Among the small number of cities in LA County that allowed dispensaries, the City of Los Angeles had a significant and potentially confounding influence that varied between odd and even years to such a degree as to effect outcomes in students’ marijuana use from year to year. This concern was addressed by using a combined 2 school year period as the unit of analysis for time, which halved the amount of data points available for the trend analysis but provided a much more reliable estimate of trends in students’ marijuana use behaviors over the twelve-year study period. CHKS data is not without limitations. The CHKS survey data set that I obtained represents only the public high schools in LA County and the marijuana use behavior of public high school students may differ from students at private high schools who are not surveyed. Furthermore, the CHKS survey and sampling strategy was designed to measure student health and school climate over time by school district, rather than by city. In many cities these units are interchangeable because there is one district high school district per city, but not this is not the case in every city. Although administering CHKS is a requirement for public schools receiving Tobacco-Use Prevention Education from the State of California , participation by school districts, schools, and students is voluntary. Participation for some schools is relatively low , although offering school districts incentives for participation after 2011 was effective in improving participation . Even though CHKS data has been sampled proportionally to generate population-based reports of student behavior at the state level, at the County level, where I have included every school that participated apart from special education schools, the protocol of voluntary participation makes the CHKS study sample more of a convenience sample. Ideally,vertical growing systems the results of this study should not be generalized outside of LA County or to students at private high schools and should be followed by more extensive data collection efforts using a large enough population-based sample to study the impacts of city policies on adolescent health behaviors. CHKS has been administered at a large enough majority of schools in LA County to provide estimates of student marijuana use that schools from 76 of the 88 incorporated cities in the County were represented in the data at some point during the study period.

In the cross sectional analysis using the 2015/2016 and 2016/2017 school years the number of cities that had schools that participated in the survey was less representative. Only 53 cities out of the 88 cities in LA County had schools that participated in the CHKS survey in the 2015/2016 and 2016/2017 school years. Although the cities that participated represent 87% of the LA County population, in an analysis of the impacts of city-level policies this is a serious limitation. Despite these limitations, the very large sample size, the opportunity to make comparisons to state-level data, and the consistency of data collected over multiple years make CHKS a valuable tool to measure substance use among California students. The Los Angeles County Department of Public Health has also used the CHKS in a recent impact assessment of the potential impacts of allowing retail and medical marijuana outlets in the unincorporated areas of LA County on the health and safety of LA County residents . The addresses of all the public high schools in LA County were obtained online from a secondary source, the California Department of Education website . School directory data is available for download as an Excel file and contains the address and geographic coordinates of each public school within California, as well as administrative details such as school type .The school directory file was downloaded, filtered to obtain all the public high schools that served LA County students during the 2015/2016 and 2016/2017 school years . The school addresses were then geocoded using ArcMap 10.4 to generate latitude and longitude coordinates for each school and thus identify where they were located within LA County. I then performed a spatial join to the city boundary shapefiles available from the LA County GIS portal to identify which city each high school was located in, as this information is not available in the CHKS dataset. The geocoded high schools were matched to schools in the CHKS dataset by their CDS code, a unique ID provided by the California Department of Education to all California public schools. Once matched to the CHKS data it was possible to link the geographic location of the school to the students’ behavioral data, which included rates of lifetime and recent marijuana use and perceptions of the risk of marijuana use, along with other behavioral and demographic data. Municipal codes and zoning laws are public information and are generally published by cities at their own expense for the benefit of city residents. The cities in LA County used online municipal database services like Municode.com and American Legal Publishing Corporation to publish searchable directories and archives of their city ordinances and zoning codes.

