Data Visualization Workshop (Politicians of LA county)

What is your story?
Our group was assigned to play the role of the Politicians of LA county. We brainstormed several storylines that a local politician could adopt, perhaps for an upcoming election campaign or as a PR activity while in office. Possibilities included adopting a “tough on crime” stance, a campaign to celebrate the accomplishments of the police department, an appeal to increase or decrease funding for the police department, to pass a piece of legislation regarding the stop & frisk programs, or to develop a more effective stop & frisk program using data-driven evidence. Given the data set we had access to, our group settled on the story of a local politician who is advocating for reform of the police department’s stop & frisk policies through enhanced sensitivity training. We sought to highlight the inconsistencies in how the program was implemented and draw attention to the wide range of individual officer practices. We felt like a politician would adopt this stance to maximize efficiency and provide legitimacy to LAPD practices with concrete data while simultaneously appealing to the potential cost savings and social justice angles of the story. We also thought that we would get more credence as a politician running for office, if we offered a constructive and proactive measure to tackle problems as opposed to simply showing data.

How does the selected data support your story?
This is a table that shows the divisions who had over 5000 stop and frisks. There is clearly a large discrepancy between the success rates. Some areas are as low as 4% and some are as high as 50%. It seems like bias may be playing a prominent role in some locations and that may need to be addressed with bias and sensitivity training for officers.

Division	Number of Frisks	Frisks Ending in Arrest
VALLEY TRAFFIC	53558	4%
METROPOLITAN DIVISN	45679	50%
WEST TRAFFIC	40917	5%
CENTRAL	34478	26%
CENTRAL TRAFFIC	34368	4%
HOLLYWOOD	32014	31%
SOUTH TRAFFIC	30556	7%
PACIFIC	27871	9%
SEVENTY-SEVENTH	26271	40%
SOUTH EAST	25575	34%
SOUTH WEST	23064	32%
NEWTON	21790	38%
NORTH EAST	19612	25%
HARBOR	19246	29%
FOOTHILL	18201	43%
WILSHIRE	17955	19%
OLYMPIC	17898	24%
RAMPART	15552	32%
HOLLENBECK	14529	28%
VAN NUYS	14506	31%
WEST VALLEY	14328	15%
DEVONSHIRE	13990	14%
TOPANGA	13698	27%
NORTH HOLLYWOOD	13567	17%
MISSION	12325	38%
WEST LA	8046	14%

What data did you omit and why?
We did not look at any of the data about the person who was stopped (e.g. their race or gender). This is because our visualization focused on the story of how successful stops and frisks are and which officers may be making too many stops without leading to an arrest or citation. This is because as politicians we are interested in using our citizens’ tax dollars efficiently so we want to conduct training to help officers make fewer and more effective stops. (Note: For the purposes of this visualization we are defining successful stops as those that lead to post-stop activity. This dataset does not allow us to address whether the stop or the subsequent citation or arrest were justified on legal or ethical grounds nor does it address the efficacy of stop and frisk programs in terms of overall well-being. These questions are very important but are outside of the scope of this dataset and visualization.) We also did not include any of the time (date and time of day) data on the stops because we are interested in identifying trends for particular officers or divisions rather than looking for time-based trends.

How does the representation support your story?
We tried various visualizations to analyze efficiency of frisks – bar charts, line graphs, scatter plots etc. However, what stood out from a specific scatter plot (after we used Tableau to obtain a first look into what the data showed) where we plotted Stops per officer v Number of Successful Stops was that the efficiency of frisks was to a great degree, poor. In the representation below, while it was reasonable to expect less successful stops with less stops, the fact that officers making high number of stops returned poor turnover rates of successful stops stood out. By investigating individual officer turnover rates in correlation with graphs that explore location, times, ethnicities of people frisked in addition to a study of demographics of people living and travelling through these locations, we would devise a sensitivity training plan. We would also use this representation to monitor and target moving the dots to the right side as a short term measure, where we can see the overall increase of the number of successful frisks, and then moving the dots downward as a long term measure, where we increase the efficiency of successful frisks.

What visual metaphor(s) did you use, and why?
Since our unit of analysis for this visualization is an individual police officer, we used a scatterplot in which each dot represents an officer. The number of stops and successful stops are represented spatially following the convention of larger numbers being further up on the y-axis and further to the left on the x-axis. We also use color to encode divisions.