Police Killings and Racial Bias

Police killings in the United States have been a source of significant concern and controversy for many years. These incidents often involve the use of deadly force by law enforcement officers, resulting in the death of individuals, frequently during encounters such as arrests or confrontations. Despite the increased data visibility into the problem in recent years, police killings of civilians continue to occur at an astonishing rate. While the increasing cases of police brutality are a menace to the society as a whole, the killings of Michael Brown, George Floyd, and Breonna Taylor, among many other, have raised questions regarding racial biases. Through further analysis, this study seeks to answer if police killings are disproportionately targeted towards certain minority races.

# Data Cleaning
# Loading input data file downloaded from the Fatal Encounters website 
data <- read.csv("FATAL_ENCOUNTERS.csv")

# Filtering required data and renaming columns
summary(data)
data_filter <- data %>% 
  select(Unique.ID,
         Name, 
         Race,
         Race.with.imputations,
         Imputation.probability,
         Date.of.injury.resulting.in.death..month.day.year.,
         Location.of.death..city.,
         State,
         Latitude,
         Longitude) # an unexplained error was encountered while renaming columns using select() or rename() and therefore the column names were left as is

# Separating the date column to get a separate year column
data_filter_1 <- data_filter %>% separate(col = "Date.of.injury.resulting.in.death..month.day.year.", sep = "/", into = c("month", "day", "year")) 

# Dropping NAs
data_nona <- data_filter_1 %>% filter(!is.na(Unique.ID))

First, we look at the total number of police killings. The chart below shows that the reported police fatalities have increased through the years. A stark rise can especially be observed after 2013. According to the dataset, over 2,000 people have died in the year 2020 and 2021 as a result of police violence.

# Grouping and summarizing total number of deaths by year
total <- data_nona %>% mutate(count = 1)
tot_yr <- total %>% group_by(year) %>% summarise(tot_deaths = sum(count))
tot_filter <- tot_yr %>% filter(year>2012)

# Plotting data
plot <- ggplot(tot_yr, aes(year, tot_deaths)) + geom_point(color = "brown", size=4) + ggtitle("Total deaths due to police encounters") +
  xlab("Year") + ylab("Deaths")
plot

But while the deaths have been increasing, it tells us little about racial bias, if any. For that we look at who are the most common victims of police violence. The waffle chart below stacks the deaths of people belonging to minority communities against the total deaths and shows that 37 out of every 100 people killed in police killings since 2000 belonged to a minority race. This proportion mostly is lower than the average for most years except the period between 2011 and 2019.

# Grouping and summarizing total number of deaths by year and race
minority_race <- total %>% mutate(minority = ifelse(Race %in% c("African-American/Black", "Hispanic/Latino", "African-American/Black African-American/Black Not imputed", "Asian/Pacific Islander", "Middle Eastern"), 1, 0))

race_yr <- minority_race %>% group_by(year) %>% summarise(tot_death = sum(count), minority = sum(minority))
race_yr1 <- race_yr %>% mutate(pct = minority/tot_death*100)
# Waffle chart
x <- c("Minority Races" = 37, "Other Races" = 63)
waffle(x, rows = 10, colors =  c("brown", "grey")) + ggtitle("People belonging to minority races every 100 people killed since 2000")

#Bar chart
ggplot(race_yr1, aes(x=year, y=pct)) + 
  geom_bar(fill = "brown", stat = "identity") + coord_flip() + geom_hline(yintercept = 36, color="darkslategray", linetype = "longdash") + xlab("year") + ylab("Percentage") + ggtitle("Proportion of minority races affected year on year")

Taking a closer look at the minority communities that we previously grouped together, shows that most of the victims were African-American followed by the Hispanic/Latino community. While a clear variation can be seen within races but there is more or less no change through the last five years.

# Grouping data by race for the past 5 years
ba_input <- total %>% filter(Race %in% c("African-American/Black", "Hispanic/Latino", "Asian/Pacific Islander", "Middle Eastern")) 
ba_input1 <- ba_input %>% filter(year %in% c("2021", "2020", "2019", "2018", "2017"))

# Converting wide table to long
total_long <- ba_input1 %>% group_by(Race, year) %>% summarise(deaths = sum(count))
#Bubble plot 
ggballoonplot(total_long, x = "year", y = "Race", size = "deaths", fill = "deaths") + guides(size = FALSE) + gradient_fill(c("#d62323", "#991919", "#5b0f0f"))

Lastly, in an attempt to look at whether certain minority races in certain states were affected disproportionately, we map the proportion of minority population affected out of the total deaths. The map shows that states like Texas, California, New York, Florida etc. have a high (more than 40%) proportion of the police killings are targeted towards people belonging to minority races.

# Grouping data by states
state_grp <- minority_race %>% group_by(State) %>% summarise(totdeath = sum(count), mindeath = sum(minority))
state_pct <- state_grp %>% mutate(minority_pct = mindeath/totdeath*100)

# Deriving census state polygons
us_states <- states(cb = TRUE, resolution = "20m") %>%
  filter(NAME != "Puerto Rico") 

# Joining the dataset to states
us_states_joined <- us_states %>%
  left_join(state_pct, by = c("STUSPS" = "State"))

top_states <- us_states_joined %>% filter(minority_pct>40)
# Mapping
tmap_mode("view")
tm_shape(us_states_joined) + tm_polygons("minority_pct", palette = "Reds") + tm_shape(top_states) + tm_borders(col = "black", lwd=2)

Although the data shows that about half of the estimated 1,000 people shot and killed by police every year in the United States are white — the proportional weight of police violence hits communities of color the hardest. The issue of race bias in police killings is a complex and contentious one in the United States. While the Black Lives Matter (BLM) movement has had a significant impact on the conversation surrounding police killings and race bias but it continues to be a significant issue in the United States. Many studies, investigations, and incidents have raised concerns about racial disparities in police use of force and fatal encounters, which this dataset may not be comprehensive enough to address as it lacks the crucial population data. Apart from this, many police killing also go unreported.