Introduction

As a city with one of the highest homicide rates and most racial segregation in the country (Bertlatsky), Chicago, Illinois is an ideal study site to investigate the relationship between race and the statistical likelihood of homicide occurrence.  The frequency of fatal shootings in on the rise, but is not felt or seen equally in all areas of the city (Chicago Tribune).  Certain police districts have seen an relatively small increases in the number of homicides while others have seen increases of up to 350% (Chicago Tribune).  This discrepancy has been noted by the city’s residents with whiter areas generally perceived as safe and blacker areas generally perceived as dangerous (Bertlatsky).  This analysis was conducted to test whether or not there is statistically significant evidence supporting the claim that areas with higher proportions of black residents do in fact have higher homicide rates.  Specifically, we wanted to see what the probability of any number of homicides occurring within the city,  the frequency of homicides in each neighborhood, and the probability of homicide occurring in each neighborhood depending on that particular neighborhood’s racial demographics. 

Methods

The study site for this analysis were all census tracts within Chicago’s city limits.  Homicide and demographic data were collected from the US Census Bureau, the Chicago Police Department, and the City of Chicago data portal.  We began by first analyzing the crime data.  This analysis was focused on the homicides that occurred in 2015, so first the crime data from CPD was filtered to contain only entries in which the crime was “Homicide” and the year was “2015.” Since the time and date of each crime was recorded in the same column, we created a new “Time” column and a new “Date” column so that we had a way to link the number of homicides that occurred on a single day to one particular date.  Once this was done we were able to determine how many days at least one homicide occurred in the city and the number of days x number of homicides occurred.   We were able to display this information in a histogram as well as plot the locations of each homicide in a latitude/longitude grid. 

For the census information, I chose to use 2014 ACS Five-Year Estimates.  I chose to use five year estimates because, for the questions we are trying to answer, accuracy is more important than currency;  in other words, this data may not me the most current, but since it is an aggregation of several years of data, it should give a more accurate measurement of each block’s demographic breakdown which is essential for our analysis to produce any meaningful results. Next we “clipped” the entire county’s census tract records to just those blocks that lied within Chicago’s city limits and changed any NA values to zero.  Then we simplified the census data set to only include the information we needed (GeoID, White population, Black population, and Total population) and merged this new data set with the data set containing individual census tracts.  With the merged data set we then had the information we needed to create a map displaying the locations of each homicide event and the proportion of each census block’s population that was black.  We then could calculate the number of homicides that occurred in each block and determine if there was any significant relationship between the demographic profile of a census block and the number of homicide events that occurred there. 

Results

Map of Chicago Homicide Location and Demographic Distribution.  Darker shades of green represent census tracts with higher percentage of its population composed by black individuals and red dots indicate the locations of 2015 homicide events.

Latitude/Longitude plot showing the location of homicides throughout the city.  The larger the circle, the more homicides occurred in that area.

Histogram of the number of homicides that occurred per day.

(1) The probability of exactly n homicides per day occurring:

(2) Probability of at least n homicides per day occurring:

(3) 10 Day Homicide Occurrence Binomial Distribution

(4) Neighborhood Homicide Frequency

(5) White Neighborhood Homicide Probability

(6) Black Neighborhood Homicide Probability

Results

The results of this analysis give us a quantitative description of Chicago’s homicides and provides us with a better understanding of the relationship between demographic characteristics of certain neighborhoods and the likelihood of homicide occurrence.  We found that city-wide there is a high likelihood of at least one homicide occurring each day (almost 70%) but that the likelihood of multiple homicides is less (about 40% for two homicides a day, and 16% for three).   That being said, the majority of the city is relatively safe with almost 600 (69%) tracts not having a single homicide.  We also found that racial demographics only seem to have a strong effect on homicide probability in the extremes.  For example all neighborhoods between ten and ninety percent black or white all had homicide probabilities under 5%.  However, the probability of homicide occurring in neighborhood that was less than ten percent white was 67.4% and the probability in a neighborhood that was between ninety and one hundred percent black was 54.7%.  Thus, our results imply that a high percentage of black residents or a low percentage of white residents increases an area’s probability of homicide occurrence.

Enhancement to this study will primarily come in controlling for other variables.  Modeling homicide probability solely off of black/white population proportions is overly-simplistic and is surely not capturing other variables that have an effect on an area’s homicide rate.  A future study should include other variables such as a census block’s average income, unemployment rate, age structure, high school graduation rate, and others to capture more of the variation in homicides across the city and provide additional explanation. 

Sources

Bertlatsky, Noah. “How Bad Is Violence in Chicago? Depends on Your Race.” The Atlantic. Atlantic Media Company, 26 Sept. 2013. Web. 10 Apr. 2016. <http:// www.theatlantic.com/national/archive/2013/09/how-bad-is-violence-in-chicago- depends-on-your-race/280019/>.

“Charting Chicago’s Violence so Far in 2015 (infographic).” Chicago Tribune. Chicago Tribune, 5 Sept. 2015. Web. 10 Apr. 2016. <http://www.chicagotribune.com/ct-charting- chicagos-summer-homicides-shootings-infographic-20150904-htmlstory.html>.