Understanding the Relationship Between Neighborhood Demographics and Allegations of Police Misconduct in Chicago ================ Sokona, Ella, and Elly
Introduction:
This project was concerned with examining a social justice issue using data. The data we used was collected by the Invisible Institute about police complaints in Chicago from 1988-2018. Through FOI requests, in 2016 more than 100,000 complaint records were released to the public.
Using this data, we examined patterns in that data around racial composition of neighborhoods, the number of complaints, and complaint outcomes. We ask three main questions of this data: What are the racial makeups of different neighborhoods in Chicago? In which neighborhoods are there the most police complaints per capita? What are the outcomes of police complaints? The methodology on how we examined these questions is detailed further below. These sorts of questions have been increasingly asked of data concerning police behavior as awareness of police brutality across America has grown. While issues of over policing and police brutality have long been things that many Americans, particularly Black Americans have long said are major community issues, data empirically backs up findings that Black people and other people of color are more likely to be targets of police abuse. The questions we asked of this data are aligned with these issues. Hopefully, these sorts of inquiries inspire policy changes that make communities safer from the police.
Methodology:
We created a number of leaflet maps to give our viewers a better sense of the demographic distribution of Chicago and help contextualize later visualizations. These showed the percentage of Black, white, and Latino residents in each neighborhood.
To understand the relationship between racial makeup of a neighborhood and complaints per capita, we filtered the data set of those accused in complaints to only include entries where the complaint could be matched to an officer from a unit matching the neighborhoods in the district demographics dataset. Then we joined this with the demographic data from the districts and mutated to create a new variable showing the complaints per capita, then created a lollipop plot. We also made a leaflet map to show this data spatially.
We next aimed to visualize the final result of the complaints by district. To do this, we cleaned up the data, sorting it into usable categories of sustained, unsustained, no affidavit/cooperation, and missing. When the complaint is missing an affidavit it means that the person who made the complaint did not provide a written statement of their complaint. We then created a filled bar plot that showed the proportion of complaints that were sustained, not sustained, missing an affidavit or had missing data.
We also took the data used to create the findings visualization and, filtering for just missing, sustained, and no affidavit complaints respectively, created visualizations of each of these complaint types per capita. In these three visualizations, we choose not to include disciplined or no cooperation as this represented a negligible number of complaints (only 6 and 29 complaints out of around 70,000).
Findings/Discussion:
We hypothesized that most complaints would happen in Black neighborhoods, given the nature of police brutality and the interactions that police have with Black people in the United States. Our lollipop plot illustrated that besides two districts (Central and Near North, the 10 districts with the most complaints per capita were majority Black. In considering the outcomes of police complaints, about 87% of the investigations ended with no finding of fault. When we look at these results more closely across district, through our bar plot, Wentworth, a majority Black district which didn’t have the most complains per capita, had the most sustained complaints out of all of the districts; despite this, most districts align with the distribution of findings among the whole data, with no one district being disproportionate. It is worth noting that Central (majority white) had the second most sustained complaints per capita. It also had the most missing complaints per capita, followed by a number of Black majority districts. It is also worth noting that the top 6 no affidavit complaints per capita were in Black majority districts. All of our findings confirmed our initial expectations about what the data would reveal. Chicago’s long history of housing segregation and current continued housing segregation allowed us to see a stark contrast between the way neighborhoods with majority Black residents are policed and how majority white neighborhoods are.
Because of the limited district data available, we only looked at 25 districts; this dataset still left us with 70,000 complaints. Regardless, unless the allegation involved criminal conduct or a residency violation, anonymous complaints can’t be investigated. Thus, even with the ability to file a complaint, 27.8% of the data includes complaints from those who didn’t provide an affidavit, mostly from majority Black districts, and therefore don’t get investigated. While it may be harder to check the validity of anonymous complaints, we wonder if this requirement is preventing complaints from filing as a result of safety or other personal reasons. When complaints are investigated, out of 47,000 complaints, 6% justified disciplinary action and .008% (6) led to Officers being reprimanded, suspended for a set number of days, or separated. Therefore, while Black complaints are more likely to complain against police officers in this dataset, it is unlikely that the complaint would even justify disciplinary action and even lead to any action regardless of race.
Further research should look at what type of allegations were more likely to be sustained and if there’s a correlation between the type of allegations and the person who made it. Research should also be done on the effectiveness of filing complaints against police officers and about the power they have, which prevents any change happening. If given more time, we would want to further explore how complaints have changed over the years, and more about the correlation between the race of the officer versus the race of the accuser in these complaints.
Our presentation can be found here.
Data
This data is publicly available at https://github.com/invinst/chicago-police-data. This dataset is massive and broken into many different dataframes so much of our time at first was spent choosing and merging the best data frames together.
References
Links to the raw data:
Community demographics: https://github.com/invinst/chicago-police-data/blob/master/data/context_data/district_demographics/Districts.csv
Accused officer: https://github.com/invinst/chicago-police-data/blob/master/data/older_data/cleaned_data/complaints-accused_2000-2016_2016-11.csv.gz
Content of complaints: https://github.com/invinst/chicago-police-data/blob/master/data/older_data/cleaned_data/complaints-complaints_2000-2016_2016-11.csv.gz
Active duty officer: https://github.com/invinst/chicago-police-data/blob/master/data/context_data/CPD%20Employees%20active-duty-only.csv