NFL Players Arrest Rates
Emmanuel Tobe
5/12/2019
Background
On February 15, 2014, NFL running back Ray Rice was arrested and charged with assault after a physical altercation in Atlantic City, New Jersey. The charges were dropped after Rice agreed to undergo counseling. However, the release of two videos by TMZ caused a widespread discussion with NFL punishments of players and a larger pattern of arrests of NFL players. Other arrests including the high profile murder charge and suicide of New England Patriots star, Aaron Hernandez, and even Houston Texans tight end, Ryan Griffin, who was arrested in Nashville during the NFL Draft keep the discussion going on the amount of players getting in trouble. News reports, viral videos and tweets about football players’ run-ins with the law makes it appear that criminal activity is an epidemic in the NFL. My hypothesis is that the NFL players get in trouble with the law just as much as the national average but media coverage makes it appear that the NFL has a big issue.
Methodology
This report is based on data from the NFL Arrests 2000-2017 dataset on Kaggle compiled by Patrick Murphy. The data contains arrests, charges and citations of NFL players belonging to an NFL team at the time of the incident or who signed with a team shortly after the incident. The data comes from media reports and public records, however some player arrests might not have been found due to no media coverage or lack of public records. The data covers January 2000 to March 2017, but I had difficulty finding another API that was up to date. There were 850 records of player arrest during that time. After I played with the data for a little bit, I imported the CSV into Microsoft Excel and cleaned up the outcomes, so I could cover the result of the results in R and see vaguely which results happened more frequently. I read each outcome and put them into a specific category with explanation of each category in the outcome section below. I used the Crime in the U.S. dataset compiled by Hax S., which featured the volume and rate of crimes per 100,000 inhabitants from 1997 to 2016. The populations are from the U.S. Census Bureau decennial census counts in 2000 and 2010, and provisional estimates the remaining years. The dataset does a good job showing Violent and Property Crime rates but misses out on more specific rates like Driving Under the Influence and Drug-related crimes. Because of that, I combined some of the relevant information received from that dataset and information from the FBI’s annual Crime in U.S. report and created my own dataset in Microsoft Excel, that doesn’t just show the amount of arrest in the U.S., but also the rate of arrest per 100,000 citizens.
The Amount of NFL Arrest to the average
There have are 850 individual incidents in the dataset between 2000 and March 2017, equaling to about 50 arrest a season. The table below shows the raw arrest data with a date, case, category and outcome of each incident.
Below that is a graph and table that shows the amounts of incidents reported each year. I had to create another variable that seperated the date and allowed me to just pull the year out since the dates were listed as 01/03/2000 in the dataset. The year is represented by the last two digits in the table and graph. That means that year 11, is year 2011. I kept 2017 in the graph but we do not have full data for that year. The data shows that 2006 was the year with the most player incidents with 71 incidents that year. The data does show that the amount of player arrest have been going down since 2013, with 2016 only having 26 incidents, which are only about 2 a month. There seems to be another spike in 2017 with 12 incidents in the first three months. If it were to continue at the pace though, they year would still be the tied for the fourth lowest amount of incidents since the spike 2006.
NFLArrest$Date = as.Date(NFLArrest$DATE, "%m/%d/%Y")
NFLArrest$Year = as.integer(format(NFLArrest$Date, "%Y"))
min_year = min(NFLArrest$Year,NFLArrest$Year)
max_year = max(NFLArrest$Year,NFLArrest$Year)
ArrestByYear <- count(NFLArrest, Year)
max_y_end = max(ArrestByYear$n)
datatable(ArrestByYear, rownames = FALSE, filter="top", options = list(pageLength = 10, scrollX=T) )AY <- ggplot(ArrestByYear, aes(x= Year, y = n)) +
geom_line() +
geom_point() +
xlab("Year") +
ylab("Number of Arrest") +
scale_x_continuous(limits = c(min_year, max_year), breaks = seq(min_year, max_year,1)) +
scale_y_continuous(limits = c(0,max_y_end), breaks = seq(0,max_y_end,10)) +
ggtitle("Total Number of Arrests by Year") +
ggthemes::theme_gdocs()
print(AY)
There were 323,127,513 people living in the United States in 2016. A FBI crime report confirmed that there were 10,662,252 arrest made in 2016. That equates to about 3,300 arrest per 100,000 citizens. We can estimate the number of NFL players as each NFL team is allowed to have 53 players on its roster plus a five-player practice squad, making a total of 1,696 players in the NFL each year. In 2016, there were 26 NFL arrest reported. Although there is not 100,000 players in the NFL, I used algebra to determine that the rate would be about 1,533 arrest per 100,000 players. In 2006, when the NFL had their highest amount of arrest in the 21st century at, 71, their rate of arrest per 100,000 players would be about 4,186. There were 14,380,370 arrest in the United States, out of 299,398,484 million people in the country. That equates to a rate of 4,803 arrest per 100,000 citizens. In fact, out of the 16 year period that we looked at NFL arrest rates compared to the national average, there was only one year that the arrest rate per 100,000 NFL players was higher than the arrest rate per 100,000 U.S. citizens.
