allowed dispensaries by reviewing municipal code texts using online municipal code databases such as Municode.com and categorizing city dispensary policies according to whether or not they allowed dispensaries and several other criteria . The number and location of dispensaries within each city were obtained from online dispensary listing and rating services such as Weedmaps.com, which I then used to map dispensary locations using ArcMap 10.4 geographic information system mapping software . The addresses of the high schools came from the California Schools Directory, which was downloaded from the California Department of Education . The CHKS survey is the largest statewide survey of resiliency, protective factors, and risk behaviors in the United States . It administered annually and anonymously at most public schools in California to measure middle and high school students’ attitudes and behaviors related to substance use and other health behaviors. Collecting data on student substance use has been an important goal of the survey dating from its inception. The precursor to the CHKS survey was the California Student Survey of Substance Use , which began collecting data from a representative state sample of secondary students in 1985. Over time, the focus of the CSS was expanded to include questions on other health-risk behaviors, resiliency, school climate, and school safety, which then formed the bulk of the CHKS Core Module when it was developed in 1998. In 2003, the California Department of Education mandated that CHKS serve as the primary data collection tool to document change in alcohol, tobacco, and drug use among California schools , which means that all school districts that receive funding under the federal Safe and Drug Free Schools and Communities Act or state Tobacco Use Prevention Education program must administer the CHKS survey at least once every two years and report the results publicly . As a consequence, the CHKS survey is administered by the majority of California secondary schools every other year on a staggered basis that means data is available for every year at state level but may need to be aggregated into two-year ranges to capture data for all the schools in a region. In terms of psychometric properties, a 2007 evaluation of the CHKS survey’s psychometric properties indicated that the survey exhibited good internal consistency, adequate reliability, and demonstrated measurement equivalence across racial/ethnic groups, males and females, and grades . The current iteration of the CHKS survey is built around a general Core Module and five optional supplements. The analyses presented in this dissertation relied on the Core Module, which assesses demographic information, substance use, exposure to school violence, and other behaviors that contribute to physical and mental health. Most of the items used in the CHKS Core Module were derived from the biennial California Student Survey of Substance Use and the Centers for Disease Control and Prevention Youth Risk Behavior Surveillance System . The analyses presented throughout this dissertation were conducted using two combined school years. This was necessary because school districts generally administer the CHKS survey every two years, at 9th and 11th grade. Therefore, on any given single school year, half of the schools in LA county may not have administered the survey, which could have introduced bias into the analysis. Preliminary analyses indicated that across all schools in the County, the number of participants for odd to even years varied in tandem with whether a majority of LA Unified School District schools had administered the survey that year. For example, the average number of City of Los Angeles schools on odd years was double the average number on even years.