Using these services, I was able to determine whether and when the cities in LA County passed ordinances banning or allowing dispensaries between 2005 and 2017. By September 2016, 79 out of the 88 cities in LA County had either specifically banned dispensaries or had zoning laws that prohibited any kind of land use not expressly listed in the municipal or zoning code. Among the remaining cities, six explicitly allowed dispensaries, and three had no business districts and therefore no commercial zoning codes. A simple tally of changes to medical marijuana ordinances within just two years documents that this was a dynamic era for city-level marijuana policy in LA County . Data characterizing city dispensary ordinances were obtained via primary data collection. I reviewed municipal codes for the 88 incorporated cities within Los Angeles County semi-annually from August 2014 and through August 2016. For cities where municipal codes were available online, search terms such as “marijuana,” “cannabis,” and “dispensary” were used to find the sections of municipal and zoning codes that regulated medical marijuana dispensaries. For the cities without municipal codes accessible online, City Clerks were contacted via email and phone to obtain the full text of the city ordinance, but this was only necessary for the cities of Avalon and Maywood. As I compiled the city policy data, I created a database listing whether the dispensary policies of the 88 incorporated cities within Los Angeles County banned or allowed dispensaries. The database included links to the full text of each ordinance and detailed notes about how it was obtained for each city. Three main categories of marijuana policy emerged as data collection proceeded; policies that addressed: storefront dispensaries, cultivation, and delivery services. In the case of delivery and cultivation policies, many communities did not explicitly state in their municipal code if these activities are allowed, but unless a local ordinance bans these activities, the local law defaults to the State law,curing marijuana which allows personal use cultivation and medical marijuana delivery to qualified medical marijuana patients. Several cities presented special cases in this analysis. In the absence of a policy banning or specifically allowing dispensaries, the cities of Long Beach, Los Angeles, and West Hollywood initially allowed dispensaries to operate according to California law before passing ordinances that restricted their number and enacted additional regulations . In the case of Long Beach, these restrictions were eventually followed by a dispensary ban , which has since been reversed again by a local ballot measure, Measure MM . As these three cities were known to have allowed dispensaries to operate openly within city borders after they became legal under state law , they were coded as allowing dispensaries starting from the 2005/2006 school year forward for this analysis. Although dispensaries likely cropped up in other cities and in the unincorporated areas of the County in advance of an official policy allowing them, my research of news reports and the background provided in city ordinance texts has not identified any other cities where dispensaries were sanctioned the way they were in Long Beach, Los Angeles, and West Hollywood prior to these cities enacting local ordinances that restricted their operation beyond California law. All the other cities and unincorporated LA County were therefore coded as not allowing dispensaries until an ordinance was passed that specifically stated that they were allowed. I recorded the location of dispensaries to measure their presence in communities directly, rather than assuming a city ban meant that there were no dispensaries operating in a city.

As soon as I started collecting data on the number and locations of dispensaries in LA County it became clear that city dispensary bans were not a reliable determinant of whether dispensaries were actively operating in a city. This backed my theory that it would be important to adjust for discrepancies between expectations based on city policy and the practical availability of marijuana from dispensaries in a particular city based on how dispensaries were operating there. Prevention research supports the idea that more convenient access to substances that are legal for adults, such as tobacco or alcohol, often has the end result of creating easier access for youth . This finding implies that youth living in or attending school in a city that allows dispensaries might obtain cannabis more easily or more often from adults in their social network. Considering that adolescents report older relatives and the illicit market as their primary sources of cannabis , a dispensary ban making access less convenient for adults could have the additional effect of making it less conveniently obtained by teens. The dispensary location data were not obtained from an official source and were intended to link medical marijuana customers to marijuana businesses rather than for research purposes. However, a greater limitation than the source of the marijuana location data is how quickly it can change. The marijuana market and policy environment in LA County is an environment where dispensaries are frequently shut down and found to crop up in other locations . Using the verified counts of the dispensaries helped address this limitation and assure that the influence of dispensaries was more contemporaneous with when marijuana use was measured among students . Once obtained and de-duplicated, the addresses of the dispensaries and LA County Public High Schools were geocoded using ArcMap 10.4. Geocoding is a process where a Geographic Information Systems software program matches an address to a database that contains latitude and longitude coordinates for all of the known addresses in an area, and then places the address locations as points on a map. For this analysis I used the “LA County Locator”, which is publicly available for download from the County of Los Angeles GIS Portal , a website that is maintained by the County of Los Angeles GIS Steering Committee to serve as a central location for GIS data created, maintained, licensed, and stored by LA County government agencies. After placing the geocoded addresses of the dispensaries and high schools within a map of LA County as points, I associated the shapefiles that placed the location of the dispensaries as points within LA County with shapefiles that defined the borders of the cities and unincorporated areas of LA County using a spatial join. When a point layer is joined to a polygon layer, a count field is created that tallies the number of points that fall within the boundaries of each polygon . I used this process to create dispensary counts per city in ArcMap. I then imported the .dbf file that ArcMap creates as part of each shapefile into SAS to be linked with the other data sources by city name. The city boundary shapefiles I used contain information about the population of the cities and the unincorporated area within LA County, which I used to account for the different sizes of the cities in LA County by calculating rates of dispensaries by the city and unincorporated area population. To do this, the counts of MMDS per city were divided by the population of the city and multiplied by 10,000 to obtain a rate of dispensaries per 10,000 residents.