For my first mapping exercise, I took the opportunity to investigate the conclusions of my charts and graphs lab, exploring a segments of the DC Marijuana Arrest Rates 2012-2017. As the tableau dashboard illustrates below (Figure 1), the overall rates of marijuana arrests decreased after marijuana possession was decriminalized in 2014 but then started to increase in 2016 for different types of marijuana offenses, including distribution and public consumption. Furthermore, the retraction of prohibition did nothing to reverse racial disparities in people arrested for marijuana crimes, with black men accounting for over 80% of defendants over the period.
Interested in delving deeper, I came across a prevalent criticism of the rising arrest rates for public consumption and distribution is due in part to a federal ban restricting the use of marijuana in public housing, forcing residents to risk consuming marijuana in public spaces or risk eviction.
I chose to explore this idea, mapping DC Marijuana Arrest Rates in 2016, the year arrests for public consumption start to increase, in relation to public housing areas and census data to explore the relationship between arrest locations, public housing areas, and the black communities of Washington, DC.
Process & Materials
To begin, I prepared the DC marijuana arrest rates dataset for CARTO, a cloud computing platform that provides GIS, web mapping, and spatial data science tools. To fit the formatting data requirements of CARTO, I transformed the Maryland Coordinate System GeoX and GeoY columns of my original dataset to latitude and longitude using a Python library. Next, I found my map base layers on Open Data DC, downloading the DC Census Tracts and DC Public Housing Areas shapefiles quickly with no additional transformations required.
Guided by the central understanding that the black community is disproportionately affected by marijuana arrests, I began my visualization with importing the census dataset as my ground layer, selecting a neutral gradient color of the census tract polygons by the value of the black population density. Next, I imported the public housing areas in CARTO, selecting the color red as a contrast to the gray background. Finally, I imported the 2016 marijuana arrest data, coloring the points by arrest type selecting colors that would contrast highly against the neutral backdrop.
In exploring the data, I asked the following questions:
- What is the relationship between the frequency, type and location of all marijuana arrests to public housing areas, if there is any?
- Is there a correlation between arrest rates for public consumption and proximity to public housing?
- How does mapping the arrest data visualize the racial disparities in marijuana arrests?
In my first survey of the uploaded map, I am struck by the clear racial divide so explicitly visualized, with almost all arrests clustered in or on the edges of predominantly black neighborhoods. Then, as I explored closer, I recognized the presence of red public housing areas as features of or near almost every arrest cluster on the map.
However, I could not discern any particular relationship between public consumption arrests and public housing areas, noting that distribution or possession with intent to distribute was as just as prevalent, if not more, in close proximity to public housing areas. To investigate trends public consumption further, I added a widget for arrest type so I could filter and view all public consumption arrests in isolation (Figure 2).
It then became clear arrests for consumption took place at the center city neighborhoods like Columbia Heights, Adams Morgan, and Logan circle. Although these arrests hold relative proximity to public housing areas, I began to suspect that the high density of public consumption arrests could be correlated to other factors not reflected in my map’s layers — like the presence of bars or restaurants in the area. I activated pop-up window to explore each arrest point, observing the average arrest times of early afternoon to early morning inconclusively supporting an idea not fully represented in the data.
Explore the final visualization below:
Reflections & Next Steps
Overall, the patterns in my data feel largely inconclusive beyond the clear visualization of policing practices falling along racial line in Washington, DC. Therefore, my next steps in this project would include adding a new layer of all DC crime data, as I suspect the pattern of arrest density and position would be quite similar to the marijuana arrest data.
This would therefore support the notion that an uptick in marijuana arrests in 2016 had less to do with contributing factors like the federal ban on cannabis use in public housing, and more to do with overriding issue of racial injustice in law enforcement.