library(readxl)
RatesPer100 <- read_excel("RatesPer100.xlsx",
col_types = c("numeric", "numeric", "numeric",
"numeric", "numeric", "numeric",
"numeric", "numeric"))
datatable(RatesPer100, rownames = FALSE, filter="top", options = list(pageLength = 10, scrollX=T) )AR <- ggplot() +
geom_line(data = RatesPer100, aes(x = Year, y = US), color = "blue") +
geom_line(data = RatesPer100, aes(x = Year, y = NFL), color = "red") +
xlab("Year") +
ylab("Arrest Rates") +
ggtitle("Arrest Rates per 100,000 for NFL vs. US Average") +
ggthemes::theme_gdocs()
print(AR)
Figure: The blue represents the U.S. and the Red represents the NFL.
To go further into comparing the NFL players to the national average, I decided to look at demographics. I created a new variable using data from the Bureau of Justice Statistics to find the number of arrests for all Men in the United States between the ages of 21 and 39. There have been no women to play a game in the NFL. I picked that age group because it is extremely rare for anyone to play in the NFL under 21 and over 40. Last season there was only one player in the NFL under 21, and throughout the NFL’s 98 year history there have been only 61 players to play over 40. I then found the arrest rate for men between 21 and 40 per 100,000 and found it much higher than the national average. In 2014, the arrest rate for Men between those ages per 100,000, is 10,650, compared to the national average at 3,514.
library(readxl)
Men_ages_21 <- read_excel("Men ages 21.xlsx",
col_types = c("numeric", "numeric", "numeric", "numeric"))
datatable(Men_ages_21, rownames = FALSE, filter="top", options = list(pageLength = 10, scrollX=T))AM <- ggplot() +
geom_line(data = Men_ages_21, aes(x = Year, y = US), color = "blue") +
geom_line(data = Men_ages_21, aes(x = Year, y = NFL), color = "red") +
geom_line(data = Men_ages_21, aes(x = Year, y = Men), color = "green") +
xlab("Year") +
ylab("Arrest Rates") +
ggtitle("Arrest Rates per 100,000 for NFL vs. US Average") +
ggthemes::theme_gdocs()
print(AM)
Figure: The blue represents the U.S., Red represents the NFL and Green represents Men Age Group.
men <- Men_ages_21$Men
us <- Men_ages_21$US
nfl <- Men_ages_21$NFL
men_box <- rnorm(200,mean=mean(men, na.rm=TRUE), sd=sd(men, na.rm=TRUE))
us_box <- rnorm(200,mean=mean(us, na.rm=TRUE), sd=sd(us, na.rm=TRUE))
nfl_box <- rnorm(200,mean=mean(nfl, na.rm=TRUE), sd=sd(nfl, na.rm=TRUE))
boxplot(men, us, nfl,
main = "Boxplots of Varying Rates per 100,000 people",
at = c(1,2,3),
names = c("Men Rate", "US Rate", "NFL Rate"),
las = 2,
col = c("orange","red"),
border = "brown",
horizontal = TRUE,
notch = FALSE
)
Violent Crimes
One major difference between NFL arrest and the national average is the amount of violent crime. Violent crime is typically defined as a crime in which an offender or perpetrator hurts or threatens to hurt someone either physically or by some other means. Violent crimes may include homicide, murder, assault, manslaughter, sexual assault, rape, negligence, endangerment, kidnapping, extortion, and harassment. A property crime is a crime to obtain or destroy money or property. These include arson, burglary, larceny, extortion, embezzlement and vandalism. The national average of property crimes is much higher than the national average for violent crimes. In the U.S., property crimes are about ten times more prevalent than violent crimes. However, the rate of violent crimes are higher than the rate of property crimes in the NFL. That might be due to the financial class of NFL players as technically the poverty rate of NFL players are 0% because the minimum salary in 2018 is 480,000$. Below are the top 5 categories for arrest in the NFL each year.
