In the District of Columbia (DC) the laws related to the recreational use and possession of marijuana changed at the effective dates of two major milestones: Marijuana Possession Decriminalization Amendment Act in 2014 and Initiative 71 in 2015, a voter-passed ballot measure to legalize the possession, recreational use, and transfer of up to one ounce of marijuana for adults over 21. However, marijuana is still illegal under federal law and Congress barred DC from regulating the sales and taxation of the substance in December 2019. Therefore, I set out to explore how both the successes and shortfalls in legalizing the use and sales of marijuana in the District has affected arrest rates and demographics by analyzing the DC marijuana arrest data from 2012-2017.
Process & Materials
To begin, I downloaded the CVS text file from DC.Gov Metropolitan Police Department and opened the file in OpenRefine, an open source desktop application for data cleanup and transformation. In previewing the data in OpenRefine, I consulted the dataset’s accompanying data sheet to gain familiarity with the data’s structure and readiness for data manipulation. Fortunately, the dataset required very little cleaning and I exported for upload to Tableau Public, a free service for publishing interactive data visualization to the web.
Within Tableau, I started by exploring the overall trends in data arrests from 2012-2017 with time-series visualizations like area graphs, bar graphs, and line graphs (Fig. 2). I could I then explored larger trend in more detail adding dimensions like arrest type, defendant age, and race asking the following questions:
- What changes, if any, can be seen in the arrest rates due to the Marijuana Possession Decriminalization Amendment Act and Initiative 71?
- How has the decriminalization and legalization of possession and private, recreational use affected the types of arrests still being made?
- Who is being arrested and what impact, if any, has been made by legalization efforts?
Guided by these questions, I identified the central narrative for my dashboard and spent considerable time considering how to illustrate it, clearly and succinctly. To aid this process, I reviewed samples of similar datasets on sources like DrugPolicy.org and in The Washington Post.
With the goal of establishing a martini glass narrative structure, I ground my dashboard with a title, caption, and annotated area chart in the top left as an introduction to the dashboard’s storyline:
Although overall rates of arrests have drastically decreased since the decriminalization and legalization of marijuana in DC, arrest types related to the sale and supply of marijuana have increased in recent years while the disproportionate rate of arrests of young, black men have remained constant over the period.
To the right of the area chart, I explore the changes in arrest type with both a stacked bar graph and a line graph utilizing color to signify type. A drastic drop can be seen in arrests for possession, the most common arrest type prior to 2015, with recent increases in the rates of arrest for distribution and possession with intent to distribute.
Next, in the lower second half of the visualization I establish that young black men are disproportionately arrested for marijuana arrests over the time period, visualizing the change in reason for arrest over the period. For clarity, I utilized Tableau’s ‘bins’ and ‘grouping’ to create age ranges and ‘other’ race categories while relying on labeling instead of introducing additional color signifiers. I worked to keep my palate coherent and limited, marking broad trends in a light purple between visualizations as a way to guide a user’s comprehension.
Reading left-to-right, top-down the user ends again with a visualization communicating a proportional growth in supply-chain related marijuana arrests, emphasizing that racial disparities in policing is not solved just by decriminalization and that arrests are likely to continue without a legal marijuana industry.
Explore the visualization here:
Reflection & Future Direction
Although I think my first visualization in Tableau is ultimately successful for the purposes of demonstrating a range of competencies in the software and communicating broad trends, I think my formatting and structuring of the dashboard leaves great room for improvement. I largely experimented and settled upon a layout and size for the visualization by accident only considering the ‘fit’ of my graphs after running into formatting issues upon publishing to my Tableau Profile. For future visualizations, I will map out and structure my dashboard much earlier in the design process.
Furthermore, I think that this project could be expanded to further highlight the disproportionate arrests of young black men by including statistical visualizations of DC’s overall population and recreational marijuana use. Mapping the data could be potentially useful, directly comparing the arrest location data with census demographic information.