DUI_arrests = filter(NFLArrest, CATEGORY == "DUI") %>%
group_by(Year) %>%
dplyr::summarise(count=n())
DUI_arrests$Category = ("DUI")
Drugs_arrests = filter(NFLArrest, CATEGORY == "Drugs") %>%
group_by(Year) %>%
dplyr::summarise(count=n())
Drugs_arrests$Category = ("Drugs")
DV_arrests = filter(NFLArrest, CATEGORY == "Domestic violence") %>%
group_by(Year) %>%
dplyr::summarise(count=n())
DV_arrests$Category = ("Domestic Violence")
Assault_arrests = filter(NFLArrest, CATEGORY == "Assault") %>%
group_by(Year) %>%
dplyr::summarise(count=n())
Assault_arrests$Category = ("Assault")
DC_arrests = filter(NFLArrest, CATEGORY == "Disorderly conduct") %>%
group_by(Year) %>%
dplyr::summarise(count=n())
DC_arrests$Category = ("Disorderly conduct")
rbind(DUI_arrests, Drugs_arrests, DV_arrests, Assault_arrests, DC_arrests) -> top_5_category_arrests
ggplot(top_5_category_arrests, aes(x=Year, y=count, fill=Category)) +
geom_bar(stat="identity") +
ggtitle("Top 5 Categories Arrests Each Year") +
ggthemes::theme_gdocs()
According to the 2006 FBI UCR, the violent crime rate was 473 for every 100,000 citizens, while the property crime rate was 3,334 for every 100,000 citizens. In 2006, when there were 71 arrest, the rate of violent crimes compared to property crimes was much higher than the national average. I created an Excel document and had to manually determine which each crime fell under which category using the FBI’s Hierarchy Rule. The table and bar graph below shows my calculations although they are not accurate. Drugs and alcohol are the number one category that people got arrested for, but violent crimes was the second highest category, with 20 out of 71 crimes being a violent crimes. That equates to a violent crime rate of 28,571 arrest per 100,000 NFL players.
library(readxl)
Crimes <- read_excel("ArrestIncidents.xlsx",
sheet = "2006 Crimes")
Crimes <- as.data.frame(Crimes)
print(Crimes)## Category Number
## 1 Drugs/Alcohol 30
## 2 Violent Crimes 20
## 3 Other 15
## 4 Guns 14
## 5 Property Crimes 2library(ggplot2)
ggplot(Crimes, aes(x=reorder(factor(Category), Number), y=Number)) +
geom_bar(stat = "identity", fill="light blue") +
xlab("Team") +
ylab("Number of Arrests by Team Through time") +
ggtitle("Arrests by Crime in 2006") +
coord_flip()
Conclusion
I believe my hypothesis was mostly proven, but the questions come when you look at the varying sample sizes. When you compared the rate of arrest per 100,000 for NFL to the national average, the national average was higher every year except 2013. When I segmented the U.S. population to match the demographics of the NFL the arrest rate was even higher. One thing that was concerning was the amount of violent crimes. When people make comments about the amount of NFL incidents they typically bring up events like Ray Rice and Kareem Hunt. The amount of violent crime is alarming, especially when you see that Domestic Abuse and Assault are the 3rd and 4th most common arrest reason in the NFL from 2000 to 2017. Creating this report was difficult because I had to make a lot of edits to the NFL data since it was a conglomerate list of a variety of different arrest reports from a variety of different sources. I am also not saying that everyone who was arrested assaulted or murdered someone. Similar to the national average, the rate of conviction in the NFL is pretty low although it is difficult to get an estimate of what that rate is.
total_arrests_by_category <- group_by(NFLArrest, CATEGORY) %>%
dplyr::summarise(count=n()) %>%
arrange(desc(count))
ggplot(total_arrests_by_category[0:20,], aes(x=reorder(factor(CATEGORY), count), y=count)) +
geom_bar(stat = "identity", fill="light blue") +
xlab("Category") +
ylab("Number of Arrests by Category Through time") +
ggtitle("Arrests by Category from 2000 to 2017") +
coord_flip() +
ggthemes::theme_gdocs()